We read the 15 Major League Baseball games ESPN's scoreboard listed as Final for September 11, 2026. In 12 of them, the winning team collected more hits than the team it beat. In 3, the hit column ran the other direction. In all 3, the team with more hits did not hit a single home run. In all 3, the team with fewer hits hit at least one.

Detroit beat Colorado 6-2. The Rockies collected 10 hits; the Tigers collected 5. Adding walks — another route to base — Colorado reached base 12 times (hits plus walks), Detroit 11 times. Detroit hit 2 home runs. Colorado hit none. Detroit scored 6 runs on 5 hits; Colorado scored 2 on 10.

Tampa Bay beat Houston 3-1. The Astros collected 9 hits to Tampa Bay's 4. Houston reached base 10 times (hits plus walks); Tampa Bay reached base 6 times. Tampa Bay hit 2 home runs. Houston hit none. Tampa Bay scored 3 runs on 4 hits and 2 walks; Houston scored 1 run on 9 hits and 1 walk.

The Dodgers beat Miami 6-2. Miami collected 10 hits; Los Angeles collected 8. Both teams reached base 13 times (hits plus walks). The Dodgers hit 1 home run. Miami hit none. Los Angeles scored 6; Miami scored 2.

Combined across the 3 games: the teams with more hits reached base 35 times and scored 5 runs. The teams with fewer hits reached base 30 times and scored 15.

Across the 15 games, the winner had more home runs in 9, an equal count in 2, and fewer in 4. In each of the 4 games where the winner hit fewer home runs, the winner collected more hits. Among these 15 games, not one winner was both out-hit and out-homered in the same box score.

The Analysis

The following is analysis, not fact.

The pattern across the 3 games is consistent with a specific feature of home-run-era baseball ^sspeculative: a home run consolidates scoring in a single plate appearance, while a single typically advances runners a base at a time. A team that reaches base 10 times through hits and walks and scores 1 run has converted 10% of its on-base events; a team that reaches base 6 times and hits 2 home runs can score 3 or more runs on fewer on-base events. The comparison is not a judgment about offensive quality in some broader sense — it is about which runs actually crossed the plate. Houston's 10 on-base events produced one run; Tampa Bay turned six into three — more baserunners, fewer runs.

The two cases where the losing team reached base more often and still lost — Detroit versus Colorado and Tampa Bay versus Houston — share a specific structure: the losing team reached base more often than the winning team (Colorado 12 vs Detroit 11; Houston 10 vs Tampa Bay 6) and still scored a fraction of what its opponent did. In both, the outcome is consistent with the winners' home runs producing more runs from fewer on-base events ^sspeculative.

One sharper line: across these 15 games, the winner always held at least one column — more hits or more (or equal) home runs. No team won while losing both.

Room for Disagreement

A single Friday's 15 games is not a trend. The run value of the home run relative to the single has been the foundation of baseball's analytical vocabulary for decades — the finding here is consistent with a known principle rather than discovering one. The losing teams' pitching may have contributed as much as their offense ^sspeculative: a team that allows a home run is penalized differently than one that allows a single, regardless of what its own lineup produces. The 3 matchups may also represent a specific pitcher-batter configuration — contact-heavy lineups against starters who surrendered fewer, harder-hit balls — that does not predictably repeat across a longer sample ^sspeculative. What is new here is the same pattern closing 3 separate games that day, and the specific arithmetic: in two of the three, the team that reached base more often scored less.

Notable

• ESPN · Colorado Rockies at Detroit Tigers — Sept. 11, 2026 recap — Game recap for Colorado at Detroit; Rockies had twice as many hits, scored a third as many runs.
• ESPN · Houston Astros at Tampa Bay Rays — Sept. 11, 2026 recap — Game recap for Houston at Tampa Bay; Astros reached base 10 times and scored once.
• ESPN · Los Angeles Dodgers at Miami Marlins — Sept. 11, 2026 recap — Game recap for the Dodgers 6-2 win on 8 hits against Miami's 10.
• ESPN · Cincinnati Reds at Milwaukee Brewers — Sept. 11, 2026 recap — Milwaukee's 20-0 win over Cincinnati (23 hits, 4 home runs).

The suicide-rate figures in this piece are WHO modeled estimates, not direct death counts — the series (World Bank code SH.STA.SUIC.P5, SDG indicator 3.4.2) derives rates from cause-of-death surveillance and statistical modeling; the economy counts and percentage changes are calculations derived from those rates, computed from the unrounded WHO estimates while the rates shown here are rounded to one decimal. SDG 3.4, adopted by all 193 UN member states in 2015, sets a one-third reduction in premature noncommunicable disease mortality by 2030; SDG indicator 3.4.2 is the designated mental-health gauge under that goal, though it carries no numerical reduction target of its own — the one-third mark is used here only as a comparison yardstick.

Lithuania's modeled rate was 37.5 per 100,000 in 2010, the highest of the 185 economies with comparable data through 2021, and fell to 22.1 by 2021 — a drop of 41 percent. ^mmodeled Mexico's was 4.5 and rose to 7.0 (up 55 percent). Thailand's was 8.9 and rose to 16.6, up 86 percent — the largest increase among economies with populations over one million. ^mmodeled The world's modeled aggregate fell from 10.67 to 9.13 per 100,000, a decline of 14.4 percent. ^mmodeled

Of the 185 economies with comparable 2010 and 2021 estimates — excluding Antigua and Barbuda, whose modeled 2010 rate of zero registers an artifactual rise — 23 cut their modeled rate by one-third or more from the 2010 baseline, 62 recorded a higher rate in 2021 than in 2010, and 100 improved by less than a third. ^mmodeled

Europe and Central Asia accounts for the largest regional share of the improvement: 11 of the 23 economies that cleared the one-third mark are in the region. Belarus fell from 34.2 to 15.6 (−54 percent); Kazakhstan from 28.1 to 14.6 (−48 percent); Russia from 35.8 to 21.4 (−40 percent). ^mmodeled Sub-Saharan Africa moved the other way: 30 of 48 economies with data recorded higher modeled rates in 2021 than in 2010. ^mmodeled In Latin America and the Caribbean, 11 of 31 economies worsened, including Brazil (5.3 to 7.6, up 45 percent) and Uruguay (16.1 to 24.8, up 54 percent). ^mmodeled Twelve high-income economies saw rates rise: the United States from 13.2 to 15.6 (up 18 percent), the United Kingdom from 7.8 to 9.6 (up 22 percent), and Spain from 7.0 to 8.7 (up 24 percent) among them. ^mmodeled

The Analysis

The following is analysis, not fact.

The steepest percentage declines were concentrated in former-Soviet economies that began 2010 at two to four times the world average — Belarus, Kazakhstan, Lithuania, Russia ^mmodeled. But because a global rate is population-weighted, steep drops in relatively small populations weigh less on the world total than more modest declines across the largest populations. India's modeled rate fell from 15.9 to 12.6 per 100,000; because it carries a population near 1.4 billion, a decline of that size exerts substantially more weight on the global aggregate than comparable movements in Belarus or Lithuania ^mmodeled.

A cross-country pattern emerges across two axes. One is regional: Europe and Central Asia improved in 41 of 48 economies, 11 by a third or more; Sub-Saharan Africa did not, with 30 of 48 worse ^mmodeled. The other is starting point: economies that began far above the world average tended to fall; those that began below it tended to rise. Whether those axes are independent — whether the former Soviet-successor states improved because of something the region specifically did — the data alone cannot establish. ^sspeculative

South Korea, which recorded 34.9 per 100,000 in 2010, cut to 27.5 by 2021 (−21 percent), tracking the right direction but short of the one-third pace used here as a yardstick. ^mmodeled

The hardest finding to explain by resources alone is the high-income cluster. The United States, the United Kingdom, Spain, Norway, and the Netherlands each had established mental health systems in 2010. Each saw its modeled rate rise. The data records that divergence; it does not identify a cause. ^sspeculative

PAHO, in its own analysis, has identified the Americas as the only WHO region where aggregate suicide mortality rose over the period it examined — consistent with the country-level pattern this data shows.

Room for Disagreement

These are modeled estimates built on civil registration systems that vary sharply in completeness. In sub-Saharan Africa especially, where many countries have low registration coverage, modeled rates can respond as much to updated data assumptions as to genuine changes in incidence ^sspeculative. The dramatic drops in Belarus and Russia may partly reflect reclassification of deaths previously coded to other causes rather than prevention progress alone ^sspeculative.

The View From

**View from Latin America**

PAHO has identified the Americas as the only WHO region where aggregate suicide mortality rose over the period of its analysis. At the country level, the picture is more fractured: Chile fell 34 percent, Mexico rose 55, Bolivia fell 27, Uruguay rose 54. What reads from outside as a regional trend maps internally as a set of divergences without a common cause health authorities can easily name. ^sspeculative

Notable

• PAHO · Suicide prevention — PAHO's regional summary includes the finding that the Americas is the only WHO region where aggregate suicide mortality has increased — corroborating the country-level data here.
• WHO · Suicide — Fact sheet — WHO current global summary: more than 720,000 deaths a year, 73 percent in low- and middle-income countries, with access to lethal means cited as a key modifiable factor.
• WHO · World Mental Health Report: Transforming mental health for all — The 2022 WHO flagship mental health report: more than 1 in 100 deaths globally is by suicide, 58 percent of suicides occur before age 50, and 20 countries still criminalize attempted suicide.

The Seahawks beat the Patriots 13-10 in a Super Bowl rematch, the first regular-season meeting between the two franchises since February's championship game [verified — ESPN game article]. New England ran 67 plays to Seattle's 48. The Patriots held the ball for 34 minutes and 17 seconds — 8 minutes and 34 seconds more than the Seahawks. They converted 5 of their 16 third-down attempts (31.2%) to Seattle's 2 of 11 (18.2%).

New England scored first. Eli Raridon caught a 2-yard Drake Maye pass for a touchdown in the second quarter; Andy Borregales converted a 50-yard field goal in the third. Seattle trailed 10-0 before Jason Myers answered with a 30-yard field goal, also in the third quarter. The score stood at 10-3 when New England's win probability reached 82.2% in the fourth quarter [verified — ESPN win probability data].

Then Maye threw 3 interceptions on 3 consecutive drives [verified — ESPN box score]. Jaxon Smith-Njigba caught a 45-yard touchdown pass from Drew Lock to tie the game 10-10; Myers's 26-yard field goal won it. Maye finished 23-for-33 for 178 yards, 1 touchdown, and 3 interceptions. Lock was 16-for-22 for 187 yards and 1 touchdown. Smith-Njigba caught 8 passes for 122 yards and that score [verified — ESPN box score and leaders data].

The Analysis

The following is analysis, not fact.

The box score marks Seattle 0-for-2 converting in the red zone — zero touchdowns from two trips inside the opponent's 20. The arithmetic behind those two field goals tells a different story.

An NFL field goal distance equals the line of scrimmage plus 17 yards: 7 for the snap, 10 for the end zone depth. A 30-yard field goal is kicked from the 13-yard line. A 26-yard field goal is kicked from the 9 ^mmodeled. Both of Seattle's field goal attempts [verified — Jason Myers 30-yd Q3, 26-yd Q4] originated inside the red zone. The two trips logged as red-zone conversion failures each produced a made field goal. Six of Seattle's 13 points came from inside the opponent's 20 — not zero.

The other 7 points arrived on a drive that never required the zone. Smith-Njigba's 45-yard catch from Lock reached the end zone without a snap inside the 20. Seattle's only touchdown bypassed the red zone entirely.

New England's geometry runs the other way. Their 50-yard Borregales field goal came from the 33-yard line ^mmodeled — outside the red zone, on a drive separate from their two red-zone trips. Their 1-for-2 red-zone rate reflects one touchdown (Raridon's second-quarter catch) and one trip that reached the Seattle red zone on a 15-play drive and ended in a Drake Maye interception [verified — ESPN drive data]. One of their two red-zone possessions turned the ball over.

The full efficiency picture: New England generated 4.13 yards per play across 67 snaps; Seattle generated 5.94 across 48. The Patriots ran 19 more plays, generated 6 more first downs, and held a nearly 9-minute possession edge — and still trailed in total yards, 277 to 285. Their last three drives all ended in interceptions, converting 11:38 of possession and 88 yards — from late in the third quarter into the fourth — into zero points [verified — ESPN drive data].

Room for Disagreement

Three interceptions explain the loss more concisely than any geometric reframing ^sspeculative. Had Maye protected the football, New England's time-of-possession edge and third-down advantage would very likely have generated additional scoring drives: the Patriots were running a high-volume, methodical offense that moved the chains consistently ^sspeculative. The red-zone arithmetic is accurate, but Seattle's fourth-quarter surge rode the picks ^sspeculative — the tying touchdown followed New England's first interception, and the winning field goal followed another that handed Seattle a short field at the New England 47 [verified — ESPN drive data]. Remove the three picks and Seattle's comeback, not the red-zone arithmetic, is what likely never materializes ^sspeculative.

Notable

• ESPN · Seahawks pick off Drake Maye 3 times to beat Patriots 13-10 in Super Bowl rematch — ESPN game article, published September 10
• ESPN · New England Patriots at Seattle Seahawks — game recap — ESPN full game recap, event ID 401872656
• Pro Football Reference · New England Patriots at Seattle Seahawks — box score, September 10 2026 — Full box score and statistical record

When a country joins the World Trade Organization, it files a schedule of concessions that specifies a maximum tariff rate — a bound rate — for the product categories it agrees to bind. A bound line is a legal ceiling, enforceable at the WTO; a line left unbound carries no ceiling. The rate a country actually charges on a given day is the applied rate, which on a bound line can sit anywhere at or below that ceiling. The gap between them is known as binding overhang, or tariff water: unused room to raise tariffs within the rules. Measured here as the gap between a country's average bound and average applied rate — the World Bank figures, which count unbound lines as zero — it is an aggregate proxy; the binding limit on any single product is that product's own line.

The World Bank compiles both rates annually from UNCTAD TRAINS. This piece pulls the two World Bank series for 2021, chosen for its bound-rate coverage, and computes the gap for every economy with both figures. That yields 140 economies with matched data; Switzerland is set aside as a data artifact (see below), leaving a 139-economy analytical panel. Both series use the simple-mean method — an unweighted average across all tariff lines — so the comparison is methodologically aligned, though the specific tariff lines each series covers may differ.

Bangladesh holds the largest gap among the 139 economies. Its WTO-bound rate averages 158.9% across all product lines. Its applied rate is 12.6%. The aggregate gap is 146.3 percentage points. Nigeria and Tanzania follow: bound at 121.3% and 120.0% respectively, applied at 12.7% and 12.0%, with gaps of 108.6 and 108.0 points. Zambia, Kuwait, the Gambia, Mozambique, Mauritius, Myanmar and Kenya round out the top ten, each carrying more than 80 points of unused tariff space.

At the other end of the distribution, the United States has a bound rate of 3.6% and an applied rate of 2.8% — a gap of 0.8 points. The European Union (which negotiates as a single WTO member) runs a bound rate of 4.4% and an applied rate of 2.2%, a gap of 2.2 points. Canada and Japan sit in the same narrow band: gaps of 2.9 and 0.9 points respectively. Australia is an outlier among rich economies — bound at 9.6%, applied at 1.9%, a gap of 7.7 points.

India, a frequent subject of WTO trade disputes, carries a bound rate of 52.0% against an applied rate of 9.9%: 42.1 points of unused room. Indonesia's gap is 31.3 points (bound 37.3%, applied 6.0%). Brazil's is 18.3 points (bound 31.5%, applied 13.2%). China's is narrow: 4.7 points (bound 10.0%, applied 5.3%).

Across the 139-economy panel, the median gap is 12.6 percentage points. Fifty-seven of 139 economies have gaps exceeding 20 points; 82 of 139 exceed 10 points; 45 of 139 show gaps below 5 points — and most of those are wealthy economies or members of customs unions where bound rates are already low. Two economies, Hong Kong and Macao, carry a bound rate of 0.0% and apply 0.0% — the only cases where the gap is precisely zero.

Two economies show applied rates nominally above their bound rates in the 139-economy panel: Norway by 0.7 points and Cote d'Ivoire by 0.9 points. These are not confirmed WTO violations. At the aggregate simple-mean level, rounding in bound-rate schedules and differences in the tariff lines covered by each series can produce small apparent inversions. Confirming a WTO violation requires a product-line analysis, which this aggregate series does not support.

Switzerland reports a bound rate of 0.0% in the World Bank series across all available years, producing a large apparent inversion against its 4.1% applied rate. The World Bank indicator records Switzerland's simple-mean bound rate as 0 consistently, suggesting a data artifact — Switzerland does have WTO commitments — and Switzerland is excluded from the panel analysis on this basis.

The Analysis

The following is analysis, not fact.

At the tariff-line level, binding overhang directly measures unused headroom; the country-average gap used here is an aggregate proxy for how much room the rules-based trading system leaves unused. On the average figures, a country's committed ceiling can sit five, ten, or twenty times above its current average rate — a commitment to the status quo only as long as the status quo suits the country, though on any individual product the binding line is reached first ^mmodeled. The WTO's dispute mechanism, the schedule of concessions, and the phrase 'rules-based order' all treat the bound rate as the system's anchor. But for 57 of the 139 economies with comparable data, that anchor sits more than 20 percentage points above the average applied rate.

The largest observed gaps are concentrated among developing economies ^mmodeled. They are predominantly low- and lower-middle-income states that joined the WTO (or its predecessor, GATT) in rounds when developing countries were not pushed toward aggressive binding commitments ^mmodeled. Bangladesh's 159% average bound rate reflects a negotiating outcome, not an intent to charge 159% ^mmodeled: the country agreed to bind its tariffs — itself a concession — without being pushed to bind them low. The ceiling is the price of membership, the applied rate is the price of imports, and the gap between them is the residue of each round of negotiations ^mmodeled.

That residue matters now in a way it did not in the 1990s or 2000s ^sspeculative. The 2025 wave of tariff escalations by major economies — the United States in particular — has focused attention on who can retaliate and how. Applied rates can be raised toward the bound level without any WTO procedure. On the average figures, a country with 80 or 100 points of tariff water could raise applied rates substantially within the rules, though line-by-line ceilings would bind before the average gap is exhausted ^mmodeled. Countries with 1 or 2 points of average water — the US, EU, Japan — have little equivalent room: their bound and applied rates are close, which may be one reason negotiators in Washington and Brussels have acted through emergency safeguard clauses and Section 232 rather than simply dialing up applied rates ^mmodeled. The asymmetry in tariff water may, in part, translate into an asymmetry in trade-war ammunition ^sspeculative.

The gap also bears on investment planning. A large country-average overhang may indicate greater tariff-policy uncertainty for some manufacturers weighing where to invest, depending on the relevant product-line bindings ^mmodeled. The WTO schedule legally commits a member to hold rates at or below the per-line ceilings; it does not guarantee they will remain near the floor. Predictability at the ceiling is not the same as predictability at the applied rate ^sspeculative.

Room for Disagreement

The strongest counter to the tariff-water framing is that the gap is not a latent threat but a design feature that has functioned as intended for three decades. The GATT and WTO rounds deliberately allowed developing countries to set high bound rates as a development prerogative — the ceiling was always understood to be a safety valve, not a target. Historically documented episodes of significant applied-rate increases exploiting large tariff water appear uncommon outside genuine economic crises (Argentina in 2001-02 among them). The existence of the gap does not predict its use, and treating it as 'unexploded ammunition' attributes intent to an accounting identity ^sspeculative.

A second counter is methodological. The simple-mean bound rate — an unweighted average across all tariff lines, including unbound lines counted as zero — is an imperfect proxy for actual legal headroom. Bangladesh's 159% figure averages across lines that include some very high agricultural bound rates and many lower industrial rates; the headline number likely overstates the room on most goods and understates the political constraint on raising rates on food, where consumer prices are politically sensitive ^mmodeled. Product-line-level analysis tells a different story than the country average.

A third counter applies specifically to the low-gap economies. The US, EU, and Japan have low gaps not because the WTO forced them there but because they chose, through eight rounds of negotiations, to liberalize deeply. The constraint they face is one they imposed on themselves as part of extracting concessions from others. From that vantage, the asymmetry is the reward for having led liberalization, not a vulnerability.

The View From

From India's vantage, the 42-point gap reads as a policy instrument, not a risk. On this view — the one a middle-income WTO member might well take — the ability to raise applied rates without WTO violation is a development tool that economies at India's income level retained in their Uruguay Round schedules. Where a critic sees policy uncertainty, this vantage sees policy space. Both readings describe the same number: the gap reads as a risk from the exporter's side of the ledger and as a right from the importer's side.

Notable

• WTO · World Tariff Profiles 2025 — The joint WTO/ITC/UNCTAD annual compilation of bound and applied tariff rates for more than 170 economies, with agricultural and non-agricultural splits. The 2025 edition covers rates through end-2024.
• WTO · Tariffs: more bindings and closer to zero — The WTO's own explanation of the bound-applied distinction and the Uruguay Round's expansion of binding coverage from 78% to 99% of product lines for developed countries.
• WTO · Tariff and trade data — The WTO tariff database gateway for members' bound schedules and applied rates.
• TradeWeave · Binding Overhang, Bound vs Applied Tariffs by WTO Member — An interactive tool for examining binding overhang by economy, sourced from WTO/TRAINS and World Bank data.
• UNCTAD · World Tariff Profiles 2024 — The 2024 joint UNCTAD/WTO/ITC edition of the same annual tariff comparison series.
• World Bank · Bound rate, simple mean, all products (%) — TM.TAX.MRCH.BR.ZS — The World Bank's WDI indicator landing page for the bound-rate series, sourced from UNCTAD TRAINS.
• World Bank · Tariff rate, applied, simple mean, all products (%) — TM.TAX.MRCH.SM.AR.ZS — The World Bank's WDI indicator landing page for the applied-rate series, the same methodology as the bound-rate series.
• Saylor Academy / International Trade: Theory and Policy · Appendix B: Bound versus Applied Tariffs — A textbook appendix explaining the mechanics of the bound-applied gap and the deliberate preservation of policy space for developing countries in WTO accession rounds.

International trade accounting carries a built-in directional expectation: the country recording imports usually shows a higher figure than the country recording exports, because imports are priced at arrival — cost of goods plus shipping and insurance — while exports are priced at the dock of departure. That holds when both sides are recording the same shipments; it need not when they attribute a shipment to different partners. For goods moving from China to India, that cost, insurance, and freight (CIF) premium is estimated at 5 to 7 percent above the export-side figure ^mmodeled.

From 2019 through 2022, both China and India filed their bilateral trade totals annually to UN Comtrade, the global repository for trade statistics. Comparing those two sets of filings — each country's own submission — produces a result that runs opposite to that expectation every year.

China's filings (exports to India): $74.8 billion in 2019, $66.7 billion in 2020, $96.4 billion in 2021, $118.5 billion in 2022. India's filings (imports from China): $68.4 billion in 2019, $58.8 billion in 2020, $87.5 billion in 2021, $102.2 billion in 2022. China's tally exceeded India's by $6.4 billion in 2019, $7.9 billion in 2020, $8.9 billion in 2021, and $16.3 billion in 2022 — China's figure on top in every case, where the freight-cost expectation would put India's higher. Applying a mid-range 6 percent freight premium ^mmodeled to China's 2022 export figure would put the expected India import total near $125.6 billion; India's actual filing was $102.2 billion — a gap of about $23.4 billion between that expectation and the record.

In 2023, the two filings converged for the first time in five years: India recorded $122.0 billion, China $117.7 billion — India's figure now the higher by $4.3 billion, or 3.7 percent, the direction the freight-cost expectation predicts, though at or below the low end of the 5-to-7-percent range.

Over the same window, India's recorded imports from Hong Kong — a major re-export platform for mainland Chinese goods — held between $14.6 billion and $19.5 billion a year from 2019 through 2022. In 2023 that figure fell to $3.6 billion, a decline of about $15.8 billion that dropped it well below its prior four-year range. India's recorded imports from the United Arab Emirates followed a similar shape: $30.3 billion in 2019, rising to $53.9 billion in 2022, then falling to $37.5 billion in 2023, a $16.4 billion drop.

The Analysis

The following is analysis, not fact.

The persistent directional reversal in the bilateral accounting from 2019 through 2022 is consistent with a structural feature of how trade statistics capture multi-leg shipments. China's customs rules record exports by final declared destination: a Chinese exporter names India on the export document regardless of which intermediate ports the goods pass through. India's customs rules record imports by immediate country of loading or last significant transformation. A Chinese-manufactured good routed through a Hong Kong bonded warehouse and loaded onto a vessel in Hong Kong would appear in China's books as an export to India and in India's books as an import from Hong Kong — the bilateral China-India pair captures it on China's side but not on India's. The same accounting divergence applies to goods re-exported through Dubai or Singapore. ^sspeculative

The 2023 pattern is consistent with a change in how India assigns country of origin at the point of import, not necessarily a change in physical routing. A stricter origin standard applied at import would reclassify goods previously counted as Hong Kong- or UAE-origin as China-origin, simultaneously shrinking those bilateral tallies and growing the direct China pair — which is the shape the data takes: a $15.8 billion drop in HK-sourced imports to India, a $16.4 billion drop in UAE-sourced imports, and a $19.8 billion gain in the direct China-India bilateral pair, all in the same year. ^sspeculative

The co-movements are striking in scale but do not establish the mechanism. China's exports to the UAE continued to rise through 2023, so any re-export pipeline through Dubai would have ample supply; the drop was on India's receiving side. Neither country's filing can establish the physical path of any specific shipment. Both record customs declarations — declared partners — which need not match actual routing.

Room for Disagreement

The most defensible counter-reading is definitional, not behavioral. China and India apply legally distinct standards for labeling a trade partner. China records the exporter's declared final destination; India records the country of last significant transformation or last point of loading. Goods assembled in China but further processed or repackaged in Vietnam or the UAE may legitimately count as Vietnam- or UAE-origin under Indian standards, with no intent to disguise origin and no circumvention of import duties involved. At the scale of the bilateral relationship, a $10-23 billion annual classification gap is large but likely arises from two countries operating incompatible counting conventions ^mmodeled, not necessarily from deliberate rerouting. The 2023 convergence may equally reflect India improving its data collection from direct importers — catching origin it had previously missed by default — rather than stricter enforcement changing behavior.

The View From

From India's trade-enforcement vantage, the 2023 convergence reads as broadly consistent with what its origin-documentation rules were designed to produce: more Chinese-origin goods captured under the direct bilateral pair. A $15.8 billion one-year drop in recorded imports from Hong Kong — well outside that pair's $14.6-to-19.5-billion band over the prior four years — is one a ministry would likely read as a classification effect ^sspeculative.

From China's perspective, its own records reflect what Chinese exporters legally declare on export documents: final destination as India for goods consigned to Indian buyers. The four-year departure from the usual accounting direction, on this reading, is a measurement artifact of India's import classification system rather than a signal in China's data ^sspeculative.

Notable

• Observatory of Economic Complexity · India and China bilateral trade profile — Background context: visualizes bilateral trade flows between India and China using the same UN Comtrade underlying data. Useful for readers replicating or extending the figures reported here.
• UN Comtrade Plus · UN Comtrade: International Merchandise Trade Statistics — Background context: the primary UN database where both India and China file bilateral trade statistics annually. This analysis drew from WITS, which aggregates the same filings.
• World Bank WITS · World Integrated Trade Solution — Trade Stats — Background context: the World Bank trade portal used to pull both countries' bilateral trade filings. Every figure in this analysis is directly reproducible from this source.

The EU cut direct exports to Russia by €59 billion between 2021 and 2025 — from €89.2 billion to €30.4 billion, a 66 percent decline. Over the same four years, EU exports to two of the routes named in EU Commission circumvention-risk guidance rose by a combined €40.0 billion: Turkey, from €79.1 billion to €114.0 billion (+€34.9B), and Kazakhstan, from €5.5 billion to €10.6 billion (+€5.1B). A third named route, the UAE, rose €22.0 billion over the same window (from €29.8 billion to €51.8 billion); we report that figure with a caveat and hold it apart from the confirmed pair. Turkey's and Kazakhstan's increases cross-verify independently against WITS and the European Commission's trade portal; the UAE's 2025 figure could not be corroborated there, so it rests on Eurostat alone. All three routes are named in EU Commission circumvention-risk guidance as priority routes at risk of channeling EU-origin goods to Russian buyers without appearing in the EU-Russia bilateral record.

In 2025 the EU-Russia goods balance turned to a surplus for the first time in the 24 annual observations Eurostat's bilateral series records, which begin in 2002. Across 2002 through 2024 the EU ran a trade deficit with Russia in every one of those 23 years — at its widest in the series, a €147.5 billion deficit in 2022, when Russian energy prices spiked in the first year of the war. By 2025, with EU total goods imports from Russia fallen from €163.6 billion (2021) to €27.9 billion, the EU ran a surplus of €2.5 billion. The series does not extend before 2002, so this is the first inversion within its window, not necessarily the first in the relationship's history.

Russia's total imports of goods and services, per World Bank national accounts, were $376.5 billion in 2021 and $380.0 billion in 2023 — effectively unchanged through the first two full years of Western sanctions.

[Eurostat ext_lt_maineu, annual bilateral trade, retrieved 2026-09-10; World Bank NE.IMP.GNFS.CD, retrieved 2026-09-10]

The Analysis

The following is analysis, not fact.

Read bilaterally against Russia alone, Eurostat's records tell a story of substantial trade disruption: direct EU exports are down 66 percent and the energy-driven deficit that defined the relationship for two decades has turned into a near-balance. Read alongside the three transit hubs named in EU circumvention-risk guidance, the same data raises a harder accounting question.

EU exports to the two routes we can state with confidence — Turkey and Kazakhstan — rose €40.0 billion since 2021, against €59 billion lost on the direct route to Russia. Both are named in EU Commission circumvention-risk guidance as priority routes at risk of channeling EU-origin goods to Russian buyers without appearing in the direct bilateral record. Add the UAE's single-source €22.0 billion and the three named routes together gained €62.0 billion, marginally exceeding the direct-export cut. Either figure is consistent with a significant portion of previously Russia-bound European goods being re-routed rather than simply withdrawn from commerce ^sspeculative; the confirmed pair alone accounts for roughly two-thirds of the decline. Eurostat's data does not establish the mechanism or decompose the re-export share from organic trade growth; it establishes the aggregate magnitudes and their coincidence across time.

Russia's own import aggregate lends circumstantial weight to the same reading ^sspeculative. A trading economy effectively cut off from a major pre-war partner would typically show a sustained import contraction ^mmodeled; World Bank records instead show 2023 imports roughly matching 2021 at approximately $380 billion, consistent with replacement supply chains rather than absent supply.

The bilateral EU-Russia ledger and the four-partner ledger do not contradict each other arithmetically — both are correct Eurostat readings of the same data. What neither can answer in isolation is the question the combined read raises: where did the goods go?

Room for Disagreement

The straightforward counter is that Turkey and the UAE are large, fast-growing economies whose import expansions are substantially self-generated. Turkey's total imports of goods and services grew from $229 billion in 2019 to $387 billion in 2022, per World Bank national accounts — an expansion of $158 billion that may partly reflect domestic inflation, energy purchases, and manufacturing inputs with no inherent relationship to Russia ^mmodeled. The EU's €35 billion gain in Turkish trade over four years is one among many growing supplier relationships in an economy that was already expanding rapidly. Kazakhstan's EU import jump may partly reflect its own infrastructure and modernization investment agenda ^mmodeled. Neither set of flows is inherently attributable to Russia re-export.

The EU Commission, while naming these countries in circumvention guidance, also reports progress in enforcing controls on dual-use and military-critical goods. The categories more exposed to re-export are civilian and commercial rather than the weapons-enabling items the sanctions regime was primarily designed to restrict ^mmodeled. Enforcement on high-priority battlefield goods has tightened through successive packages.

The View From

**View from Ankara** — Turkey has consistently maintained that it observes international law, has not ratified Western sanctions, and trades under its sovereign rights as an independent state. Turkish officials have rejected 'conduit' framing as an attempt to extend EU regulatory jurisdiction beyond EU territory, noting that Turkey's trade expansion predates 2022. Ankara points to its active mediation between Russia and Ukraine as evidence of genuine neutrality rather than alignment with either side.

Notable

• Council of the EU · EU Restrictive Measures: Russia — Council of the EU overview of the full Russia sanctions framework and successive packages, including the 14th Package anti-circumvention provisions.
• Eurostat · International Trade in Goods -- Data Browser (ext_lt_maineu) — The bilateral EU partner-country trade series used to derive all export and import figures in this report.
• Kiel Institute for the World Economy · War Against Ukraine -- Research and Analysis — Independent research on sanctions effectiveness and Russia trade flow analysis.
• World Bank · Imports of Goods and Services -- Russian Federation (NE.IMP.GNFS.CD) — National accounts series used to assess Russia aggregate import resilience, 2021-2023.

The Federal Reserve began cutting its benchmark rate in September 2024. From its August 2024 level of 5.33 percent, the federal funds rate had fallen 170 basis points to 3.63 percent by August 2026. The average credit card rate — the Federal Reserve's quarterly H.15 series, which reads on different observation dates — fell 82 basis points over a closely overlapping span, from its August 2024 high of 21.76 percent to 20.94 percent in its latest quarterly reading, May 2026 [series TERMCBCCALLNS, FRED].

The resulting gap as of May 2026 — 17.31 percentage points — is the second-widest in the 127 quarterly readings the Federal Reserve has published since it first measured the average credit card rate in November 1994. The record came one quarter earlier: in February 2026, the average card rate of 21.00 percent against a federal funds rate of 3.64 percent produced a spread of 17.36 percentage points. The two most recent quarterly readings are the two widest in the 32-year series.

The trajectory ran in reverse on the way up. The Fed raised its benchmark 525 basis points between November 2021 and August 2023, from 0.08 percent to 5.33 percent. Credit card rates rose more over their own cycle — 725 basis points, from 14.51 percent in November 2021 to a 21.76 percent high in August 2024, continuing to climb for about a year after the Fed stopped hiking.

Across the full 32-year history of the series, the narrowest spread between the policy rate and the average credit card rate was 7.81 percentage points, recorded in August 2006, when the federal funds rate sat at 5.25 percent and the average card rate was 13.06 percent.

The Analysis

The following is analysis, not fact.

Two asymmetries stack here. On the way up, credit card rates rose 725 basis points against the Fed's 525 — 38 percent more, in raw basis-point terms. On the way down, they have moved less than half as far. The result is a spread that hit a 32-year record in February 2026 — 17.36 percentage points — and retreated by only five basis points in the three months that followed.

One framework ^sspeculative consistent with this pattern: credit card rates are indexed to the prime rate, which typically moves closely with the federal funds target, but issuers set the margin above prime based on operating costs and credit-risk expectations that do not move one-for-one with the benchmark. During the near-zero-rate period from 2020 to 2022, card rates held in the low-14-percent range despite a near-zero policy rate — consistent with a substantial non-policy-rate component in card pricing ^mmodeled. As benchmark rates rose, issuer margins and other components may also have shifted; over their respective cycles, the card-rate increase was larger in basis-point terms than the federal-funds increase ^mmodeled. On the way down ^sspeculative, the same structural stickiness may operate in reverse: once margins have repriced higher, competitive pressure to lower them may be slower to build than the pressure to raise them when funding costs spike.

Credit card delinquency rates reached a 2024 high of 3.22 percent in April and have since declined to 2.85 percent as of April 2026, per FRED series DRCCLACBS. Charge-off rates reached a 2024 high of 4.69 percent in July and now stand at 3.82 percent as of April 2026, per series CORCCACBS. Both remain modestly above their pre-2020 baselines — delinquency ran 2.61 percent in mid-2019 — but the direction in both is down. The direction of travel in the credit-quality data weakens a simple contemporaneous credit-risk explanation for the rate stickiness: the spread has widened to a 32-year record even as the metrics that would justify a risk premium decline from their peaks ^mmodeled.

Room for Disagreement

The strongest counter to the record-spread framing is compositional. The credit card market in 1994 was narrower: fewer subprime and secured-card products existed. Those products often carry rates in the 25-to-30-percent range and pull the all-accounts average upward regardless of the policy rate ^sspeculative. Part of the widened spread ^sspeculative may reflect structural expansion of credit access over three decades — more people with credit cards, more of them at the higher-rate end of the market — rather than a change in what any individual borrower type is paying.

A methodological limit also applies to the comparison: the H.15 series tracks the average rate across all credit card accounts, including those carrying a zero balance at the statement date, where the stated APR costs the holder nothing that billing cycle. Borrowers who actually revolve a balance can face a materially higher effective rate than the all-accounts average captures. The spread calculation here may therefore be conservative relative to the experience of revolving borrowers ^mmodeled.

Notable

• Consumer Financial Protection Bureau · The Consumer Credit Card Market (annual report) — CFPB's annual review of credit card pricing, terms, and consumer outcomes — the primary regulatory lens on rate trends and market structure.
• Federal Reserve · H.15 Selected Interest Rates — The H.15 release, published since 1975; the all-accounts credit-card-rate series used here (TERMCBCCALLNS) begins in November 1994.
• Bankrate · Average credit card interest rates this week — Weekly rate tracker that contextualizes the quarterly H.15 reading against real-time issuer offers; provides a higher-frequency view of the same trend.
• NerdWallet · Average Credit Card Interest Rate in America Today — Consumer-facing rate tracker covering the current-account and new-offer landscapes that the all-accounts H.15 average blends together.

The Bureau of Labor Statistics Job Openings and Labor Turnover Survey (JOLTS) put the national quit rate — voluntary separations as a share of employment, seasonally adjusted — at 1.9% in July 2026. In July 2019, before the pandemic, it was 2.4%.

We read the quit rate for all eleven major JOLTS industry supersectors in both months. Ten sat at or below their July 2019 level. One did not: mining and logging rose to 2.6% from 1.9%, a 0.7-point increase, up 37% — the only major industry where workers quit more often in July 2026 than before the pandemic.

The full cross-section, July 2026 against July 2019, ordered by the current rate:

- Leisure and hospitality: 3.4% vs. 4.8% - Trade, transportation, and utilities: 2.6% vs. 2.8% - Mining and logging: 2.6% vs. 1.9% - Other services: 2.2% vs. 2.5% - Construction: 1.9% vs. 2.3% - Professional and business services: 1.8% vs. 3.3% - Private education and health services: 1.8% vs. 2.0% - Manufacturing: 1.4% vs. 1.5% - Financial activities: 1.2% vs. 1.5% - Information: 1.0% vs. 1.5% - Government: 0.8% vs. 0.8%

Among the ten at or below their July 2019 rate, the steepest decline was professional and business services — down 1.5 points, or 45.5%, the largest in the set on both the point and the percentage measure. Leisure and hospitality remains the highest-quit supersector at 3.4%, though its distance from the field has closed: it ran 2.4 points above the national rate in July 2019 and 1.5 points above it now, with trade/transportation/utilities and mining and logging just behind at 2.6%. Government returned exactly to its 2019 rate, 0.8% in both months. Within trade, the retail-trade sub-industry ran 3.1% in July 2026, down from 3.4%.

The figures are the seasonally adjusted quit-rate series (JTS...QUR) for total nonfarm and the eleven supersectors — mining and logging (110099), construction (230000), manufacturing (300000), trade/transportation/utilities (400000), information (510000), financial activities (510099), professional and business services (540099), private education and health services (600000), leisure and hospitality (700000), other services (810000), and government (900000) — retrieved from the BLS API on September 9, 2026. July is held constant in both years to fix the pre-pandemic reference month.

The Analysis

The following is analysis, not fact.

The national number and the cross-section tell slightly different stories. A quit rate falling half a point, 2.4% to 1.9%, reads as broad cooling. Read across all eleven supersectors at once, the cooling is nearly total — ten of eleven at or below their pre-pandemic mark — with a single sector pointing the other way. That exception is the part a single-industry glance would miss ^sspeculative.

Mining and logging is the outlier worth naming. It is the one major industry where voluntary quitting now runs above its 2019 norm, 2.6% against 1.9%. What drives it is not in the series: a commodity-and-energy hiring cycle, wage competition for a relatively small workforce, or ordinary volatility in a small sector all fit, and the quit rate records the movement, not the cause ^sspeculative. The sector is small enough that its rate can move sharply from month to month, which is a caution as much as a finding.

The sharpest normalization is not the one usually told. Leisure and hospitality — the face of the 2021 quitting surge, once above 5% — fell 1.4 points and is still the highest-quit sector. But the largest decline in the set belongs to professional and business services, down 1.5 points to 1.8%, a category that includes temporary-help and staffing, where churn is structurally high and where a return toward the pre-pandemic baseline shows up as a large move ^sspeculative. The cross-section's real content is that "quits fell everywhere" is close but wrong: quits fell almost everywhere, and the one place they rose is the tell.

Room for Disagreement

A quit rate at or below its pre-pandemic level does not by itself signal that workers are content. Quits are read alongside job openings: thinner openings can reduce workers' outside options and therefore may contribute to lower quitting ^mmodeled. JOLTS openings have fallen from their 2022 peak, so part of the broad decline is consistent with reduced worker leverage rather than settled satisfaction ^mmodeled. The mining-and-logging rise cuts the other way and carries its own caveat: it is one month, one relatively small supersector off a low base, and JOLTS estimates are revised — an annual benchmark can narrow, widen, or unwind a single-month gap this size.

The View From

From the headline quit rate alone, July looks like uniform late-cycle cooling: 1.9%, down from 2.4%, a labor market where fewer people are walking. Set the eleven industries beside it and the same month reads as near-uniform normalization with one genuine exception — mining and logging, the lone sector quitting more than before 2020. The reassuring aggregate and the one sector moving against it are the same month, counted whole rather than in the single line most readers see.

Notable

• U.S. Bureau of Labor Statistics · Job Openings and Labor Turnover Survey — program page, methodology, and current release tables — Context link: the primary JOLTS program page, including the industry breakdown and seasonal-adjustment methodology behind the series used here.
• FRED / Federal Reserve Bank of St. Louis · JOLTS: Quits Rate — Total Nonfarm, Seasonally Adjusted (JTSQUR), December 2000 to present — Context link: the full historical national quit-rate series for readers who want the long-run trajectory behind the July reading.
• Economic Policy Institute · JOLTS indicators — monthly tracking of job openings, hires, quits, and separations — External analysis: EPI's monthly JOLTS commentary, including industry-level and labor-market-power framing of the quits data.

We read the 15 games ESPN's scoreboard lists as completed for September 9, 2026. The box score leads with hits. That column pointed to the right winner in 10 of 15 games — and the wrong winner in 4.

Four times on September 9, the team that collected more hits lost. In all 4 cases, the winning team converted its baserunner opportunities — hits plus walks plus hit batters — into runs at a higher rate. The metric that tracked the winning team in 14 of 15 games: conversion rate, computed as runs divided by total on-base events (hits + walks + hit-by-pitch).

Across those 15 games, winning teams produced 113 runs from 208 on-base opportunities: a 54.3% conversion rate. Losing teams produced 56 runs from 166 opportunities: 33.7%. That 20.6-point aggregate edge came with conversion rate tracking the winner in 14 of 15 games.

The widest hit-vs-outcome divergence came at Miami. The Mets won 15-14 with 15 hits to the Marlins' 20 and 23 total bases to Miami's 30. New York converted 65.2% of its 23 on-base events (15 hits, 6 walks, 2 hit batters) into runs. Miami converted 53.8% of 26 opportunities. The Marlins out-hit the Mets in both hits and total bases and lost by one.

Three other games matched the pattern. Kansas City beat Arizona 5-2 with 7 hits to the Diamondbacks' 9, converting 50.0% of 10 on-base events to Arizona's 15.4% of 13. Seattle beat Texas 3-2 on 6 hits to 7, 33.3% to 20.0%. The Athletics beat Toronto 2-0 on 4 hits to 6, converting 33.3% of their 6 on-base events while Toronto reached base 9 times and scored none.

Philadelphia and Houston hit identically — 12 hits apiece. Houston hit for more power: 22 total bases to Philadelphia's 19. But Philadelphia walked 4 times; Houston walked zero. Those 4 extra baserunners gave Philadelphia 16 on-base events to Houston's 12, a conversion rate of 68.8% to 58.3%, and an 11-7 win.

The one exception to the conversion pattern: the Angels beat the Red Sox 6-4 with both teams converting at exactly 40.0%. With rates identical, the larger opportunity set produced more runs — Los Angeles reached base 15 times to Boston's 10.

The Analysis

The following is analysis, not fact.

Conversion rate is not a box-score column. It is arithmetic applied to three published statistics — hits, walks, and hit batters — to measure how efficiently a lineup turned its time on base into runs. On September 9 it tracked the winning team in 14 of 15 games; hits did so in 10.

The pattern carries a consistent shape: in every game where hit counts and the final score disagreed, conversion rate agreed. In all 4 losses suffered by the team with more hits, the winner also had fewer total on-base events — so within this metric, its edge came from the higher conversion rate, not a larger opportunity set.

The Philadelphia-Houston game is the clearest case. Both teams recorded 12 hits. Houston hit for more power — 22 total bases to Philadelphia's 19. But Houston's hitters drew zero walks, compressing the Astros' own opportunity set to 12 on-base events. Philadelphia's 4 walks expanded its set to 16. With equal hits, the walk gap widened Philadelphia's opportunity set.

^sspeculative Whether September conditions amplify this effect — full bullpens, pitching staffs at depth, tighter high-leverage management — is a question one slate cannot answer. What September 9 shows is that in more than a quarter of games, conversion rate distinguished the winner where hit count did not.

Room for Disagreement

The 14-of-15 figure is a single-day snapshot across 15 games. Over a 162-game schedule, hits may be a more durable predictor, and the variance visible in one slate may wash out over hundreds of games ^sspeculative.

Conversion rate has a structural limitation: it treats every on-base event as equivalent regardless of inning, game state, or opponent. A team that converts well may have faced a depleted bullpen in high-leverage spots rather than executing better with runners on base. The metric cannot separate those effects. It also conflates pitching efficiency (limiting opponent baserunners) with offensive efficiency — a team's own conversion rate and its opponents' are set simultaneously.

Notable

• ESPN · Lindor and Morel power the Mets to a 15-14 win over the Marlins — ESPN's game story on the day's widest hit-vs-outcome gap — New York won 15-14 despite Miami's 20 hits to New York's 15.
• ESPN · Schwarber hits grand slam for 44th homer in Phillies' 11-7 win over Astros — The game where equal hit counts (12-12) and a walk gap (4 to 0) produced a 4-run difference in outcome.
• ESPN · Rookie Kade Anderson picks up first win with 6 sharp innings as Mariners beat Rangers 3-2 — Seattle won with 6 hits to Texas's 7 and a 33.3%-to-20.0% conversion advantage.
• ESPN · Henry Bolte's homer in the 4th inning helps Athletics beat Blue Jays 2-0 — The Athletics' 4-hit attack — a solo home run plus 3 other hits and 2 walks — converted 33.3% of 6 on-base events; Toronto reached base 9 times and scored none.
• ESPN · Witt homers to help the Royals beat the Diamondbacks 5-2 — Kansas City won with 7 hits to Arizona's 9, converting 50.0% of 10 on-base events to the Diamondbacks' 15.4% of 13.

We read the 45 tight ends who saw 30 or more targets in the 2024 NFL regular season, per the nflverse season log, so you didn't have to — and the scoring composition told a different story than the box score suggests.

The pool's top two finishers built their seasons on yards and catches. Brock Bowers logged 153 targets, 112 receptions, and 1,194 receiving yards — and 5 touchdowns; those touchdowns made up 11.4% of his 262.7 PPR total. Trey McBride logged 147 targets, 111 receptions, and 1,146 yards — and 3 touchdowns (two receiving, one rushing); his touchdown scoring accounted for 7.4% of his 243.8 PPR total. Travis Kelce, fifth in the pool at 195.4 PPR, scored 3 touchdowns — 9.2% of his total. The pool's top two finishers averaged 9.4% of their PPR from touchdowns.

Mark Andrews finished sixth (188.8 PPR) on 11 touchdowns and 69 targets. Those 11 scores contributed 66 points — 35.0% of his total. His non-touchdown production (122.8 points) ranked him 11th among the 45 qualifying tight ends. The five tight ends who finished sixth through tenth averaged 26.6% TD scoring — nearly three times the 9.4% average of the top two.

Across the 45-TE pool, 16.4% of PPR scoring came from touchdowns. Within the top 12, individual TD dependence ranged from 7.4% (McBride) to 35.0% (Andrews) — a 27.6-percentage-point spread. The median qualifying tight end derived 11.9% of PPR from touchdowns.

The Analysis

The following is analysis, not fact.

The conventional tight end picture centers on the touchdown — the view that the position runs fewer routes per game than wide receiver but sees more red-zone looks per touch, so an end-zone opportunity outsizes the payoff per snap ^sspeculative. The 2024 scoring composition complicates that picture at the top of the position.

Bowers and McBride each reached at least 147 targets, 111 receptions, and 1,140 receiving yards — exceeding every other qualifying tight end in the pool by more than 40 yards and more than 14 receptions. Their model is volume underneath: high catch rates, consistent opportunity, high target share from their respective offenses ^sspeculative. Touchdowns were a component, not the spine. George Kittle finished third (236.6 PPR) on 8 touchdowns, with 20.3% of his scoring from the end zone — between the two archetypes.

Andrews' profile inverts the Bowers-McBride model. His 69 targets were the fewest of any top-12 finisher — and 19 of the 45 qualifying tight ends drew more targets over the full season. What his targets disproportionately produced was touchdowns: 11 in 69 opportunities, a rate consistent with a player whose snaps concentrate near the goal line, though target location is not available in this dataset ^sspeculative. Remove the 11 touchdowns and his remaining 122.8 non-TD points rank him 11th in the pool — five spots below his actual finish.

The 27.6-percentage-point spread within a single position's top 12 is consistent with two structurally distinct production models operating inside the same fantasy position: one that accumulates a volume floor without scoring often, and one that scores enough to carry a lower volume floor ^sspeculative. Whether either model is more stable from year to year is a question one season's cross-section cannot answer ^sspeculative.

Room for Disagreement

The counter-view holds: the TE6-through-TE10 tier averaging 26.6% TD scoring is a meaningful share at any position, and touchdowns are what separate a viable TE start from a borderline one in most weeks. Bowers and McBride are exceptional cases — both operated in offenses that deployed them at target counts (153, 147) far above the pool median, and both were rookies or second-year players in breakout years. One season in which two volume-dominant tight ends finished at the top does not mean the position's traditional reliance on end-zone production has diminished. The median pool player's 11.9% TD share is lower than the TE6-10 tier, but two or three touchdowns still represent a meaningful swing in the PPR standings at the position level.

Notable

• NFL.com · Raiders' Brock Bowers sets new record for receptions by rookie TE, passes Lions' Sam LaPorta — NFL.com reporting on Bowers surpassing LaPorta's record for most receptions by a rookie tight end
• Sports Illustrated · Brock Bowers makes NFL history with another huge game for Raiders — Sports Illustrated on Bowers' record-setting 87th reception in Week 14 of the 2024 season
• ArizonaSports.com · Trey McBride behind only 1 TE in fantasy football ranking — McBride's 2024 per-game average of 15.6 PPR, 28.4% target share, and projected 7.9-touchdown expectation
• FantasyPros · 2024 Fantasy Football TE Statistics — Season-wide 2024 TE stats including targets, receptions, yards, and touchdowns for all qualifying players

Every game ESPN's MLB scoreboard lists as completed for Monday, Sept. 7, and Tuesday, Sept. 8, 2026 — 26 contests, 11 on Monday and 15 on Tuesday — was pulled inning by inning from that scoreboard API. For each game, cumulative scoring was tracked across every half-inning to find the moment the first lead was established and every time it changed hands afterward.

Of 26 games:

- **18 of 26 (69.2%)** recorded zero lead changes — the first team to take a lead was never overtaken. - **The first team to score won 21 of 26 (80.8%).** Five times the first team to put a run on the board still lost, among them Cincinnati, which scored first against the Dodgers on Sept. 7 and fell 6-3. - **13 of 26 (50.0%)** ended with a margin of exactly one run.

Inning by inning, across the 52 half-innings for each of innings 1 through 8:

- **2nd inning:** 33 total runs, 0.635 runs per half-inning — the highest among innings 1 through 8. - **4th inning:** 12 total runs, 0.231 runs per half-inning; scoreless in 45 of 52 half-innings (86.5%) — the lowest among innings 1 through 8, about a third of the 2nd inning's output.

Across innings 1 through 8, per-half-inning scoring ranged from the 4th inning's 0.231 to the 2nd inning's 0.635.

Eight games had at least one lead change: New York Mets over Miami 9-4 (2 changes, Sept. 7); Milwaukee over Chicago 4-3 (1 change, Sept. 7); Detroit over Minnesota 5-4 (1 change, Sept. 7); San Francisco over St. Louis 5-4 in eleven innings (2 changes, Sept. 7); Los Angeles over Cincinnati 6-3 (1 change, Sept. 7); Minnesota over Detroit 3-2 (2 changes, Sept. 8); Milwaukee over Chicago 4-3 in ten innings (1 change, Sept. 8); and San Diego over Washington 5-4 (1 change, Sept. 8).

Two additional records from the slate:

- Philadelphia held Atlanta scoreless through seven full innings on Sept. 7, then scored once in the bottom of the eighth to win 1-0. - Milwaukee beat Chicago 4-3 on Sept. 7. The clubs met again Sept. 8: tied 1-1 after one inning, scoreless from the second through the eighth, tied 3-3 after the ninth — Milwaukee won in the bottom of the tenth, 4-3.

The Analysis

The following is analysis, not fact. The margin column from this two-day slate looks like competitive baseball: 13 of 26 games decided by a single run, several deep into extra innings or the final third. The inning-by-inning record describes something closer to the opposite. Once a team went ahead in this sample, it rarely fell behind again: 18 of 26 games ended with the first lead standing, and the first team to score won 21 of 26.

The 4th inning offers a structural thread, though the causal direction is ^sspeculative. In this slate, that half-inning ran nearly silent: 45 of 52 iterations were scoreless, producing just 12 total runs at 0.231 per half-inning. If a stretch where a trailing team might chip into an early deficit instead generates almost nothing, early advantages face less competition than the close final scores suggest ^sspeculative.

The 2nd inning sits at the top of the production table (0.635 R/half-inn): this slate's most active scoring window arrived before many bullpen adjustments and before deeper pitchers had settled in. Runs came early — and in this sample, those early runs rarely had to be defended a second time.

The Philadelphia-Atlanta game shows the shape in its plainest form: scoreless through seven, Philadelphia scored first in the eighth and won 1-0, its lead never overtaken. The two Milwaukee-Chicago games cut the other way — Chicago scored first both nights and lost both 4-3, the second in ten innings — a reminder that the slate's reversals were real games, not rounding.

Room for Disagreement

26 games across two September days is a limited sample. Late-season rosters carry expanded bullpen depth, tired starting rotations, and clubs with diminished playoff stakes — all factors that could suppress lead reversals relative to a May or June slate. Over a full season the first-scoring team wins meaningfully more often than not, but at a rate well below this slate's 80.8% ^sspeculative; the two-day figure likely reflects both the small sample and possibly the particular mix of teams and conditions.

The 4th-inning quiet also does not by itself establish that trailing teams fail specifically in the 4th. The distribution of run-scoring across 26 games contains natural variance: a slate with more blowouts than usual would suppress scoring in the middle innings, and the 4th-inning figure could be partly an artifact of game state rather than a standalone structural feature. "Zero lead changes" also does not mean a game was decided early in the sense of scoreline comfort: Philadelphia and Atlanta were locked in a 0-0 game until the eighth, and Milwaukee and Chicago played to a 3-3 ninth-inning tie before extra innings settled it.

Notable

• ESPN · Dodgers vs. Reds — Sept. 7, 2026 — One of five games where the first team to score still lost; Cincinnati scored first and fell 6-3.
• ESPN · Phillies vs. Braves — Sept. 7, 2026 — The 1-0 contest that stayed scoreless through seven innings before Philadelphia scored in the eighth.
• ESPN · Cubs vs. Brewers (10 inn.) — Sept. 8, 2026 — The extra-inning rematch that ended 4-3 for Milwaukee a second straight night.
• ESPN · Twins vs. Tigers — Sept. 8, 2026 — One of the games with two lead changes; Minnesota won 3-2 after leading, trailing, and retaking the lead.

The Bureau of Economic Analysis publishes two figures for the size of the U.S. economy, built from independent data. Gross domestic product (GDP) adds up spending; gross domestic income (GDI) adds up earnings — wages, profits, interest, rents, taxes on production. Because one person's spending is another's income, the two are the same quantity measured from opposite sides, and in theory they are equal. The distance between them has a name: the statistical discrepancy. We pulled every quarter BEA has released — 318 of them, 1947 first quarter through 2026 second quarter — from the agency's series via the St. Louis Fed.

In the second quarter of 2026, GDP ran to $32,486.1 billion at an annual rate; GDI, $32,245.1 billion. The spending side was larger by $240.9 billion — 0.742% of GDP — the value on BEA's own statistical-discrepancy line, computed from the agency's unrounded levels; subtracting the one-decimal figures shown here gives $241.0 billion, off by $0.1 billion from rounding. In inflation-adjusted terms the gap was $180.0 billion in chained 2017 dollars.

The direction is the story. GDP has exceeded GDI for 15 consecutive quarters, from the fourth quarter of 2022 through the second quarter of 2026 — the last quarter before the run, GDI was the larger of the two. That is the longest stretch of the spending measure topping the income measure since a run that ended in 1997. Across the 15 quarters the gap averaged $279.0 billion, or 0.951% of GDP, and was widest in the third quarter of 2023 at $459.2 billion, 1.64% of output.

The persistence shows up in growth, not just levels. Since the run began, real GDP has grown 9.69%; real GDI, 8.50% — a 1.19-point gap over roughly four years, with the income side growing more slowly over the 15-quarter period. BEA, aware the two rarely match, publishes both.

The Analysis

The following is analysis, not fact. The arithmetic is settled — GDP minus GDI is $240.9 billion, and it has carried the same sign for 15 quarters — but the meaning is not, and the honest reading is narrow. The two measures should describe one economy. When the income side comes in lower quarter after quarter, the pattern is consistent with the income tally being understated, the spending-and-output tally being overstated, or a combination of the two ^sspeculative. These data do not separate those cases. The discrepancy is a residual, not a diagnosis; it says the books do not close, not why.

What is unusual here is duration, not size ^sspeculative. A single quarter's gap of three-quarters of a percent of GDP is ordinary — the country has seen wider in both directions many times. What the full record surfaces is that the sign has not flipped since 2022, and that over the run the income tally has grown more than a point less than the spending tally in real terms. Read only the headline GDP print and the expansion looks steady; read the income side beside it and the same years look a shade weaker ^sspeculative. Which of the two better tracks the economy is a genuine open question in the research, not one this table resolves — any claim that GDI is the truer signal here is a hypothesis, not a finding ^sspeculative.

Room for Disagreement

The gap does not by itself establish manipulation or identify an error in either measure — it is a known feature of measuring one thing two ways from different source data, which is why BEA reports the discrepancy openly and averages the two. Magnitude, in particular, should not be oversold: 89 of the 318 quarters on record carried a larger gap than this run's average, and the single longest stretch of GDP above GDI was not now but the 62 quarters running from 1973 into 1988. Both figures are also heavily revised — the most recent quarters most of all — so annual revisions can narrow, extend, or unwind the current streak. The levels quoted here are the present vintage, retrieved September 1, 2026. The durable claim is the one the complete series supports: the sign has held for 15 quarters and the income side has grown more slowly than GDP — not that either number is the right one.

The View From

From the topline alone, the last three years read as an ordinary expansion ^sspeculative: nominal output up, quarter after quarter of positive growth (real output rose in all but one quarter, dipping once in early 2025). Set the income measure beside it and the same span looks slightly softer and, more to the point, internally unsettled ^sspeculative — two official tallies of the same economy that have leaned the same way, income below output, for 15 quarters straight. The reassuring headline and the quieter disagreement underneath it are the same economy, counted two ways that have diverged in the same direction.

Notable

• Federal Reserve Bank of Cleveland · The Discrepancy Between Expenditure- and Income-Side Estimates of US Output — Economic Commentary explaining why GDP and GDI diverge, how the statistical discrepancy behaves, and what research says about which measure better tracks turning points — the standard reference on the gap this piece measures.
• U.S. Bureau of Labor Statistics · GDP, GDI, and GDO: an evaluation of output measures for productivity analysis — Monthly Labor Review analysis of the two output measures and their average (informally GDO), and why the choice between them matters for reading the economy.
• Marketplace · GDP should be roughly the same as gross domestic income. It's not. — Plain-language reporting on what GDI is and why it can tell a different story than GDP — external context on the same divergence.
• Thornburg Investment Management · What Does the Gap between GDP and GDI Signal for Investors? — A market-side reading of the same gap, included as an independent interpretation of what a persistent GDI shortfall may or may not imply.

We ran the MLB's own expected-win formula through all 30 clubs so you didn't have to — and one of the table's starkest splits between run margin and standing sits inside the AL East, where Tampa leads New York by 4 games despite trailing the Yankees by 70 runs in run margin.

The formula, called xWinLoss, estimates how many games each club should win based on run-scoring context ^mmodeled. Apply it to the full 2026 standings through September 7 and two clubs are winning 8-9 games ahead of what the math says, while two others are losing 10-12 more than expected ^mmodeled.

Tampa Bay (85-58) leads the AL East with a run differential of +47. The formula puts them at 76-67 — 9 games behind their actual record ^mmodeled. The Yankees trail by 4 games despite a run differential of +117: the formula pegs New York at 84-59 ^mmodeled, and they sit at 81-62, 3 games short of that expectation ^mmodeled. The gap in run differentials between the two clubs is 70 runs; the gap in the standings is 4 games, and it runs the other way.

Two other cases frame the full range. The Cincinnati Reds (69-75) have allowed 108 more runs than they've scored and should be 61-83 by the formula — 8 wins below their actual record ^mmodeled. Their one-run record is 24-14 (.632). The Los Angeles Angels (54-90) are one of the worst teams in baseball by run margin (-80), and they're losing 10 more games than even that bleak projection — the formula predicts 64-80, and they've won 10 fewer ^mmodeled. Arizona (77-68) occupies a category of its own: the Diamondbacks have scored 641 runs and allowed 641 exactly, a run differential of zero, and have still won 4 more games than a break-even scoring record should produce ^mmodeled.

Across all 30 clubs, 12 have outperformed the formula and 14 have fallen short; 4 are exactly at their projected win total ^mmodeled. The range runs from Tampa's +9 surplus to Detroit's -12 ^mmodeled, with the Braves, Brewers, and Dodgers — each with run differentials of +119, +169, and +155 — clustered within 1 win of their expected totals ^mmodeled.

The Analysis

The following is analysis, not fact. Most of the biggest overperformers carry winning records in one-run games — Tampa at 21-14 (.600), Cincinnati 24-14 (.632), Philadelphia 25-13 (.658), the Rangers 22-14 (.611), and Arizona 27-19 (.587) — though the tendency is not universal, as the exceptions below show. The four teams with deficits of 5 or more — Detroit, the Angels, Washington, and San Francisco — all have one-run records at or below .400: 14-26 (.350), 16-26 (.381), 14-21 (.400), and 13-25 (.342).

The pattern is consistent with close-game performance accounting for much of the formula gap ^sspeculative. A team that wins one-run games accumulates wins without proportional run margin — every 1-0 result counts the same in the standings as a 10-0 blowout, but contributes almost nothing to the run differential total. A team that loses one-run games sheds wins without the margin deteriorating to match. Eight of the nine biggest overperformers and underperformers combined follow the same direction in their one-run records ^mmodeled.

The Yankees case is particularly specific. Their one-run record is 17-21 (.447), losing more close games than they win, while Tampa is 21-14 (.600); both clubs have played roughly 35-38 one-run games. On run differential the formula orders New York ahead of Tampa by 8 games (expected 84-59 to 76-67); in the actual standings Tampa leads by 4 (85-58 to 81-62) — a 12-game reversal in relative position ^mmodeled. That swing runs in the same direction as their opposite one-run records, consistent with close-game performance driving most of it ^sspeculative.

Room for Disagreement

The correlation is not universal. Houston is 5 games above expectations yet 13-15 in one-run games — the inverse of the pattern — which suggests overperformance can come from sources the one-run record alone doesn't capture ^sspeculative. Pittsburgh is 4 below expectations with a 24-17 one-run record; their gap and their close-game record point in opposite directions.

Atlanta is the clearest limit case for the entire frame: the Braves are 85-59 with a +119 run differential, and their expected record is precisely 85-59. When the underlying dominance is that consistent, the formula doesn't need close games to explain anything — and the standings confirm it. Across the table, the divergence between expected and actual tends to compress as run differential grows, with the outliers concentrated in the middle of the standings rather than at the extremes ^mmodeled.

Notable

• ESPN · 2026 MLB Standings — Live standings for all 30 clubs, updated daily
• FanGraphs · 2026 MLB Standings and Projections — Expected-win projections and standings context for the full season
• Baseball Reference · 2026 Major League Baseball Season — Season-level statistics, run differentials, and historical comparisons

We read the 13 Major League Soccer games completed on September 5, 2026. The ESPN scoreboard for the date lists 14 scheduled fixtures — 13 completed and one postponed (D.C. United at FC Cincinnati) — so the 13 are the complete finished set. Combined: 47 goals, 136 shots on target, a 34.6% aggregate conversion rate.

Six finished level. In the 7 that produced a winner, the team with more shots on target won 4 and lost 3.

The three reversed finishes:

**LA Galaxy 2, New England Revolution 1.** The Galaxy put 3 shots on target and scored twice (67%). The Revolution put 7 on target and scored once (14%). LA won 2-1 at home despite being outshot on target by 4.

**FC Dallas 4, Sporting Kansas City 3.** Sporting KC had 10 shots on target — the most of any team Saturday — and scored 3 (30%). Dallas had 9 shots on target and scored 4 (44%). The team that put the most on-target efforts of any in the 13 completed games lost the match.

**Portland Timbers 5, Minnesota United 4.** Minnesota had 8 shots on target and converted 4 (50%). Portland had 5 — three fewer — and all five produced goals per ESPN's box score, anchored by a hat trick from Vincent Janssen; one of Portland's goals was credited to Minnesota goalkeeper Drake Callender as an own goal, with ESPN's tracking counting the preceding Portland effort among those 5 shots on target. Portland won 5-4.

In the 4 decided games where the shot-on-target leader won, the losing side failed to convert any on-target attempts in 3 of them: Columbus beat Colorado 3-0, Philadelphia beat Montreal 2-0, and Orlando beat San Diego 1-0.

Two of the six draws produced no goals despite multiple on-target attempts: Seattle put 6 shots on target against New York Red Bulls and scored none (0-0); Houston put 5 against Charlotte and scored none (0-0).

The Analysis

The following is analysis, not fact.

Shots on target is what the MLS broadcast reaches for when it wants to describe control — the effort that cleared the first line of defense, that wasn't wide or high or blocked, that reached a goalkeeper. By that measure, Sporting Kansas City had the best day of any team across the 13 completed games. They lost.

Eight teams across Saturday's 13 games put 6 or more shots on target without winning: KC (10), RSL (9), Toronto (8), Minnesota (8), New England (7), Seattle (6), Chicago (6), and LAFC (6). KC, Minnesota, and New England lost outright; the other five drew. On this slate, that on-target volume did not translate into wins.

The Galaxy-Revolution result is the sharpest edge. New England hit 2.3 times as many shots on target and scored once. The Galaxy hit 3 and scored twice. The conversion gap — not the volume gap — separated the result. ^sspeculative

Portland is the volume paradox in miniature: Minnesota generated more on-target attempts across the full 90 minutes and lost by a goal. In the two 0-0 draws, high shot-on-target output (6 for Seattle, 5 for Houston) produced nothing at all. Across all 13 games, 34.6% of on-target efforts became goals — but that aggregate conceals a range from 0% to 100% at the per-team level on the day.

Finishing rate separated 3 of 7 Saturday winners from the team that had put more shots on target. ^sspeculative

Room for Disagreement

Seven decided games is a thin basis for a conversion-rate argument. The likeliest explanation is ordinary finishing variance — conversion rates are streaky, and a single round of fixtures says little about what holds across a season; establishing how unusual a 4-of-7 slate actually is would require season-level shots-on-target and outcome data this piece does not include. The stronger claim this data supports is narrower: on this particular slate, shot count alone did not settle results.

Notable

• Yahoo Sports · Vincent Janssen scores first 3 MLS goals in Timbers' wild 5-4 victory over Minnesota United — Match report on Janssen's hat trick and the 5-4 result
• Yahoo Sports · Portland Timbers 5 - 4 Minnesota United: Final score, results, recap, box score, stats — Full box score and recap for the Portland-Minnesota match
• Yahoo Sports · LA Galaxy Score Solid 2-1 Home Win — Recap of the Galaxy win over New England, the slate's starkest shot-reversal
• ESPN · Soccer Scores - September 5, 2026 — Full September 5 MLS scoreboard
• Fox Sports · LA Galaxy vs. New England Revolution - September 05, 2026 boxscore — Detailed boxscore for the Galaxy-Revolution result

The Federal Reserve's G.19 Consumer Credit release, published monthly and aggregated in FRED series REVOLSL, tracks revolving consumer credit outstanding — primarily credit card balances. After growing 27.6 percent from January 2022 ($1.043 trillion) to June 2024 ($1.331 trillion), the series' year-over-year growth turned negative in December 2024. Outstanding revolving credit came in below its year-earlier level every month from December 2024 through November 2025 — twelve consecutive readings. The widest gap was October 2025, when balances ran $35.4 billion — 2.62 percent — below October 2024's $1.352 trillion.

The reversal ended in December 2025. By June 2026, revolving credit stood at $1.351 trillion, growing 3.8 percent year-over-year and nearly back to the October 2024 level.

Monthly year-over-year changes, computed from FRED REVOLSL: December 2024 −0.15% · January 2025 −0.49% · February −1.16% · March −2.13% · April −1.78% · May −2.33% · June −2.19% · July −2.04% · August −2.38% · September −2.22% · October −2.62% · November 2025 −1.92%. December 2025: +2.11% (ending the run).

Non-revolving credit — auto loans, student loans, and other fixed-term borrowing — grew throughout: 1.3 percent year-over-year to June 2025 and 2.0 percent to June 2026 (FRED NONREVSL). Total consumer credit (FRED TOTALSL) grew 0.4 percent year-over-year to June 2025, its slowest pace since the pandemic.

A separate FRED series — DRCCLACBS, the delinquency rate on credit card loans at all commercial banks — moved in the same arc but on a different ledger and at a different pace. That rate hit a trough of 1.53 percent in Q3 2021, peaked at 3.22 percent in Q2 2024, and retreated to 2.85 percent by Q2 2026. The Q2 2024 peak was the highest quarterly reading since 2012, per the Kansas City Fed.

Scope note: REVOLSL covers revolving consumer credit across commercial banks, credit unions, finance companies, and federal programs. DRCCLACBS covers commercial bank delinquency only — a subset of the REVOLSL lender universe; the two series are not directly comparable as rates. The G.19 was revised in July 2025 to exclude nonfinancial business revolving credit; FRED REVOLSL reflects the revised historical data throughout.

The Analysis

The following is analysis, not fact.

The 12-month streak sits in the same class as two other periods in the FRED REVOLSL series: the post-2008 financial-crisis paydown, which extended for roughly three years, and the COVID-19 shutdowns, which coincided with a sharp decline across 2020. Between 2013 and 2019 — outside of those events — the series grew year-over-year in every comparable month. In the REVOLSL series since 1968, sustained multi-month year-over-year declines had previously appeared only in the post-2008 paydown and the 2020 shutdowns; the 2025 reversal shows the same directional sign without such a downturn ^mmodeled.

The delinquency rate's arc — peaking in Q2 2024, roughly six months before the year-over-year balance decline widened — is consistent with ^sspeculative a charge-off sequence: elevated delinquencies can generate charge-offs, which would reduce outstanding balances, working through the G.19 series with a lag. But the aggregate data cannot decompose this from voluntary paydown — consumers reducing balances through accelerated payments — or from tighter lending standards — banks issuing fewer new cards or at lower limits. Each of the three mechanisms carries different implications for household financial health; none is separately visible in the G.19 total.

The 2026 recovery — 3.8 percent year-over-year to June — began while commercial bank delinquency rates were still declining (2.91 percent in Q1 2026, 2.85 percent in Q2). Whether this reflects renewed borrowing demand, normalization of the post-reversal baseline, or a lagged effect of easing lending standards cannot be established from these two series alone.

Room for Disagreement

The reversal does not undo the post-pandemic expansion. Revolving credit at $1.351 trillion in June 2026 stands 37 percent above the June 2021 level of $985 billion and 29.5 percent above January 2022. The delinquency rate at 2.85 percent, while declining, remains above the pre-pandemic 2019 range of 2.54–2.61 percent and well above the 2021 trough of 1.53 percent. Analysts who emphasize distributional credit burdens argue that the balance reversal reflects stress rather than discipline ^sspeculative: on this reading charge-offs and tightened standards did much of the work, while underlying demand never materially changed. On that reading, the 2026 recovery is simply borrowing resuming at a smaller lender base and higher rates — not a structural improvement.

Notable

• Bankrate · Credit Card Balances Fell In Q1 2025. Borrowers Aren't Out Of The Woods Yet — Covers the Q1 2025 G.19 release; context on delinquency and interest-rate burden
• Federal Reserve Bank of Kansas City · Subprime Credit Card Delinquencies Have Fallen — Confirms the Q2 2024 peak was the highest delinquency reading since 2012
• Wolf Street · Credit Card Delinquencies, Balances, Burden, Credit Limits, and Collections in Q3 2025 — Independent analysis of the same series through Q3 2025
• PYMNTS · Revolving Credit Drops 12 Percent in November as Consumers Trim Balances — Covers the November 2025 G.19 release, the sharpest annualized monthly drop in the streak
• Eye on Housing (NAHB) · Credit Card and Auto Loan Balances Continue to Slow — Parallel analysis of REVOLSL and NONREVSL through November 2025
• eMarketer · Credit card charge-offs totaled $46B through Q3 2024 — Reports commercial bank charge-off volume in the quarters leading into the balance reversal
• Philadelphia Fed · Large Bank Credit Card and Mortgage Data 2025 Q1 Narrative — Lender-level delinquency and charge-off data for Q1 2025

In May 2012, all 194 WHO member states endorsed resolution WHA65.6, which set six global nutrition targets for 2025. Target 2 was a collective goal: a 50% reduction in anemia among women of reproductive age — defined as hemoglobin below 12 g/dL for non-pregnant and below 11 g/dL for pregnant women ages 15–49 — measured against 2012 levels. It was a target for global prevalence, not a per-country quota; countries set national targets aligned to it. The 2025 deadline has passed; the latest WHO country estimates run through 2023.

WHO's Global Health Observatory publishes modeled annual prevalence estimates for this exact population; the World Bank carries the series as SH.ANM.ALLW.ZS. Every country anemia-prevalence figure in this piece is a WHO modeled estimate from that series — modeled, not directly measured. Of the World Bank's 217 economies, 192 have estimates for both the 2012 baseline year and 2023 — the most recent year available; 25 lack one or both. Measured country by country against the 50% benchmark, not one ^mmodeled of the 192 had cut its own rate in half by 2023. The Philippines recorded the deepest cut: 18.7% to 12.0% ^mmodeled, a 35.8% reduction — the best of the 192 economies with both-year data, and still about 14 percentage points short of a halving. Sri Lanka was next at −18.0%. Every other economy either improved by less or moved in the wrong direction.

152 of 192 economies (79%) ^mmodeled recorded a higher anemia prevalence in 2023 than in 2012. The unweighted median across the 192 — a Claridas aggregation of WHO country estimates — rose from 22.0% to 25.5% ^mmodeled. WHO's own population-weighted global prevalence, the metric the target actually names, has been roughly flat to rising rather than the 50% cut the goal sought ^mmodeled.

The Analysis

The following is analysis, not fact.

The benchmark was a single global 50% cut. Read country by country against it, the 192-economy panel — every income tier, every region — shows none reaching it by 2023.

Among the 53 economies that entered 2012 with the heaviest burden — anemia rates above 30% — not one ^mmodeled recorded a 50% reduction by 2023. The median change in that group was 0.0% ^mmodeled: flat, not declining. Twenty-six of those 53 worsened. India — at a 53.7% modeled anemia rate among its women of reproductive age — moved from 50.1% to 53.7% ^mmodeled, a 7.2% increase; halving its own rate would have meant reaching 25.1%, leaving it 28.6 percentage points above a halved rate. Afghanistan moved from 36.5% to 45.4% ^mmodeled, the sharpest deterioration among high-burden economies.

Among the high-burden group, Gambia (−13.2%), Senegal (−13.1%), Nigeria (−12.9%), and Ghana (−12.8%) ^mmodeled led the improvers — each a genuine decline over the period, and each achieving roughly one-quarter of the 50% benchmark. The pattern across the full panel points to a gap between the pace the global target assumed and what the underlying nutrition systems delivered [speculative — the data shows the outcome, not the mechanism].

Room for Disagreement

Large increases in high-income economies — Canada's modeled rate rose from 7.7% to 14.0%, France's from 8.0% to 13.1%, Germany's from 9.1% to 14.0%, Australia's from 6.9% to 11.1% ^mmodeled — warrant caution. WHO substantially revised its anemia estimation methodology in 2021, and the retrospective 2012 values now in the series may differ from the baseline estimates that existed when WHA65.6 was adopted. If the revision lowered historical baselines for some economies and raised current-year estimates, the observed increase in high-income countries may be partly a measurement artifact. The headline finding — none of the 192 cut its rate in half by 2023 — is unaffected: no high-income economy with single-digit baseline anemia halved its rate, which from those baselines would have required reaching 3–5%. But the 152-of-192 count includes these revisions; the fraction of economies that genuinely deteriorated is likely lower ^sspeculative. Proponents of global nutrition targets argue that without the 2012 resolution, political and financial attention to anemia would have been further reduced — and that the Philippines' 30%+ reduction suggests such a cut is achievable under concentrated intervention ^sspeculative.

The View From

From India's position, the gap between the goal and the outcome is especially large in absolute terms. With a 2023 modeled rate of 53.7% across an estimated 350 million women of reproductive age (UN population data) — roughly 190 million affected ^mmodeled — India carries a large absolute burden. A policymaker in New Delhi would likely point to the structural scale of India's anemia programs — iron supplementation, deworming, and dietary diversification across 1.4 billion people — as evidence that the 2025 timeline was insufficient for a problem rooted in diet, healthcare access, and food systems that take decades to shift [speculative — the WHO data shows the gap, not its cause].

Notable

• WHO · Comprehensive implementation plan on maternal, infant and young child nutrition (WHA resolution 65.6) — The 2012 resolution that set the 50% anemia target — the primary commitment measured here.
• WHO Global Health Observatory · Anaemia in women of reproductive age — global estimates — The underlying tracker; country-level modeled estimates and methodology notes.
• World Bank · Prevalence of anemia among women of reproductive age (% of women ages 15-49) — SH.ANM.ALLW.ZS — Re-pullable source for the 2012 and 2023 country panels used in this analysis.

A widely shared X thread — a Norwegian post by DNB Asset Management portfolio manager and Tesla bull Audun Wickstrand Iversen (@IversAudun), quote-tweeting a Tesla site in San Antonio — argues that robotaxis will make car ownership, parking, accidents, insurance, tickets, DUIs and even snow plowing 'gradually disappear' over 10 to 15 years. Its load-bearing number is a per-mile cost of 20 to 30 cents, set against roughly 70 cents for a private car and 200-plus cents for a human-driven taxi. We read the thread line by line against the record and pulled the actual fares.

The 20-cent figure traces to Elon Musk at Tesla's 'We, Robot' Cybercab unveiling in October 2024, where he put operating cost at 'probably around 20 cents a mile,' explicitly 'over time' and 'at scale' [marketing]. It is a projection, which no fleet tracked here has demonstrated.

The largest at-scale, fully driverless paid US fleet tracked here is Waymo, which reported about 500,000 paid rides per week across roughly 10 metros as of March 2026, up about tenfold from May 2024 [verified, company-disclosed]. Analyst estimates put Waymo's fully loaded cost near $1.40/mile today [modeled, analyst estimate, no company disclosure] — roughly 5 to 7 times the 20-to-30-cent target. Tesla's own Austin robotaxi fares, from two legible receipts in the X thread under review, were $6.99 for 2.5 miles and $6.84 for 2.4 miles — about $2.80/mile on those two short trips, base fare included [reported, operator-captured]. Tesla's metered Austin rate runs nearer $1.25 to $1.40/mile on longer trips [reported].

The private-car '70 cents' matches the IRS 2025 standard mileage rate of 70.0 cents and the AAA 2025 full-ownership average of 77 cents at 15,000 miles/year. Both are full-cost figures. The marginal cost of driving a mile in a car you already own — fuel plus maintenance — is about 24 cents [modeled from AAA components]. Rideshare and metered taxis run roughly $2 to $4/mile as a retail price, not a cost [reported].

Tesla's Austin service is geofenced to about 245 square miles and remains predominantly supervised: as of March 31, 2026, only about 4 to 8 of roughly 37 to 42 Austin vehicles ran without an in-car monitor, and those are remotely supervised. In San Antonio, a Zoning Commission voted 8-2 on July 21, 2026 to recommend rezoning a parking-and-charging depot near the Pearl in Tobin Hill — city staff had recommended against, City Council had not voted, and it authorizes no driverless operation.

The Analysis

The following is analysis, not fact.

The thread's core comparison appears to mismatch its terms three ways ^mmodeled. First, cost versus price: the 70-cent car is a cost, the 200-cent taxi is a retail price that includes a driver's wages and a platform margin. Second, full versus marginal: the honest 'should I drive or ride' number for a car you own is about 24 cents, which is within the 20-to-30-cent target range, not above it. Third, demonstrated versus projected: the car and taxi numbers are measured today, while 20 to 30 cents remains an unproven future target ^mmodeled. Compare like with like and the story inverts — today a car costs about 77 cents/mile all-in versus an estimated $1.40 for a robotaxi, and two Austin receipts read about $2.80.

The 20-to-30-cent figure is a forecast, not a measurement. At-scale manufacturing has driven costs down before, so the target is not far-fetched ^sspeculative. But no fleet in this review has demonstrated it, and the fleet closest to scale appears not yet profitable ^mmodeled.

The 'everything disappears' framing reads as a category error ^sspeculative. These functions are more likely to restructure than vanish ^sspeculative: liability would shift from millions of individual auto policies toward concentrated operator and manufacturer product-liability — a different insurance market, not the absence of one ^sspeculative. Robotaxis still crash: Waymo recalled 3,067 vehicles in December 2025 over software that let them pass stopped school buses, under an open NHTSA investigation. Snow may be the thread's weakest claim ^sspeculative: heavy snow appears to remain a significant operating limitation for the systems tracked here ^mmodeled — Waymo pauses in heavy snow, and a robotaxi that stops in a blizzard could mean more plowing, not less ^sspeculative. And we identified no peer-reviewed measured decline in car ownership or miles driven attributable to robotaxis — the honest answer is that it is too early. Modeling actually points the other way, toward more miles via empty repositioning and induced demand ^mmodeled.

The economics argue for a slower, ODD-limited path ^sspeculative. Alphabet's Other Bets, which houses Waymo, posted a 2025 operating loss of about $7.5 billion [verified, SEC segment]; Sundar Pichai has said he expects Waymo to contribute meaningfully to Alphabet's financials as soon as 2027 [reported] — a contribution target, not a profitability guarantee. Cruise, once among the largest US robotaxi operators, spent more than $10 billion, admitted a false report to regulators over an October 2023 pedestrian-dragging and paid a $500,000 criminal fine, and GM stopped funding it in December 2024. The timeline record is worse still: at Tesla's April 2019 Autonomy Day, Musk said there would be 'over a million robotaxis' on the road the next year. Waymo eventually delivered a real service — this is not vaporware — but the field's hard timelines have repeatedly slipped.

Room for Disagreement

The strongest honest case for the thread is that the tech is real and improving fast, and that costs at scale historically fall. Waymo's roughly tenfold ridership growth in under two years is not a slide-deck number — it is paid rides on public streets, and peer-reviewed work (Kusano et al., 2025, in Traffic Injury Prevention) found about 79 to 81 percent fewer injury crashes for the driverless system, a large effect even after discounting for the authors being Waymo employees and the operating domain excluding freeways and severe weather. If Cybercab reaches purpose-built, mass-produced, single-purpose vehicles running dense city loops, per-mile costs genuinely could fall toward the marginal cost of a private car — and at that point ownership economics in a few dense metros really would come under pressure ^sspeculative. Fewer humans owning idle cars could reduce parking demand ^mmodeled, the one second-order effect with a plausible mechanism. The direction the thread points is not crazy; the record simply does not yet show the arrival. A lower crash rate within a geofence is not the elimination of crashes; a pilot in 10 metros is not a finished national transition; and the front-runner apparently not yet profitable ^mmodeled while a major competitor shut down points toward gradual, city-by-city, weather-limited expansion ^sspeculative — not the wholesale disappearance of cars, parking, tickets and plows on a 15-year clock.

The View From

From China, the framing looks less like a Tesla story and more like a two-country race ^sspeculative. Baidu's Apollo Go reported a weekly-order peak above 300,000 in late 2025 and more than 20 million cumulative rides as of its February 2026 earnings, across roughly 26 cities [reported] — a driverless network of a scale that appears comparable to Waymo's ^mmodeled, built under a different regulatory posture. The at-scale question the thread treats as settled looks, on the global evidence, still contested in two markets at once ^sspeculative.

Notable

• TechCrunch · Waymo's skyrocketing ridership in one chart — Company-disclosed ~500,000 paid rides/week across ~10 US cities as of March 2026, up ~10x from May 2024.
• Electrek · Tesla expands unsupervised 'Robotaxi' area in Austin with only a handful of cars — Only ~4-8 of ~37-42 Austin vehicles ran without an in-car monitor; 'unsupervised' cars are still remotely supervised.
• San Antonio Report · Tesla robotaxis could come to San Antonio as Waymo prepares for a return — Zoning Commission 8-2 recommendation (July 21, 2026) to rezone a charging/parking depot near the Pearl; city staff recommended against; Council had not voted.
• CBS News · Waymo recalls more than 3,000 vehicles over faulty software following school bus violations — 3,067 vehicles recalled over software that let robotaxis pass stopped school buses; NHTSA probe opened in October.
• CBS News · Cruise admits to false report in 2023 dragging of San Francisco pedestrian — Cruise (GM) admitted a false report over the Oct. 2, 2023 pedestrian-dragging and agreed to a $500,000 criminal fine.
• NBC News · Musk claims Tesla will have 1 million robotaxis on roads next year — At Tesla's April 2019 Autonomy Day, Musk projected over a million robotaxis the following year; it did not happen.

We read all 15 MLB games played on September 2, 2026.

Five went to extra innings. The 2026 season rate through that date, across 2,098 completed games, was 8.8% — 184 extra-inning games. On September 2 the rate was 33%.

Every one of the five games that needed extra innings was tied after nine regulation frames. Four of the five were already knotted before the ninth inning began. Minnesota and Detroit sat at 5 apiece after eight innings. St. Louis and Los Angeles at 5. New York and Los Angeles at 1. San Francisco and Pittsburgh at 3. In none of those four games had either team taken a lead into the ninth inning.

Baltimore was the exception. The Orioles led Colorado 3-2 entering the ninth, and the Rockies scored once to force extras. Colorado won in 11.

Away teams won 4 of the 5 extra-inning games: the Giants over Pittsburgh in 10 innings, Detroit over Minnesota in 12, the Yankees over Los Angeles in 10, the Cardinals over Los Angeles in 10. Colorado's walk-off in the 11th was the lone home win.

When those games finally broke open, scoring was not modest. Three of the five ended in the first extra frame, each with multiple runs scored. The Yankees put up 5 in the 10th to go from 1-1 to 6-3. The Cardinals scored 3 to turn a 5-5 tie into an 8-6 win. Minnesota and Detroit required two more innings: both teams scored in the 11th, then Spencer Torkelson's home run opened a 5-run 12th for Detroit, ending a game that had been level 5-5 for the last two regulation innings. The 5 games combined for 26 runs across 8 extra frames.

The ten regulation-game results ranged from Toronto 11, Cleveland 0 at one extreme to Philadelphia 0, Arizona 1 at the other. Four teams scored zero runs in regulation: Washington (in a 9-0 Atlanta blowout), Cleveland, Philadelphia, and Chicago's White Sox.

The Analysis

The following is analysis, not fact.

Five games on a 15-game slate reaching extra innings tied through regulation is statistically anomalous. Given the season's 8.8% extra-inning base rate, the probability of seeing five or more such games in a random 15-game draw is roughly 0.7% ^mmodeled. That makes September 2 an outlier on paper — not a trend, one day out of a 2,430-game schedule.

The pattern inside those five games is worth noting on its own: four pairs arrived at the ninth inning already level, so those ties were not created by a ninth-inning equalizer — the score was even before the final regulation frame. Whether that differs materially from the usual extra-inning game is not something this slate can establish: we have the season's extra-inning rate, not the inning-by-inning path of the other 184 games.

September roster rules allow expanded bullpens, which could concentrate specialist-matchup usage and, in turn, might suppress scoring in regulation's final innings ^sspeculative. Whether that contributed to five simultaneous regulation ties, or whether September 2 was simply an unusual binomial outcome, is not something one slate can distinguish.

Room for Disagreement

A statistician would lead with the sample size. Fifteen games is a single daily draw from a 2,430-game season, and binomial variance on any one slate is large. Getting five extra-inning games when the season rate implies 1.3 is genuinely rare — roughly 1-in-134 slate days at the season rate — but rare is not impossible. Whether five simultaneous tied-through-regulation games reflects something structural about late-season baseball or is simply what a 0.7% outcome looks like when it occurs, a single September day cannot say.

Notable

• ESPN · Torkelson homers to start 5-run 12th, Tigers top Twins 11-6 — Game recap for the 12-inning Detroit-Minnesota game
• ESPN · Cardinals at Dodgers game summary, September 2, 2026 (10 innings) — STL 8, LAD 6 in 10 innings
• ESPN · Yankees at Angels game summary, September 2, 2026 (10 innings) — NYY 6, LAA 3 in 10 innings; Yankees scored 5 in the 10th
• ESPN · Giants at Pirates game summary, September 2, 2026 (10 innings) — SF 5, PIT 4 in 10 innings
• ESPN · Orioles at Rockies game summary, September 2, 2026 (11 innings) — Colorado walk-off win 6-5; only home team to win in extras that day

In May 2015, the World Health Assembly — all WHO member states — adopted the Global Technical Strategy for Malaria 2016–2030. The strategy established an intermediate 2025 milestone: cut global malaria incidence and mortality rates by at least 75% compared to 2015. The 2025 checkpoint was designed as an early signal of whether the eventual 90% reduction target, due in 2030, was reachable. The strategy was updated in 2021 and its milestones were maintained.

The WHO Global Health Observatory tracks malaria incidence as estimated cases per 1,000 population at risk — the metric the strategy uses. The December 2025 vintage of that series, underpinning the World Malaria Report 2025 published the same month, covers data through 2024.

The global malaria incidence rate stood at 59.01 per 1,000 population at risk in 2015. By 2024 it had risen to 63.95 — an 8.4% increase. Meeting the 75% reduction milestone would have required the rate to fall to 14.75 or below. The actual 2024 rate is 4.3 times that threshold.

Among the 89 countries that recorded a positive 2015 incidence rate and have a 2024 WHO estimate, 16 cleared the 75% reduction milestone. The other 73 did not: 34 saw their rates rise, and 39 declined but by less than 75%. [verified, WHO GHO MALARIA_EST_INCIDENCE, December 2025 vintage]

The same pattern holds for deaths. The WHO World Malaria Report 2025 estimates 610,000 malaria deaths globally in 2024, up from 438,000 in 2015 ^mmodeled — a 39% increase in absolute deaths. The report states the 2024 malaria mortality rate is more than three times the 2025 milestone target.

Of the three GTS milestones, the elimination target was already met: the strategy called for malaria eliminated in at least 10 countries by 2025, and since its adoption 14 countries have received WHO malaria-free certification — including China (2021), El Salvador (2021), Belize (2023), Cabo Verde (2024), and Georgia (2025). For the two rate milestones, the latest available 2024 data left the world far off the 2025 targets. [verified, WHO certification record]

The Analysis

The following is analysis, not fact.

The global trajectory appears to be heavily shaped by a single region ^mmodeled. The WHO African Region carries 94% of the world's malaria cases, and its own incidence rate barely moved — from 243.2 per 1,000 in 2015 to 237.6 in 2024, a 2.3% decline against a 75% target.

The 20 countries with the highest 2015 incidence rates are all in Africa, each above 260 per 1,000. Among them, 4 — Democratic Republic of Congo, Nigeria, Burundi, and Malawi — saw their rates increase. Of the 16 that declined, only Rwanda approached the milestone, cutting its rate by 74.96% — from 308.04 to 77.14 per 1,000, a fraction below the threshold; the rest ranged from under 4% (Cameroon) to 46% (Liberia). Rwanda's result is consistent with ^sspeculative sustained national investment in surveillance, treatment coverage, and net distribution across the same period.

The 16 countries that cleared the 75% milestone divide into two groups. Twelve had 2015 incidence rates below 1 per 1,000 at risk — they were already approaching elimination. The 4 that cleared it from higher baselines — 2015 rates above 1 per 1,000 at risk — were Cambodia (19.7 → near zero), Laos (17.7 → near zero), Zimbabwe (94 → 11, −88%), and India (6.4 → 1.5, −77%). The WHO South-East Asia Region moved from 6.2 to 1.9 per 1,000 — a 69% decline; India, the region's dominant population, cut its own rate 77% (6.4 → 1.5) over the same span.

The Eastern Mediterranean ran in the opposite direction: its regional rate more than doubled, from 9.0 to 19.1 per 1,000. Sudan's rate rose 127%, Yemen's 101%, Pakistan's 173%. The pattern is consistent with ^sspeculative the collapse of national malaria control programs in conflict-affected settings.

The global trajectory also marks a pivot year. From 2015 through 2019, the global rate edged modestly lower, from 59.0 to 58.0 per 1,000. Since 2019 it has risen 10%, reaching 63.95 by 2024. Against that trajectory — the latest WHO data runs through 2024 — the world was left far off track for the 2025 milestone, not just in the final measured year but across the decade.

Room for Disagreement

The steelmanned counter is structural: the 75% milestone was designed without anticipating COVID-19. The pandemic disrupted bed-net distribution, suppressed care-seeking, and diverted health infrastructure across high-burden countries — on that reading, the visible plateau and reversal in global incidence around 2020 is not simply an accountability failure ^sspeculative. Rwanda's 74.96% reduction — missing the milestone by 0.04 percentage points from a 2015 rate of 308.04 per 1,000 — is consistent with the target having been nearly attainable from high endemicity in at least one case ^sspeculative; a single country cannot establish that it was generally feasible. The WHO World Malaria Report 2025 attributes part of recent progress to new tools — including the R21 vaccine rolled out in high-burden countries from 2023. The WHO's 2030 target of 90% reduction remains in force; the 2025 checkpoint is a course-correction signal for a decade that has not ended.

The View From

Health ministries across the WHO African Region operate with the funding shortfall in plain view: the WHO World Malaria Report 2025 puts available malaria financing in 2024 at $3.9 billion against the GTS target of $9.3 billion — less than half. That shortfall is a global figure, not a per-country measure — but it is the fiscal backdrop against which high-burden health ministries run their malaria programs. From that position, a 75% milestone set in 2015 can read as a commitment that was never fully provisioned ^sspeculative. The strategy's architects would likely counter that the funding gap is itself the failure — not an excuse for missing the target it was designed to fund ^sspeculative.

Notable

• World Health Organization · New tools saved a million lives from malaria last year but progress under threat as drug resistance rises — The WHO's own December 2025 press release on the World Malaria Report — the institution that set the milestone assesses its own progress.
• Target Malaria · World Malaria Report 2025 — Independent malaria-research partnership summarizing the report's main findings — context on how the science community reads the milestone miss.
• United to Beat Malaria · 2025 World Malaria Report: What You Need to Know — Advocacy-network summary of the report aimed at donor governments — the frame through which Western funders are hearing these numbers.
• Medicines for Malaria Venture · Malaria facts and statistics 2025 — Product-development nonprofit's public fact sheet — independent compilation against which the WHO series can be cross-referenced.
• European Medical Journal · Global Malaria Report 2025: Progress and Challenges — Clinical-audience summary that states the mortality rate figure (13.8 per 100,000) and its relation to the 4.5 milestone target — the number context this piece uses.

Across the 19 completed games from the first two playing dates of the 2026 college football season — August 29 and September 3 — the team that eventually loses accounts for 16.9% of first-quarter scoring and 39.3% of fourth-quarter scoring. The loser's share nearly doubles across the game, from 34 of 201 total Q1 points to 107 of 272 total Q4 points.

Ten of the 19 losing teams scored more in Q4 than they did in Q1. The sharpest case: San José State trailed USC 28-0 at halftime and 35-3 after three quarters. The Spartans scored 23 of their 26 points in Q4 and the final score read 42-26. The margin at the whistle was 16 points. The margin after three quarters had been 32.

In the 10 games whose halftime margin exceeded 14 — effectively settled before intermission — the trailing teams accounted for 5.3% of first-quarter scoring (7 of 133 points) and 39.4% of fourth-quarter scoring (63 of 160 points). Akron trailed Wake Forest 31-3 through three quarters and scored 13 of its 16 season-opening points in the final 15 minutes; the box score final of 38-16 suggests a competitive second half that wasn't playing for the lead. Bethune-Cookman was shut out in the first three quarters at UCF and scored 6 of its 6 total points in Q4 of a 73-6 game.

The Q4 surge is not limited to true blowouts. Hawaii trailed Stanford 23-7 entering the fourth quarter and scored 20 points — a genuine effort at a comeback — before falling 37-27. In the 6 competitive games (final margin 14 or fewer), the eventual loser accounted for 44% of Q4 scoring — higher than in the decided games — because those fourth quarters were genuine contests rather than structured damage control.

The full 19-game sample spans the Week 0 Saturday slate of August 29 and a midweek Thursday, September 3 card — not the marquee Week 1 Saturday slate of September 5, typically the highest-profile and most evenly matched, which had not been played at time of pull. The figures here reflect the full population of games ESPN tracked on those two opening dates: 8 on August 29, 11 on September 3, all completed.

The Analysis

The following is analysis, not fact.

The fourth-quarter loser-share gap in the decided games is consistent with a structural dynamic that repeats every Week 1: when winning teams protect a lead late, they shorten the game and substitute players, while losing teams — with starters still in, no deficit pressure on clock management — continue to move the ball on a less-experienced defensive front ^mmodeled. The data cannot separate coached strategy from personnel-driven opportunity, but the 7.4x shift in the loser's share from Q1 to Q4 in the already-decided games is consistent with that mechanism, not with genuine late-game competition ^sspeculative.

The college football box score is a final-score instrument. It reports how many points each team scored, not when they scored relative to the game's inflection point. A game that reads 42-26 and one that reads 73-6 both produced different fourth-quarter performances — one by a team that moved the ball after the game was over, and one where both teams effectively stopped. The average final margin across these 19 games was 24.8 points; the median was 19. Neither figure tells you whether the fourth quarter mattered to the outcome.

The Q4 figure also matters for cumulative scoring statistics. Every touchdown scored in garbage time enters the season-opening totals for points allowed, yards gained, and scoring average — numbers that recruiting, rankings, and conference metrics will cite. A program that allows 23 fourth-quarter points to a team it led by 32 will see a different defensive average than one that shuts down entirely. The separation only becomes visible in the quarter-by-quarter data.

Room for Disagreement

The sample is 19 games from the Week 0 and midweek openers, which skew toward the scheduled mismatches programs book for home-guarantee revenue — weaker opponents invited for the first week paycheck. If the marquee Week 1 Saturday slate runs more balanced, the loser share gap will likely compress. The 44% loser share in competitive close games (final margin 14 or fewer) is actually higher than in decided games — Hawaii's 20-point fourth quarter against Stanford was a genuine comeback attempt, not padding, and those points read correctly as live-game scoring. And some Q4 "loser" scoring occurs in games where neither team scores much (both UCF and BCU went quiet after UCF's third-quarter surge), so the share reflects a mutual late-game fade as much as a losing-team surge.

Notable

• ESPN · San Jose State at USC — Game Summary, August 29 2026 — ESPN game report for the SJSU-USC matchup; Spartans scored 23 of 26 Q4 points after trailing 35-3.
• ESPN · Bethune-Cookman at UCF — Game Summary, September 3 2026 — ESPN game report for the 73-6 result; BCU scored all 6 of its points in Q4 after being shut out in Q1-Q3.
• ESPN · Colorado at Georgia Tech — Game Summary, September 3 2026 — ESPN game report for the 14-13 contest; representative of genuine Q4 competition versus garbage-time scoring.
• College Football Reference · College Football Statistics and History — CONTEXT: Primary statistics database for college football; season totals incorporate cumulative Week 1 figures.
• Football Outsiders · College Football Analytics — CONTEXT: Analytics publication covering scoring efficiency and situational metrics in college and professional football.

In 2024, just over half of all consumer complaints filed with the Consumer Financial Protection Bureau were resolved with some form of relief — a corrected account, a waived fee, a reversed charge. By 2025, after complaint volume nearly doubled, that share was 41 percent.

The bureau's public complaint database records every filing since 2011 and exposes the full record through a keyless API. Querying it by calendar year: 800,245 complaints in 2022; 1,292,049 in 2023; 2,734,269 in 2024; 5,442,964 in 2025 — a 6.8-fold increase over three years. Through September 4, 2026, an additional 4,863,672 have been filed.

The concentration is extreme. In 2025, 4,810,302 of the 5,442,964 complaints — 88.4% — were credit-reporting disputes. Three companies absorbed the bulk of those: Equifax drew 1,680,536 complaints, TransUnion 1,601,849, and Experian 1,388,821. Together, those three accounted for 4,671,206 complaints — 85.8% of all filings for the year.

Complaints are submitted through the bureau's public portal. The CFPB requires companies to respond; it does not validate whether a complaint is accurate or the underlying claim is meritorious. Of the 5,442,677 resolved 2025 complaints (287 were still in progress as of the September 2026 pull), 26,125 — 0.48%, or 1 in 208 — closed with monetary relief. Another 2,212,234 (40.6%) closed with non-monetary relief such as account corrections or record updates, and 3,194,721 (58.7%) closed with an explanation from the company and no remedial action. Those three outcomes account for 5,433,080 of the resolved complaints; the remaining 9,597 (0.2%) closed under other dispositions the database records separately.

For comparison, in 2024: monetary relief applied to 23,745 of 2,734,269 resolved complaints (0.87%); any form of relief to 50.9%; explanation only to 49.0%. Through September 4 of 2026, of the 4,248,240 complaints already resolved (615,432 remain in progress), monetary relief applied to 20,887 (0.49%) and any form of relief to 29.4%, with 70.3% closed by explanation only. The 2026 figures will shift as in-progress complaints close.

The Analysis

The following is analysis, not fact.

Reading the full four-year complaint record surfaces a single directional pattern: as volume multiplied, the fraction of filings producing any substantive resolution declined. The any-relief rate moved from 50.9% in 2024 to 41.1% in 2025; the explanation-only share of resolved complaints rose from 49% (2024) to 59% (2025). In 2026's resolved complaints — still an incomplete year — explanation-only has reached 70%, though the in-progress pool will shift that figure.

The data establish the pattern but cannot isolate its cause, and at least two readings are consistent with the numbers ^sspeculative. Under one ^sspeculative, the shift reflects a change in bureau posture: reduced staffing, fewer formal enforcement actions, and lighter institutional pressure on companies to provide substantive relief. Under another ^sspeculative, the shift is a compositional artifact: a growing share of credit-reporting complaints are filed by third-party credit-repair services on clients' behalf — a practice the bureau's own annual reports have noted — and those filings may be more likely to yield an explanation than a correction. A higher proportion of non-meritorious or bulk-filed complaints would reduce the any-relief rate arithmetically even with no change in company behavior or bureau enforcement.

No causal test was conducted here. No filing-source variable was controlled. The data do not distinguish a consumer-filed complaint from one filed by a credit-repair service on a consumer's behalf, and so the relative weight of these two readings cannot be resolved from this dataset alone. What the record establishes without ambiguity: the explanation-only share of resolved complaints rose in 2025 — from 49% to 59% — alongside a complaint population that roughly doubled in each of the last two years.

Room for Disagreement

Consumer advocates read the falling relief rate as evidence of a bureau operating below capacity at a moment of the highest annual complaint volume in the 2022-2025 record queried here. The CFPB shed most of its examination staff in early 2025 under a new administration; critics argue that companies facing less formal oversight have less incentive to provide substantive resolution. The credit bureaus and their industry representatives contest this reading: they point to the large volume of credit-repair-service submissions as the primary driver of both the complaint surge and the declining relief rate, arguing that disputes filed on consumers' behalf by third-party services tend to target information the bureaus are legally entitled to report. The empirical gap is genuine — the current dataset does not allow attribution — but advocates note that the explanation-only rate was falling even before the staffing changes, and the bureau's own consumer response reports have previously noted that non-consumer originators account for a substantial share of credit-reporting submissions, without attributing the resolution mix to that composition.

The View From

Equifax, TransUnion, and Experian each contend that their high complaint volumes reflect the size of the credit-reporting industry and the role of third-party credit-repair services, not the prevalence of bureau error. Their published responses to the bureau have consistently described the bulk of disputed items as information they are obligated by creditor reporting to maintain. Those public responses do not address the 2025 aggregate resolution figures reported here.

Notable

• ProPublica · Experian and TransUnion Are Leaving More Mistakes on Credit Reports — Investigation into the persistence of credit report errors after consumer disputes
• ProPublica · Lawmakers Demand Answers About Unfixed Credit Report Mistakes — Congressional response to the credit-bureau accuracy investigation
• ProPublica · Bilt Fiascos Raise Concerns About Trump CFPB's Enforcement Policies — Reporting on CFPB enforcement posture under the current administration

At the 10th Conference of the Parties to the Convention on Biological Diversity — held in Nagoya, Japan, in October 2010 — 196 governments adopted Aichi Biodiversity Target 11: conserve at least 17% of the world's terrestrial and inland-water areas by 2020. The target set a single global figure and a firm deadline; it did not assign each country its own 17% quota — governments set national targets separately. The deadline is five years past, which makes the target a fixed benchmark to measure the world against.

The World Database on Protected Areas (WDPA), maintained by IUCN and UNEP-WCMC, is the international ledger for formally designated protected areas. The World Bank indicator ER.LND.PTLD.ZS — protected areas as a share of total land area — draws from it; it is the standard proxy for the target, though the target's scope is terrestrial and inland-water areas together, a slightly broader base. Measured against the 17% benchmark, most economies with data sit below it. Of the World Bank's 217 economies, 212 reported a 2020 value — only the Channel Islands, Macao, Nauru, San Marino, and Kosovo did not — so this is a near-complete census, not a sample. Among the 212, 95 are at or above 17% and 117, or 55%, fall below. Among the 191 of those economies that are CBD parties, the split is 90 above and 101 below (53%). The median economy protected 15.6% of its land in the deadline year, 1.4 percentage points below the benchmark.

Across those 212 economies, the gap runs through every income group but falls unevenly. Among high-income economies, 50 of 82 cleared 17%. Among upper-middle-income, 17 of 58. Among lower-middle-income, 22 of 47. Among low-income, 6 of 25. Several major economies landed below the line: China at 15.6%, India at 6.0%, Russia at 11.5%, Indonesia at 12.2%, Mexico at 14.5%, Argentina at 8.5%, South Africa at 8.6%. Among those that cleared it: Brazil at 30.3%, Germany at 37.8%, Japan at 29.4%, France at 27.3%, the United Kingdom at 28.7%.

Seventy-six economies — more than a third of all 212 with data — protected less than 10% of their land in 2020. By 2022, two years after the deadline, the count had barely shifted: 97 of 213 economies had reached 17%.

The Analysis

The following is analysis, not fact.

The income-level pattern holds but imperfectly. That 61% of high-income economies cleared the bar against 24% of low-income economies points to administrative capacity as a factor — designating, mapping, and reporting protected areas requires resources and institutional continuity ^sspeculative. But 32 of 82 high-income economies also missed, and some low-income economies like Zambia (41.3%) and Bhutan (49.7%) cleared it by margins that eclipse wealthy neighbors.

The subtler issue is what 'cleared' actually measures. The WDPA counts areas that governments have formally designated and submitted to the international database. IUCN protection categories range from strictly managed nature reserves to areas where sustainable use is permitted. A government reporting 22% protected land may have filed extensive paperwork while a different government at 14% enforces tighter boundaries within a smaller fraction. The Aichi target specified 'effectively and equitably managed' areas — a standard that a percentage alone cannot capture.

In December 2022 — at COP15 in Montreal — the same governments adopted a successor: the Kunming-Montreal Global Biodiversity Framework, with a 30% terrestrial target by 2030. The bar nearly doubled while the 17% benchmark still sat unmet by most: 55% of economies with data were below it in 2020, when 76 protected less than 10% of their land; 68 remained below 10% in 2022.

Room for Disagreement

The per-economy count weights Seychelles (61.5% of 451 km²) and China (15.6% of 9.6 million km²) identically. In land-area terms, the world's largest states diverge: Brazil (30.3%) and Australia (20.3%) together cover significant territory above the threshold. But Russia, China, and the United States — among the largest countries by land area — are all below it, at 11.5%, 15.6%, and 11.8% (the United States is not a CBD party). The unweighted mean of economy percentages (17.6%) exceeds the 17% threshold; an area-weighted average, if computed, would likely not ^sspeculative.

The most searching counter holds that coverage is not the right measure for biodiversity outcomes. The CBD's own Global Biodiversity Outlook 5 (2020) found none of the 20 Aichi targets fully achieved — including Target 11 — and noted that protected-area extent without effective management does not reliably conserve biodiversity. Some economies above 17% may shelter less functional habitat than some below it ^sspeculative.

The View From

View from India: India protected 6.0% of its land in the deadline year, well below the 17% benchmark. As we read India's own position ^sspeculative, the government's argument is qualitative rather than quantitative: its protected areas are concentrated in globally recognized tiger reserves and wildlife sanctuaries — small in extent, intensive in management — rather than broadly designated across a larger fraction of territory. Whether that approach conserves more biodiversity per hectare than a government reporting 20% of loosely managed land is a real empirical question. It is not the question the Aichi threshold was designed to answer.

Notable

• The Guardian · Biodiversity: landmark deal agreed at COP15 in Montreal to protect nature — Reports the Kunming-Montreal 30% target, adopted after the 2020 Aichi 17% deadline passed with the benchmark unmet.
• Convention on Biological Diversity · Global Biodiversity Outlook 5 (2020) — The CBD's own assessment finding no Aichi target fully achieved by 2020, including Target 11 on protected areas.
• Protected Planet / UNEP-WCMC · World Database on Protected Areas — The source database underlying the World Bank terrestrial-protection indicator used in this analysis.

In 2013 the World Health Assembly — all 194 WHO member states — endorsed the Global Action Plan for the Prevention and Control of Noncommunicable Diseases. Its fifth voluntary target is a single, falsifiable number: a 30% relative reduction in the prevalence of current tobacco use among people aged 15 and older by 2025, measured against a 2010 baseline. The deadline year has now passed.

The World Bank carries the WHO's own estimate of that prevalence — series SH.PRV.SMOK, 'Prevalence of current tobacco use (% of adults),' compiled by the WHO and last updated 2026-07-13. We read the 2010 and latest-available (2024) value for every economy. These are WHO modeled trend estimates, not counted surveys, so every figure below is tagged accordingly. ^mmodeled

Globally, the WHO series falls from 27.5% of adults in 2010 to 20.3% in 2024 — a 26% relative decline. A 30% cut would have required roughly 19.2% by 2025. The world is short. ^mmodeled WHO's own headline figure, age-standardized, tracks the same shape — 26.2% in 2010 to 19.5% in 2024, about a 27% relative fall — and the agency projects the 30% target will not be reached until around 2029. ^mmodeled

Restricting to the 140 economies over a million people that report both years, 48 hit the 30% cut, 78 declined but fell short, and 14 saw tobacco use rise. The unweighted median fall was 23.8%. ^mmodeled

Wealth does not sort the progress. By income group, the median relative change runs: low income −38.8% (12 of 18 hit the target), lower-middle −29.2% (17 of 34), upper-middle −18.2% (just 5 of 41), high income −19.4% (14 of 47). The low-income bloc's median fall alone already clears the 30% bar; the upper-middle-income group is furthest from it, and the high-income group is barely better. ^mmodeled

The 14 economies where smoking rose cluster in one region: seven are in the Middle East and North Africa — Lebanon (+21%), West Bank and Gaza (+17%), Egypt (+16%), Oman (+9%), Algeria (+8%), Jordan (+4%) and Saudi Arabia (+1%). The single largest increase is the Republic of Congo (+26%); Portugal (+13%) is the only sizeable high-income riser. Among the biggest single-market populations, China (−9%) and Indonesia (−5%) barely moved, while India (−41%), the United States (−40%) and the United Kingdom (−47%) fell steeply. ^mmodeled

The Analysis

The following is analysis, not fact. Read every economy's trend rather than the global headline, and the intuition that the countries with the most resources would be furthest along inverts. The two lowest-income groups lead the field; the upper-middle-income world — the emerging economies with money to spend on public health — is the laggard bloc, and the richest group ranks third of four. Income is not doing the sorting.

One caution keeps this honest: the global −26% is population-weighted, while the −23.8% median is an unweighted count across economies over a million people. They land close together but answer different questions — how the world's smokers fared versus how the typical country did. ^mmodeled And the reduction leaders span the wealth scale — Norway (−52%), New Zealand (−49%) and the United Kingdom (−47%) sit beside Paraguay (−60%), Rwanda (−60%) and Sierra Leone (−55%). Steep progress in both rich and poor countries is a pattern these estimates show, not a mechanism they isolate: we joined no measure of taxation, advertising bans, or treaty implementation, so why some blocs moved and others stalled is a hypothesis, not a finding. ^sspeculative

What the numbers do settle is narrower and firmer: a target 194 governments signed will be missed globally, the miss is not concentrated where a resource story predicts, and in 14 economies — half of them in one region — the trend ran the wrong way entirely.

Room for Disagreement

The load-bearing caveat is in the data class. These are WHO modeled estimates, and the model leans hardest where survey data is thin — which is exactly the low-income group posting the largest declines. That bloc's apparent lead is the softest figure here; a skeptic can fairly argue it reflects modeling as much as measured behavior. The 30% figure is also a global voluntary target — measuring each country against the same bar is our analytical frame, not a per-nation legal pledge, though the WHO urged countries to set aligned national targets. And 2024 is one year short of the 2025 finish line; the final scorecard will move at the margin. What survives all of that is the direction: a global miss, and an income gradient that runs opposite to the obvious guess. ^sspeculative

The View From

From a health ministry in a high-prevalence Middle Eastern state, as we read it, the ranking would land as an unwelcome outlier rather than a verdict. Several of these governments have leaned on tobacco excise for revenue even while endorsing the reduction target, and would likely stress what the panel also shows — that the target is voluntary and that comparable data in the region is itself contested. The vantage from those capitals is that a rising line a distant desk files as policy failure looks, up close, like a collision between a public-health pledge and a fiscal habit — a reading these estimates are consistent with, not one they prove.

Notable

• UN News · WHO: Despite smoking decline, tobacco still hooks one in five adults worldwide — The UN's own coverage of the October 2025 WHO trends report — confirms the world is off the 30% pace, with prevalence at 19.5% in 2024 on the age-standardized measure.
• Pan American Health Organization · WHO tobacco trends report: 1 in 5 adults still addicted to tobacco — The WHO regional office's summary of the same report, noting the Americas already cleared a 36% relative reduction — independent corroboration of the wide regional spread.
• Tobacco Asia · Global Tobacco Use Reduction Off-Target — Trade-press framing of the missed 2025 target and the slow pace in the Western Pacific — the industry-facing read on the same shortfall.
• News-Medical · New WHO report highlights tobacco and e-cigarette trends — Coverage adding the vaping dimension the prevalence series does not capture — context for why the headline decline may understate total nicotine use.

We pulled the complete 2025 regular-season play-by-play from nflverse and read every rushing attempt charged to a quarterback — the 38 who threw at least 200 passes, the full-time-starter pool. Each carry falls into one of two kinds the box score sums into a single line: a designed run the offense called, or a scramble the charting flags as a quarterback bailing out of a broken pass play. We set kneel-downs aside — 433 of them across the pool, clock plays, not rushing offense — which is why our per-quarterback totals run a touch under the season aggregate that folds them in: Josh Allen's 94 designed-and-scramble carries plus 18 kneels equal the 112 the aggregate lists, and his 599 rushing yards less 20 lost to those kneels equal its 579.

Across the 38, the two halves are lopsided in opposite directions. Scrambles produced 7,072 of the pool's 8,801 rushing yards — 80% — on 944 of the 1,552 carries, gaining 7.49 yards a pop. Designed runs produced 1,729 yards on 608 carries, just 2.84 apiece — but they produced the scores: 58 of the pool's 97 quarterback rushing touchdowns, 60%, on carries that averaged under three yards. Priced on the two positive rushing inputs alone — 0.1 point per yard, 6 per rushing touchdown, a rushing figure identical in standard, half-PPR and PPR because no reception enters it — scrambles account for 941 of the pool's 1,462 rushing fantasy points and designed runs 521.

Josh Allen led all 38 with 143.9 rushing points, and his line is the split in miniature: 426 of his 599 rushing yards came scrambling, but 11 of his 14 rushing touchdowns came on designed runs — the short-yardage keeper. Jalen Hurts scored 6 of his 8 on called runs and took 54 of his 94 carries by design (57.4%). Patrick Mahomes is the mirror: 72.9 rushing points, but 4 of his 56 carries were designed (7.1%) and 420 of his 429 rushing yards came improvising. Among the 13 quarterbacks with 40 or more non-kneel carries, the designed-run share of carries ran from Mahomes's 7.1% to Justin Fields's 59.4% (41 of 69). Fields (62.7 rushing points) and Mahomes (72.9) finished about 10 points apart on the ground and built it from opposite ends.

The Analysis

The following is analysis, not fact. A quarterback's rushing line is two different things wearing one number. One is a play the coordinator drew up — a keeper, a sneak, a read-option give kept — initiated by the offense before the snap. The other is what happens when a pass play collapses and the quarterback takes off, initiated by the pass rush. The nflverse charting separates them; the counting stat a single-team manager reads does not. So the phrase "rushing quarterback," the thing drafts pay a premium for as a floor, resolves into two mechanisms that in 2025 pointed opposite ways: the yards lived in scrambles, and the touchdowns — the six-point events that decide a fantasy week — lived in the called runs inside the five.

That is why Allen and Mahomes read so differently under the same heading. Allen's rushing points are anchored by a designed goal-line role his offense chose to give him, 11 scores on it; the yards on top are scramble variance. Mahomes reached nearly as many rushing points with almost no designed role at all — 7.1% of his carries — meaning essentially all of it came off broken plays. Which of those two inputs repeats from one season to the next is not a question a single year can answer, and we are not claiming it here ^mmodeled. What one year can show is that the pool's rushing production is not one signal but two, sorted by who started the run.

Room for Disagreement

This is a description of how 2025's quarterback rushing points were assembled, not a forecast and not a draft board ^mmodeled. Whether a designed-run role is stickier than a scramble rate — the intuition that a schemed goal-line keeper repeats while broken-play yards scatter — is a claim about the future this one season does not test, and outside analysts who model it find both inputs predictive, with only a modest edge to designed runs ^mmodeled. The 200-attempt cutoff is a choice; a lower line would add part-time runners and could shift the pool's edges, though the named quarterbacks sit far from any boundary. Game script muddies the split, too: a quarterback trailing scrambles more and one leading gets more called runs to bleed clock, so a single season's designed-versus-scramble mix carries some of each offense's win-loss shape, not only its intent ^sspeculative. And nothing here says which quarterback to draft — only that two players can reach the same rushing total from opposite halves of the position.

The View From

A manager who rostered a rushing quarterback in 2025 saw a rushing total and a floor, and mostly got one. The pool read is the part a single roster cannot show: that the floor was built two different ways, and that the yards a scrambler piles up and the touchdowns a schemed short-yardage runner scores are not the same asset stacked under one number.

Notable

• Fantasy Points · Statistically Significant: Scrambles/QB Runs — External analysis separating designed quarterback runs from scrambles and weighing each as a fantasy predictor.
• Pro Football Focus · Ranking the NFL's best scrambling quarterbacks using PFF data — Independent charting of scramble production by quarterback — the improvisational half of the split described here.
• The Huddle · Analytical Darlings - The Best Predictors of Success at QB and RB — External take that rushing inputs top the quarterback predictor list, with designed runs edging scrambles.
• Wikipedia · Quarterback scramble — Context link: the definitional line between a scramble and a designed run used to classify carries here.
• nflreadr · Play-by-play data dictionary — qb_scramble, rush_attempt, qb_kneel, rushing_yards, rush_touchdown — Context link: field definitions for the qb_scramble flag and rushing columns re-derived in The Facts.

The World Bank carries an intentional-homicide rate for nearly every economy — victims per 100,000 people, series VC.IHR.PSRC.P5, compiled from national criminal-justice and public-health records by the UN Office on Drugs and Crime, last updated 2026-07-13. We read the latest available value for every economy; 196 non-aggregate economies report one. Values are recorded statistics, not model estimates, but they carry two caveats worth stating up front: the latest year differs by country (most of the load-bearing figures here are 2022 or 2023, a few are older), and national definitions and recording practices are not perfectly comparable.

To keep the ranking from being driven by tiny populations where a handful of killings swings the rate, we restrict every comparison below to the 144 economies with at least a million people.

In September 2015 all UN member states adopted Sustainable Development Target 16.1: 'significantly reduce all forms of violence and related death rates everywhere,' measured by the intentional-homicide rate. The pledge word is 'everywhere.' The record is not everywhere.

Of the 20 highest homicide rates among economies over a million people, 13 are in Latin America and the Caribbean; of the top 10, eight are. The list runs Jamaica 49.4 (2023), Ecuador 45.7 (2023), South Africa 43.7 (2022), Haiti 41.2 (2023) and Trinidad and Tobago 40.4 (2022). Twenty-five of these economies post a rate above 10 per 100,000, and 16 of the 25 are in the region.

The regional gap is stark. Latin America and the Caribbean (23 economies over a million people) has a median rate of 12.6; the rest of the world (121 economies) has a median of 1.7, and the global median over-a-million figure is 2.2 — a roughly seven-to-one gap between the region and everywhere else. These medians are our calculation from the World Bank series as of its 2026-07-13 update.

Income does not sort the ranking the way the poverty-drives-crime intuition predicts. Across the 144 economies the correlation between the homicide rate and (log) national income per capita is only -0.20 — weak. High-income economies do sit lower as a group (median 0.9), but the three lower income tiers barely separate from one another: upper-middle 3.2, lower-middle 4.1, low income 4.2. The world's poorest region, sub-Saharan Africa, carries a median of 5.0 — less than half Latin America's.

The cleanest cut isolates the region from the income tier. Among upper-middle-income economies over a million people, the 12 in Latin America and the Caribbean have a median homicide rate of 15.1; the 29 upper-middle-income economies elsewhere have a median of 2.2 — the same income class, seven times the violence.

The Analysis

The following is analysis, not fact. Read every economy's rate rather than the handful a desk usually names, and the geography of lethal violence separates far more sharply by region than by wealth. Income groups run from 0.9 (high income) to about 4 across the three poorer tiers — a modest, roughly four-fold spread that does not even fall cleanly from rich to poor. Regions run from 1.1 (Europe and Central Asia) to 12.6 (Latin America and the Caribbean), an eleven-fold spread. Region is doing the work income is often assumed to do.

The point that the region, not the income line, is the axis holds inside a single income tier: Latin America's upper-middle-income economies are seven times as lethal as upper-middle-income economies everywhere else. It holds at the top of the wealth scale too — the region's high-income members are not spared. Trinidad and Tobago (40.4), Costa Rica (17.7), Panama (11.7) and Uruguay (11.2) all sit far above the global high-income median of 0.9. Whatever concentrates homicide in this hemisphere is not simply the absence of money.

What the ranking does not settle is why. UNODC and the wider literature attribute the region's rate to organized crime and illicit drug markets — the agency's 2023 study reports Latin America and the Caribbean had the highest share of homicides tied to organized crime of any region. ^mmodeled That is a reading these rates are consistent with; it is not one this panel isolates. We joined no predictors — no measure of trafficking, state capacity, firearm availability or inequality — so the concentration is the finding, and the mechanism behind it is a hypothesis, not something the numbers here demonstrate. ^sspeculative

One caution against reading a rate as a body count: this piece ranks intensity, not scale. A rate says nothing about which countries carry the largest absolute number of killings — naming those would need the counts and the populations, which a per-100,000 figure does not supply.

Room for Disagreement

The load-bearing caveats are in the data. The homicide figures come from different latest years — most of the countries anchoring the regional story report 2022 or 2023, but a few values elsewhere are older and were left out of the load-bearing claims. Cross-country comparability is imperfect: definitions of intentional homicide and the completeness of recording vary, and weaker vital-registration systems in some poorer countries could understate their rates, flattering the contrast. And a rate is not a count.

What survives those caveats is not any single decimal but the shape robust to the population filter: a seven-to-one regional gap, a top-of-table dominated by Latin America and the Caribbean, and a within-income-tier gap that no GNI-based wealth measure explains. A skeptic can fairly argue the true global map is fuzzier than any one year's numbers — but would struggle to relocate the concentration away from this hemisphere. ^sspeculative

The View From

From a public-security ministry in the region, as we read it, the ranking is not news and not a poverty story. Several of these governments already frame their homicide burden around organized crime and transnational drug markets rather than deprivation, and would likely note what this panel shows plainly: their middle-income and even high-income neighbors carry rates that the same income brackets elsewhere in the world do not. The vantage from those capitals is that a figure a distant desk might file under 'the developing world' is, on the ground, a regional phenomenon that tracks trafficking geography more than any national income line — an attributed reading, not one these rates settle on their own.

Notable

• InSight Crime · InSight Crime's 2024 Homicide Round-Up — The trade press's annual country-by-country tally for Latin America and the Caribbean, reporting a regional median around 20.2 per 100,000 and over 121,000 murders in 2024 — independent corroboration of the concentration read here.
• UN Office on Drugs and Crime · Global Study on Homicide 2023 — Executive Summary — The compiler's own analysis: Latin America and the Caribbean has the world's highest subregional homicide rate and the highest share of killings tied to organized crime — the source of the attributed mechanism this piece declines to assert as fact.
• Inter-American Development Bank · Citizen Security in Latin America and the Caribbean — A development-bank framing of the region as home to a disproportionate share of the world's homicides despite a small share of its population — context for why the burden is regional rather than simply economic.
• International Monetary Fund · Violent Crime and Insecurity in Latin America and the Caribbean — A Macroeconomic Perspective — IMF departmental paper treating the region's violence as a structural economic drag, independent analysis of the same concentration from a macroeconomic angle.

The three-point shot is the loudest idea in basketball, and the WNBA has spent this season leaning into it — the Golden State Valkyries alone fire a league-leading 30.6 threes a game. We read the shooting record of all 15 teams through the games of Aug 30, each 39 or 40 games deep, essentially the entire 2026 regular season to date. What tracks a team's record is not how many threes it takes, but how accurately it shoots them.

Measure a team's shot diet as the share of its field-goal attempts launched from three. Across the league that share runs from 28.6% (Connecticut) to 44.9% (Golden State), a median of 37.9%. Line that share up against each team's winning percentage and the two barely move together: the Pearson correlation is 0.06, and a tie-aware rank correlation (Spearman) is 0.04 — effectively no relationship.

The four most three-reliant offenses make the point on their own. Golden State (44.9% of its shots from three) is 28-11; the New York Liberty (43.3%) are 24-16; the Portland Fire (43.1%) are 16-23; the Toronto Tempo (41.3%) are 11-29. The same shot diet sits atop the standings and near the bottom. Minnesota, which owns the league's best record at 31-8, ranks 12th of 15 in three-point rate — it takes a smaller share of its shots from deep than all but three teams.

Now measure accuracy instead. Three-point percentage runs from 29.3% (Connecticut) to 39.5% (Indiana), a median of 34.0%. Against winning percentage it lines up at 0.61 (Pearson), and the tie-aware rank version agrees at 0.63 (Spearman) — a moderate, consistent link. The three most accurate shooting teams are Indiana (39.5%, 26-14), Minnesota (39.3%, 31-8) and Las Vegas (35.5%, 27-13); the least accurate are Connecticut (29.3%, 10-29), Washington (31.2%, 24-16) and Chicago (31.7%, 15-25).

Raw made-three volume lands in between — it correlates 0.35 with winning, more than shot diet does but well short of accuracy, because a made three folds accuracy into volume — and its loudest cases run the other way: Portland ranks fourth in the league in three-pointers made yet sits at 16-23, and Toronto ranks fifth at 11-29.

The Analysis

The following is analysis, not fact.

The three-point revolution reached the WNBA as a volume story — take more of the most efficient shot — and this season that volume is nearly universal: 13 of 15 teams launch at least a third of their shots from deep. What the whole-league read shows is that the volume dial, the one every broadcast fixates on, is not the dial that separates this season's contenders from its also-rans ^mmodeled. The teams that shoot the highest share of threes span almost the entire standings.

A fair objection is that the accuracy link is partly mechanical: making a higher share of your threes literally puts more points on the board, so some of that 0.61 is arithmetic rather than strategy ^mmodeled. That is exactly why the flat result on rate is the more informative one. Rate is closer to a choice than accuracy is — how a team elects to distribute its shots, though game state, defensive pressure, personnel and shot availability all shape it too — and it carries almost no information about whether the team wins ^mmodeled. Accuracy is less a lever a coach pulls than a readout of shot quality and personnel: open looks, the right shooters taking them ^mmodeled. The association is real; the causation is not something these season totals can settle ^sspeculative.

Washington is the instructive outlier. The Mystics take the second-smallest share of threes in the league and shoot them the second-worst, yet sit at 24-16 — a reminder that a shooting profile is one column of a team, not the team, and that defense and shot-making inside the arc carry records this data does not measure ^mmodeled.

Room for Disagreement

A 15-team, roughly 40-game snapshot is a small sample, and a mid-season correlation across the league is a different question from what wins a title ^mmodeled. The volume case is strongest at the very top: the league's current front-runners sit in the upper half of threes made even when their shot-diet rank varies — Golden State ranks second in threes made and Minnesota sixth, though each takes a middling share of its shots from deep. Winning percentage also folds in defense, health and schedule, none of which a shooting table captures, so a low rate-to-wins correlation does not show that volume is worthless — only that, in a league where nearly everyone shoots a lot of threes, how many a team fires no longer distinguishes it ^mmodeled. What survives is the arithmetic: across all 15 teams, three-point rate correlates 0.06 with winning and three-point accuracy 0.61.

The View From

Watch Golden State or New York bomb away and win, and the lesson looks obvious — shoot more threes. Read all 15 shot charts and the same shot diet turns up at 28-11 and at 11-29. The volume is everywhere now; what still separates teams is the part the highlight reel already shows you going in.

Notable

• ClutchPoints · 5 stats that have shaped Valkyries' impressive 2026 season — Reports Golden State's league-leading three-point volume (30.6 attempts per game), the top of the shot-diet table analyzed here.
• Yahoo Sports · WNBA power rankings: Who are the league's best 3-point shooters in 2026? — External coverage of 2026 three-point shooting across all 15 teams, framed by volume and efficiency.
• WNBA.com · 2026 WNBA Power Rankings: Week 11 — League context on the standings and current form referenced in the win-percentage comparison.

A fantasy manager sizing up a receiver reads two things about his offense: is it good, and does it throw. Neither is the number that decided whether that receiver was worth starting in 2025. Read all 32 passing offenses in nflverse's complete 2025 regular-season file and the figure that tracked a lead receiver's fantasy output was how concentrated his team's targets were — the share commanded by its single busiest pass-catcher — its alpha.

That top share ran from 35.8% (Seattle, Jaxon Smith-Njigba) down to 14.0% (the New York Jets, Adonai Mitchell), a league median of 22.5%. Six of the 32 lead targets weren't wide receivers at all — five tight ends and Christian McCaffrey. Rank the offenses by that top share and the lead receiver's full-PPR points follow it: the correlation between a team's top target share and its alpha's fantasy total was 0.74, against just 0.38 for the raw number of passes the team threw.

The split is cleanest at the ends. The 11 most target-concentrated offenses (25.4% and up) returned a startable finish — top-24 among receivers or top-12 among tight ends — from all 11 of their lead pass-catchers, averaging 278 PPR points; the 10 most spread offenses (21.1% and down) returned one from just 3 of 10, averaging 160. New Orleans and Buffalo make the point directly: Buffalo's passing game charged 504 targets to New Orleans' 474 — more volume — but Buffalo spread them, its leader Khalil Shakir seeing 19.8% and finishing WR36 (166 PPR), while New Orleans funneled 27.6% to Chris Olave, WR6 (268) on fewer total team targets. Green Bay, another high-volume passing offense, left its leader (Romeo Doubs, 18.5%) at WR37.

The Analysis

The following is analysis, not fact. The opportunity-versus-output spine of fantasy usually gets read one player at a time: my receiver saw eight targets, yours saw ten. The 2025 league file reframes it as a division problem. Targets are a roughly fixed pie inside each offense; what a manager is really buying is a slice of that pie, and the slices are cut very differently. A 20% share of a busy offense and a 20% share of a quiet one sit closer to each other than either does to a 30% share on the next depth chart ^mmodeled.

That is why real-life offensive quality is a poor guide to fantasy startability. Buffalo and Green Bay ran two of the league's more productive passing attacks and produced no receiver worth starting; Buffalo's production split four ways, Green Bay's five. The concentration a manager can read straight off a depth chart — one clear alpha versus a rotation — tracked the lead receiver's season-long PPR production more tightly than the offense's raw volume did ^mmodeled. The whole-league read surfaces what a single roster hides: the useful question isn't how good the offense is, it's how few mouths it feeds.

Room for Disagreement

A big slice of a broken passing game is still limited. Minnesota handed Justin Jefferson the league's second-highest target share, 30.1%, and he finished WR21; the Giants' Wan'Dale Robinson drew 27.8% and finished WR14 — concentration bought them opportunity, but a struggling quarterback and offense were associated with a lower return, the second pillar analysts attach to any WR1 case alongside target share. The 0.74 correlation is real but partial: about 45% of the variance in alpha PPR is unexplained, sitting in efficiency, health and touchdowns. 'Startable' is also a line that moves with league size — a 14.0% share plays differently in a shallow 10-team league than a deep one — and these are completed 2025 season totals, not a forecast of how any 2026 target tree will be cut.

The View From

From inside one roster, a touch is a touch and a good offense looks like a good bet. From the whole 2025 league, what most separated an ownable lead pass-catcher from an unstartable one looked to be concentration more than the quality or volume of the offense around him ^mmodeled — Buffalo threw more than New Orleans and had no one worth starting.

Notable

• DraftSharks · 2025 NFL Target Leaders: What They Mean for 2026 Fantasy Drafts — Trade-press treatment of the same 2025 data through total targets, target share and targets per route — the player-level view this piece rolls up to the offense level.
• Pro Football Network · Bills WRs Fantasy Outlook 2025: Khalil Shakir, Keon Coleman, Joshua Palmer — External reporting on the exact mechanism at the spread end of the finding — Buffalo entering 2025 with no clear alpha and targets distributed across the room.
• Yahoo Sports · The top 16 most likely candidates to be fantasy football's overall WR1 in 2026 — States the two pillars this piece separates — a dominant target share and an efficient offense to convert it — the second being the counter-view for Jefferson's concentrated-but-capped 2025.
• FantasyPros · 2025 NFL Targets By Team — Data tool (context, not reporting) presenting 2025 target distribution team by team — an independent surface for the same per-offense concentration measured here.

The FedEx Cup finale is down to 30 players with no cut, everyone starting level at even par — straight stroke play at East Lake Golf Club in Atlanta. We scored every hole of every player who has completed all three rounds: 29 golfers, 1,566 holes. (J.J. Spaun withdrew after five holes Thursday and is excluded; the field is 30.) Through 54 holes the group stands a combined 231 strokes under par.

Nearly half of that came on two holes. East Lake's par-5s — the 6th and the 18th, 11% of the course — account for 112 of the 231 under-par strokes, one shy of what all twelve par-4s produced combined (113). The four par-3s, across 348 plays, ran just six strokes under par — near-even against the course.

On the par-5s the field is close to automatic. Of 174 plays, 111 finished birdie or better — a 63.8% rate — with four eagles and just three scores worse than par. The par-5 6th surrendered one bogey in 87 tries and played to 4.28 against a par of 5, a birdie or better 71% of the time. The home 18th, also a par-5, drew 49 birdies-or-better in 87 plays and two bogeys.

The rest of the course held. The hardest hole at East Lake is the par-4 1st, which played 0.18 over par — 21 bogeys-or-worse against six birdies. The par-3s barely moved: the 9th and 11th played fractionally over par, the 15th fractionally under. Viktor Hovland leads at 15 under (195) after a third-round 65, one clear of Ryan Gerard, with Scottie Scheffler, Chris Gotterup, Ludvig Åberg and Adam Scott at 12 under. Every player in the top eight is at least three under par on the two par-5s.

The Analysis

The following is analysis, not fact.

Golf's own scoring conventions predict that par-5s play under par — they are the reachable holes, the designed birdie chances. What the whole-course read surfaces is the magnitude: two holes did as much for the field's ledger as the twelve par-4s did, and the four par-3s did essentially nothing. On 16 of East Lake's 18 holes the best 30 players in the sport were, in aggregate, within a fraction of a stroke of par. Two holes, the 6th and the 18th, made nearly half the board's red numbers — 112 of 231 strokes, as much as all twelve par-4s combined.

That reframes what a level-start, no-cut finale rewards. When every hole but two plays near even for the whole field, the separation available is compressed onto the ground where birdies are effectively expected — and a bogey there is a two-shot swing against a field that almost never makes one ^mmodeled. The three worst-than-par scores on the par-5s in 174 attempts is the tell: missing the birdie is the cost, and making one only keeps pace ^mmodeled. The leaderboard is consistent with that shape — the top eight are all at least three under on the two holes — though it does not prove the par-5s decided the order, since those players also separated elsewhere ^mmodeled.

Room for Disagreement

This is 54 holes of a single event on one course, and par-5s playing well under par is the normal texture of professional golf, not an East Lake peculiarity — the finding is the size of the concentration, not its direction ^mmodeled. The par-5 birdie rate can also compress the story misleadingly: a hole every elite player birdies creates little separation between them even as it dominates the field's aggregate, so 'where the scoring is' is not the same as 'where the tournament is won' ^mmodeled. Sunday's final round, played to a different pin sheet and, often, different wind, can redistribute all of it ^mmodeled. What survives is the arithmetic through three rounds: two holes, 112 under-par strokes; the other sixteen, 119 between them.

The View From

On the broadcast this is a leaderboard race decided by Hovland's putter and a chasing pack. Read across all 18 holes for the whole field, it is a tournament where two par-5s make nearly half the field's red figures — as much as all twelve par-4s combined — while the four par-3s barely move the ledger.

Notable

• Golf Channel · 2026 Tour Championship: Viktor Hovland surges to lead with slew of challengers behind him — Round-3 recap of the FedEx Cup finale, with Hovland at 15 under and the group at 12 under.
• PGA Tour · TOUR Championship 2026 — official leaderboard and overview — The tour's own leaderboard for the event; confirms round scores and the level-start, no-cut format.
• Yahoo Sports · 2026 Tour Championship leaderboard, Round 3 live updates, tee times, highlights — Live coverage of the third round at East Lake.
• Golf Monthly · Tour Championship 2026 Tee Times: Round Three — Field and pairings for the round scored here (context for the 30-player, no-cut format).

The United States and Mexico ran roughly $840 billion in goods across their shared border in 2024, by the US government's own count — enough to make Mexico the United States' largest trading partner. Both governments file that trade to the United Nations' Comtrade database. The two filings do not match.

We pulled every leg of the bilateral flow for 2021 through 2024. On goods moving north — Mexico to the US — the two sides agree almost exactly. Mexico reported exporting $475.2 billion in 2023 and $503.4 billion in 2024; the US reported importing $480.1 billion and $510.0 billion. The importer's figure sits about 1% above the exporter's in every year — roughly what freight and insurance add to a shipment's value.

On goods moving south — the US to Mexico — the two ledgers come apart. For 2024 the US reported exporting $334.0 billion; Mexico reported importing $262.7 billion from the US. That is a $71.3 billion gap, and it runs backward: the importer's figure, which normally carries freight and should be the higher of the two, is a fifth *lower* than what the exporter says it shipped. It is not a one-year blip. The southbound gap was $55.1 billion in 2021, $57.5 billion in 2022, $61.4 billion in 2023 and $71.3 billion in 2024 — widening every year.

The Analysis

The following is analysis, not fact. The asymmetry is directional, and that is the whole finding. The same two statistical offices, reading the same border in the same year, agree to within about 1% in one direction and split by a fifth in the other. Whatever is happening is specific to goods the US declares as bound for Mexico.

The direction rules out the usual first explanation — the FOB/CIF freight convention. Trade statistics value exports at the seller's border (FOB) and imports with freight and insurance added (CIF), so an importer's number is normally the larger of the two — as it is northbound. Southbound it is smaller, so freight cannot account for it: the gap is there despite the accounting convention, not because of it.

What the gap *is* remains an open question these two ledgers cannot settle. National statisticians have documented several candidate causes for exactly this kind of US-Mexico asymmetry: goods that pass through the US and are re-exported onward but recorded as US exports to Mexico; exporters coding a final destination less precisely than importers, who must declare origin to clear customs; and goods entering Mexico under temporary manufacturing-import programs that the two countries book differently. ^sspeculative Which of these carries the $71 billion — or how much is plain measurement error — cannot be isolated from the totals alone. We did not decompose the flow by commodity or firm. ^sspeculative

Room for Disagreement

No goods went missing. A bilateral gap of this kind is a measurement seam, not evidence that $71 billion of exports vanished or that anyone miscounted on purpose. Statisticians expect mirror data to disagree, and the US Government Accountability Office flagged the specific limitations of US-Mexico trade figures — the maquiladora treatment, the weaker reliability of export-destination coding — as far back as 1993. On a value-added view, which nets out the imported content embedded in what crosses the border, the balance can look different again. The honest reading is narrow: the two governments' headline southbound goods numbers have not reconciled for at least four years, and the size of the 'US trade deficit with Mexico' depends on whose ledger you read. It is not a claim about who is right.

The View From

**View from the statistical agencies.** Neither the US Census Bureau nor Mexico's INEGI would call this a contradiction. Each measures what crosses its own frontier under its own rules, and the OECD treats reconciling the two — chiefly by stripping out re-exports — as one of the hardest standing problems in trade measurement, not a scandal. On that reading the $71 billion is the visible seam between two honest accounting systems that were never built to produce the same number. It is a lens on the ledgers, not a verdict they hand down.

Notable

• U.S. Government Accountability Office · U.S. Trade Data: Limitations of U.S. Statistics on Trade With Mexico — The US audit office on the specific data problems in US-Mexico trade statistics — maquiladora treatment and the reliability of export figures. The asymmetry is longstanding, not new.
• OECD Statistics Blog · The Re-Export Puzzle: how the OECD addresses the largest source of distortion in merchandise trade statistics — The OECD on re-exports — goods routed through one country and out again — as the biggest documented distortion in partner-country mirror statistics.
• UN ESCAP (Statistics Division) · Asymmetries in International Merchandise Trade Statistics — Working paper explaining why two countries' figures for the same trade — mirror statistics — routinely diverge, and how the gaps are measured.
• Banco de Mexico · The US-Mexico bilateral trade relation through a value-added lens — Mexico's central bank on this exact bilateral relationship, and how imported content embedded in cross-border flows distorts the gross trade numbers.

To keep agencies' foreign-currency reporting consistent, the U.S. government converts such amounts through one quarterly table: the Treasury Reporting Rates of Exchange, maintained by the Bureau of the Fiscal Service. It is the standard rate federal agencies use to report their foreign-currency transactions in U.S. dollars — not a market feed, and not the same thing as a country's official or market exchange rate. It reflects, in Treasury's own words, "the exchange rates at which the U.S. government can acquire foreign currencies for official expenditures as reported by disbursing officers for each post" — one rate per currency, set at quarter-end. Treasury lists specific exceptions where a different rate applies: amounts fixed by international agreements, conversions of one foreign currency into another, foreign currencies sold for dollars, and transactions affecting dollar appropriations — and it directs that a non-current (amended or prior) rate not be used to value transactions affecting those appropriations.

It moves between quarters only on a tripwire. Treasury's rule: "If current rates deviate from the rates in this report by 10 percent or more, Treasury will issue amendments to this quarterly report." Since April 2021 an amendment appears as a second line for that currency, carrying a later effective date (Treasury Financial Manual, Volume I, 2-3200).

We pulled all 171 lines of the June 30, 2026 report — 167 currencies, four of them carrying an amended second line. Every one of the four cleared the 10-percent trigger, base rate to amendment:

- **Bolivia — boliviano:** 6.85 → 10.35, **+51.1%** (amended effective July 15, 2026) - **South Sudan — pound:** 4,800 → 5,700, +18.8% (effective Aug. 14, 2026) - **Venezuela — bolívar soberano:** 621.465 → 723.933, +16.5% (effective July 15, 2026) - **Burundi — franc:** 3,000 → 3,443, +14.8% (effective July 15, 2026)

Three of the four were already drifting: Venezuela carried an amendment the prior quarter too (472.686 → 547.998), and South Sudan and Burundi have crept down for years. Bolivia is the outlier. Across every quarterly report the dataset returns from 2016 through March 2026 — roughly 40 straight quarters — the boliviano sat on a single line between 6.77 and 6.90. June 30, 2026 is its first amended line in that record, and the largest single move of the four.

The Analysis

The following is analysis, not fact.

The 10-percent tripwire makes an amendment worth reading as a signal in itself. The table does not chase daily foreign-exchange moves; it re-marks a currency only when the reported rate has moved at least 10 percent from the figure on its books. So a second line means just that — the quarter's rate had drifted past the tripwire, nothing more inferred about why. Four such amendments this quarter is a small set against 167 currencies — and Bolivia is the one that changes character, from a rate that had held near 6.85 for a decade to a 51-percent revaluation in a single filing.

The why sits outside our pull, and it is well reported: Bolivia's economy ministry ended the country's roughly 15-year dollar peg in late June 2026 — a devaluation of the boliviano of roughly 30 percent, its official rate rising about 40 percent, from near 6.96 to about 9.73 per dollar — after a dollar shortage drove the parallel market toward 20. The Treasury amendment lands weeks later and is consistent with that break; it does not prove it, and the table records the move rather than causing it. ^sspeculative

One comparison needs its dates attached. Treasury's 10.35 mark for the boliviano took effect July 15; the ~9.73 figure was Bolivia's reference rate when the float opened in late June — about three weeks earlier, a window in which the new official rate kept climbing. The two are not a same-day pair, so Treasury's July mark cannot be read as sitting "above" the June announcement; it is a later snapshot of a rate that was still moving. What the amendment shows is narrower and solid: by mid-July the government's own booked rate for the boliviano had crossed 10 — a currency it had carried near 6.85 for a decade.

Room for Disagreement

Read the amendment as a market quote and you will overread it. It is a single per-post figure, updated in discrete 10-percent-or-more steps, not a live rate — so it lags continuous moves and understates volatility between amendments. A currency that drifts 9 percent every quarter can fall by a quarter over a year and never trip the wire, which means the four amendments undercount, not overcount, real movement across the table. And whether four amendments in one quarter is high or ordinary is not something a single report can settle — we did not build the multi-year amendment-frequency series that claim would need. The defensible statement is narrow: in the June 30, 2026 report, four currencies carry an amendment, and Bolivia's is both the largest and the first the boliviano has drawn in the decade the dataset returns.

The View From

A desk watching Bolivia saw a peg break; a desk watching Venezuela saw one more crawl. Neither saw the instrument they share. The Treasury table converts all of it — the embassy in La Paz and the one in Caracas — through one quarterly page, on one 10-percent rule, and reading the whole page is what surfaces that four currencies tripped it at once and that only one of the four had been standing still.

Notable

• France 24 · Bolivia removes 15-year dollar peg in bid to revive economy — Wire coverage of the June 2026 policy change behind the Treasury amendment.
• FXStreet · Bolivia ends Dollar peg, devalues currency — Market-desk account of the move from ~6.96 to a flexible rate.
• Americas Quarterly · Bolivia's Reform Agenda Is Moving, but Slowly — Policy context on the dollar shortage and reform push.
• The Rio Times · Bolivia Ends Its Dollar Peg, Devaluing the Boliviano 30% — Reports the ~30% official devaluation figure referenced above.

The debate everyone is having about the Federal Reserve — cut rates now, or hold them higher — runs on a handful of public numbers. As of late August 2026 the Fed's target range for the federal funds rate is 3.50% to 3.75%, with the effective rate trading near 3.63%. The federal funds rate is the overnight rate banks charge each other that the Fed steers; it is not a rate the Fed sets directly for consumers. The trending peg is the sitting chair's own words: in his first Jackson Hole address as Fed Chair, on 28 August 2026, Kevin Warsh — who this year succeeded Jerome Powell — said the summer's better inflation readings "do not tell me that underlying trends have meaningfully improved," language the market read as the Fed still having "work to do."

The Fed's formal goal is 2% inflation measured on headline — total — personal consumption expenditures, the PCE price index including food and energy. Over the year to July 2026 headline PCE rose 3.70%, well above that 2% target. Core PCE — the same index stripped of food and energy — is the gauge the Fed leans on to judge the underlying trend, because it forecasts where headline is heading; it rose 3.34% and was unchanged from the month before. The more familiar Consumer Price Index, a separate Bureau of Labor Statistics measure, rose 3.4% at the headline and 2.5% at the core over the same year, on the BLS's published 12-month figures — the core CPI reading about 0.8 of a percentage point below core PCE.

The job market — the other half of the Fed's mandate — turned down in July. Employers shed 23,000 jobs, and the two prior months were revised lower by a combined 103,000: May from +129,000 to +63,000, June from +57,000 to +20,000. The unemployment rate edged down to 4.1%, but the BLS tied the dip largely to people leaving the labor force rather than to hiring — so the rate understates the softening the payroll figures show.

So the raw picture: a target range of 3.50-3.75%, inflation above the 2% goal on every gauge (headline PCE 3.70%, core PCE 3.34% and not falling), and a labor market that lost jobs in July and lost more ground once revisions landed. The gap between the policy rate and core PCE — one ex-post proxy for the "real," inflation-adjusted rate — is about 0.3 of a percentage point.

The Analysis

The following is analysis, not fact. Read the two positions at their strongest and each now argues from a different half of the Fed's mandate.

**The strongest case for holding higher.** The Fed's target is 2% inflation, and inflation is above it on every gauge — headline PCE, the measure the 2% goal is actually written on, at 3.70%, and core PCE, the underlying-trend gauge, stuck at 3.34% and not moving. A year and a half into "the last mile" and it has not been covered. You do not cut into inflation that is still running above target, because the risk is asymmetric: ease now, and if price growth re-accelerates the Fed has to reverse and hike into a slowing economy. Chair Warsh's own read — that better summer prints "do not tell me that underlying trends have meaningfully improved" — is this case stated from the chair. ^mmodeled

**The strongest case for cutting now.** The labor market just turned. Employers shed 23,000 jobs in July and the prior two months were revised down by a combined 103,000; the 4.1% unemployment rate fell only because people left the workforce, not because hiring improved — so the rate flatters a job market that is actually weakening. Rate moves hit the economy with a lag of a year or more, so waiting until the softening shows up in the headline unemployment rate is waiting too long to prevent it. And core CPI, at 2.5%, reads about 0.8 of a point cooler than core PCE — evidence that the disinflation is closer to done than the Fed's preferred gauge admits. ^mmodeled

**The middle — where the numbers actually sit.** This is not the clean "lean hold" a glance at the inflation line suggests; it is a genuine conflict between the Fed's two mandates. The hold case is right that inflation is not beaten — both PCE gauges sit above 2%, and the cut camp's "inflation is normalized" premise is simply not in the data. But the cut case no longer rests on that premise: it rests on a labor market that lost jobs in July and gave back 103,000 more once revisions landed, which is real and current. What each side overstates is certainty about the other half. The hold camp waves off a payroll turn that has, in past cycles, preceded downturns; the cut camp waves off inflation that is still measurably above target. The settled part is narrow: "inflation is beaten, so cut" is unsupported — but "the job market is fine, so hold" is now unsupported too.

Room for Disagreement

The argument can hinge on which number you privilege — and now on which mandate. On inflation, core CPI (2.5%) and core PCE (3.34%) are both "core," but they weight housing and healthcare differently and sit about 0.8 of a point apart. The Fed's 2% goal is written on headline PCE, which is why its official read runs hawkish; a cut advocate anchored on the cooler CPI and a hold advocate anchored on PCE can both be honest. On the labor side, the unemployment rate (4.1%, and lower than in spring) and the payroll count (negative, with heavy downward revisions) point in opposite directions in the same month — the rate says steady, the establishment survey says weakening. Two further caveats cut against certainty. Monetary policy acts with long and variable lags, so the "right" rate depends on where inflation and jobs will be in a year, not where they are now — a forecast, not a fact. ^sspeculative And a single month is noisy: one firm jobs report or one hot inflation print would move this debate. What is not in dispute is the snapshot: inflation above target on every gauge, and a labor market that stopped adding jobs in July.

The View From

**View from the mortgage desk and the credit-card statement.** To a household, "the Fed's real rate is 0.3%" is a monthly payment — but the two big consumer rates do not move with the policy rate the same way. A credit-card APR is set as the prime rate plus a margin, and prime moves almost step-for-step with the fed funds rate, so a Fed cut passes through to revolving balances fairly directly. A 30-year mortgage does not: it tracks long-term Treasury yields, which move on the market's expectations for inflation and growth. So the Fed can cut and mortgage rates can hold — or even rise — if bond investors think easing will let inflation run, and with inflation still above target on every gauge, that is a live risk. The lesson the whole-data read offers a borrower is the one the cable debate skips: the Fed sets one short-term rate, but what you pay to borrow long-term is set by the market's read on inflation — and that read is still looking at numbers that start with a 3.

Notable

• The Washington Post · Fed chair Warsh, concerned about inflation, says bank 'has more work to do' — The trending peg — coverage of the Chair's 28 August Jackson Hole speech, where he said the summer's better inflation prints did not show underlying trends improving; the hold-side case straight from the chair.
• CNBC · Jobs report July 2026: payrolls fall 23,000 as prior months are revised sharply lower — Coverage of the July payroll decline and the 103,000 downward revision to May and June — the cut side's strongest current evidence.
• CNBC · CPI inflation report July 2026: prices rose 0.1%, annual rate 3.4% — Coverage of the July CPI release (3.4% headline, 2.5% core) — the cooler of the two inflation gauges the debate cites.
• NPR · Fed's Kevin Warsh warns inflation is too high, sparking bets rate hikes are coming — The market reaction — investors moving to price a possible September hike after the speech, the hawkish end of the live debate.

The European Union writes two numbers into its own treaties as the outer edge of a sound government budget: debt no higher than 60% of GDP, and a deficit no wider than 3%. They are the reference values of the Stability and Growth Pact, fixed at Maastricht three decades ago and carried into Article 126 of the bloc's governing treaty.

We pulled Eurostat's own excessive-deficit-procedure dataset — the figures the European Commission uses to police those limits — updated 22 April 2026, and kept every 2024 value for all 27 member states. Sixteen breach at least one line. Only 11 clear both.

Twelve carry debt above 60% of GDP: Greece highest at 154.2%, then Italy 134.7%, France 112.6%, Belgium 103.9% and Spain 101.6%; Germany (62.2%) and Cyprus (62.7%) sit just over. Eleven ran deficits wider than 3%: Romania widest at 9.3%, then Poland 6.4%, France 5.8%, Slovakia 5.3% and Hungary 5.1%. The EU27 as a single economy breaches both — 80.7% debt on a 3.1% deficit.

The Analysis

The following is analysis, not fact. Read the two lists side by side and they only partly overlap. Twelve countries are over on debt, 11 over on deficit, but only seven are over on both — five breach the debt line alone, four the deficit alone. The two halves of the same rulebook catch overlapping but noticeably different sets.

Greece is the tell. It carries the bloc's largest debt pile, 154.2% of GDP — two and a half times the ceiling, the face of the last euro crisis — and in 2024 it ran a budget surplus of 1.3%. Cyprus, also above the debt line at 62.7%, posted a 4.1% surplus; Portugal, at 93.5%, a 0.6% surplus. At the other end, the widest deficit breachers, Romania and Poland, each carry debt of 54.8% — comfortably under the 60% ceiling.

A country can be buried in accumulated debt yet balancing its books this year, or lightly indebted yet borrowing heavily. The debt ceiling measures a stock built over decades; the deficit limit measures a single year's flow. In 2024 they pulled apart — which is why a rule-by-rule read names one set of countries, and the bloc-wide aggregate hides all of it.

Room for Disagreement

Exceeding a reference value is not the same as breaking the law. The Stability and Growth Pact, reformed in 2024, judges compliance on a multi-year net-expenditure path and lets debt above 60% pass if it is 'sufficiently diminishing' toward the value.

The bloc's actual enforcement tool, the excessive deficit procedure, covered eight states in 2024 — the seven the Council named that July (Belgium, France, Italy, Hungary, Malta, Poland, Slovakia) plus Romania, under procedure since 2020. That list turns on the deficit, not the debt ceiling, and omits some breachers on these numbers: Spain, Austria and Finland each ran deficits above 3% in 2024 yet were not among the seven states placed under procedure that July. A raw breach and a formal finding of an excessive deficit are different things — and a ratio can also climb simply because GDP fell.

The View From

**View from Brussels.** The Commission would say these numbers were never meant as automatic pass-fail switches. A country above 60% on a Council-agreed correction path is treated differently from one simply ignoring the line; the reformed pact weighs the medium-term trajectory, not a single year's snapshot. On that framing, France at 112% inside an agreed path is a managed case, not a violation. It is the institution's own reading of its rules — a lens on the ledger, not a verdict these figures hand down.

Notable

• France 24 · EU reprimands France and six other EU countries for breaking budget rules — Wire coverage of the July 2024 decision opening excessive-deficit procedures — the enforcement side of the breaches counted here.
• Council of the EU (Consilium) · Stability and growth pact: Council launches excessive deficit procedures against seven member states — The official record naming Belgium, France, Italy, Hungary, Malta, Poland and Slovakia — the seven cited in Room for Disagreement.
• Council of the EU (Consilium) · Stability and growth pact: Council adopts recommendations to countries under excessive deficit procedure — The follow-on corrective paths and deadlines set for the states under procedure, Romania included.
• EU Law Live · Council Decisions on excessive deficit in France, Hungary, Italy, Belgium, Malta, Slovakia and Poland, published in OJ — Legal-press note on the Official Journal publication of the individual decisions.

On 28 August, President Trump posted that the meatpacking business is "essentially 4 of them, a very non competitive number," a "nasty Monopoly" with "much of its ownership based outside of the U.S." that makes "life miserable for our wonderful Farmers and Ranchers." His remedy, in his words: "I am authorizing legal documents to be drawn in order to allow Farmers and Ranchers to be given the right to PROCESS THEIR OWN FOOD. This should move quickly."

The rebuttal, raised by ranchers replying to the post, was blunt and specific: the right to process your own beef already exists. The catch is that such meat is stamped "Not For Sale" — and without a change to the federal rule requiring inspection to sell it, a restated "right to process" changes nothing.

Both statements are checkable against the record. The concentration is real: the four largest packers account for about 85% of U.S. steer and heifer slaughter, per the USDA's 2019 Packers and Stockyards report — the "4" Trump names. In May 2026 the Justice Department confirmed an antitrust investigation of that concentration, a price-fixing investigation under Section 1 of the Sherman Act tied to a November 2025 executive order; Acting Attorney General Todd Blanche, who said investigators had reviewed more than three million documents, described it as carrying a criminal component running "in parallel with the civil investigation" and declined to call it predominantly one or the other. Two of the four — JBS, headquartered in Brazil, and National Beef, majority-owned by Brazil's Marfrig — carry major foreign ownership, so the "ownership outside the U.S." claim is partly borne out; Tyson and Cargill are U.S.-based.

The rancher's point is also on the books. Federal law already lets a livestock owner have an animal slaughtered and processed for personal use at a "custom-exempt" facility — but that meat must be stamped "Not For Sale" and stay with the owner, the owner's household, employees, and non-paying guests. To sell beef to the public, the animal must be slaughtered and processed under continuous USDA or equivalent state inspection. That inspection-for-sale requirement is written into federal statute, the Federal Meat Inspection Act.

There is legislation built precisely to change that. The standalone PRIME Act (Processing Revival and Intrastate Meat Exemption Act, H.R.4700), reintroduced in 2025 by Rep. Thomas Massie with a bipartisan Senate companion, would let states authorize custom-processed meat to be sold within their own borders; on its own it remains only introduced. A pilot version was folded into the farm bill the House passed in spring 2026, but it has not cleared the Senate and has not become law.

The Analysis

The following is analysis, not fact. Read the post and the reply as a single argument and they are not really contradicting each other — they are each holding one true half.

**The strongest case for the president's framing.** The market power is not imagined. A four-firm, ~85% grip on fed-cattle buying is the kind of concentration antitrust law exists to scrutinize, which is why the DOJ opened a probe; ranchers have complained for years that the packers set the price of the animal and the price of the beef at both ends. Foreign ownership of two of the four is a real and politically salient fact. Aiming policy at breaking that grip — and at giving producers an alternative to selling into it — is a coherent goal, and expanding on-farm and small-plant processing capacity is one lever that could, at the margin, loosen it. ^mmodeled

**The strongest case for the ranchers' rebuttal.** "The right to process their own food" is not the binding constraint, because ranchers already have it — custom-exempt slaughter is legal in every state. What they cannot do is *sell* that meat, and the wall there is the federal inspection requirement, not a missing "right." An announcement that grants a right producers already hold does nothing about the barrier that actually keeps their beef off the market. ^mmodeled

**The middle — where the settled law puts it.** The concentration Trump names is real; the fix he announced points at the wrong lock. The barrier the rancher names — USDA inspection to sell — sits in a federal statute, and changing a statute takes an act of Congress, not "legal documents" drawn by the executive. That is exactly what the PRIME Act would do, and it is stalled in Congress, not on the president's desk. An executive order can direct agencies, expand inspection capacity, or push antitrust enforcement; it cannot, by itself, repeal the inspection-for-sale rule the rancher is pointing at. So the honest read is: right problem, and a remedy that — as worded — restates a right ranchers already have rather than lifting the sale barrier they actually hit. ^mmodeled

Room for Disagreement

What the record does not settle is intent and effect, and both cut against a confident verdict. The "legal documents" have not been drawn — Trump's post is an announcement, not a text — so it is not yet knowable whether they will merely restate the processing right, or reach further: directing USDA to expand small-plant inspection, backing the PRIME Act, or opening an antitrust front on the packers. If they do the latter, the "changes nothing" verdict would be too harsh. ^sspeculative It is also true that more processing capacity, even without a change to the sale rule, has some value — it gives producers more of their own freezer beef and marginally less dependence on the Big Four. And a president cannot pass the PRIME Act, but a president leaning on Congress can move it: Rep. Massie has explicitly urged the bill's passage in the wake of the announcement, so the post may function as pressure on the exact fix the critics say is missing. The claim here is narrow and legal, not political: the specific thing announced — a "right to process their own food" — is a right that already exists, and the sale barrier lives in a statute the executive cannot rewrite alone.

The View From

**View from the inspection line.** To a small state-inspected processor, the fight is over a word the debate keeps blurring: "process." Slaughtering an animal for a rancher's own freezer and slaughtering one whose cuts will be sold at a farmers' market are, in federal law, two different acts under two different regimes — custom-exempt for the first, continuous inspection for the second. The president's phrase, "process their own food," lands squarely in the first regime, which was never closed. The second — the one that would put a rancher's beef in a neighbor's grocery cart — turns on who is standing on the kill floor, and that is set by statute, not by will. From that seat, the question that decides whether any of this reaches ranchers is not whether they may process, but whether the inspection-for-sale line moves — and that line moves in Congress.

Notable

• The Hill · Massie urges passage of PRIME Act following Trump move to ease meat processing restrictions — Ties the president's announcement to the actual legislative vehicle — the PRIME Act — and the lawmaker pushing it, the connection the machine read draws between the post and the statutory fix.
• Congress.gov (Library of Congress) · H.R.4700 — 119th Congress: PRIME Act (Processing Revival and Intrastate Meat Exemption Act) — The bill text and status — the exemption that would let custom-processed meat be sold intrastate, and the record showing it has not become law.
• National Agricultural Law Center · 'Custom Exempt' Slaughter: The Exception, or the Rule? — The neutral legal explainer for the core fact both sides argue over: custom-exempt processing is legal but 'Not For Sale'; selling requires inspection.
• USDA FSIS · Custom and Retail Exemption from Federal Inspection (NACMPI presentation) — The inspecting agency's own description of the custom exemption and its 'not for sale' limit — the primary source behind the 'Not For Sale' stamp.

Every rushing touchdown starts somewhere on the field, and in 2025 nearly three-quarters of them started close. We read the complete play-by-play of the 2025 NFL regular season — 14,633 rushing plays — and of the 510 rushing touchdowns, 375 (73.5%) were scored from inside the 10-yard line and 301 (59.0%) from inside the 5. The touchdown, the single most valuable event on a fantasy back's line — worth six points in standard scoring — lives on a sliver of the field.

That sliver is a sliver of the workload, too. Across all 49 running backs who carried 100 or more times in the regular season — the workhorse and committee-lead tier — carries from inside the 10 numbered 880 of their 9,493 total carries, or 9.3%. Yet that 9.3% is what the scoring followed. Within the 49-back pool, rushing touchdowns correlated 0.885 with a back's inside-the-10 carries and only 0.636 with his total carries — the touchdowns tracked the goal-line role far more tightly than the overall volume a draft board is built on.

And the two roles are not the same depth chart. For 22 of the 49 backs, their rank in inside-the-10 carries sat 10 or more places away from their rank in total carries; for 17 of them the gap was 15 or more. Jonathan Taylor is the case where both align: he led the pool in carries (324), tied for the lead in inside-the-10 carries (40), and led in rushing touchdowns (18). Derrick Henry did the same on a smaller base — 307 carries, 40 inside the 10, 16 touchdowns. Those are the alignments the correlation is anchored on, not the rule.

The Analysis

The following is analysis, not fact. Fantasy scoring is a repeatable signal — snaps, carries, targets — with a noisy layer of touchdowns and long gains stacked on top. The whole-pool read locates where, in 2025, that noisy layer sat: not on the bulk of a back's carries, but on the handful that begin inside the 10, and those are handed out on a separate ledger. Whether that goal-line work recurs as its own usage input would take multiple seasons to establish; this is a single-season cross-section, not a 2026 forecast ^mmodeled. Two 2025 backfields show the split cleanly.

In Seattle, Kenneth Walker carried 221 times to Zach Charbonnet's 185 — Walker owned the workload. But Charbonnet took 32 of the Seahawks' 49 inside-the-10 rushes, 65% of them, to Walker's 10, and out-scored him 12 rushing touchdowns to 5. The higher-volume back was not the one the rushing scores went to.

In Atlanta the gap ran the other way against workload. Bijan Robinson carried 288 times, Tyler Allgeier 143 — Robinson nearly doubled him. At the goal line they were even: Allgeier led the team with 19 inside-the-10 carries to Robinson's 18, and 10 inside-the-5 carries to Robinson's 11, and finished with 8 rushing touchdowns to Robinson's 7. The lighter-used back scored more on the ground. Breece Hall is the same story without a named partner: 243 carries, a top-dozen workload, but only 10 inside-the-10 carries and 4 rushing touchdowns. The yards were there; the goal-line work, and the scores, were not.

Room for Disagreement

Goal-line role is not independent of quality or volume — the correlation with total carries is still a positive 0.636, and the backs who own both, like Taylor and Henry, are rewarded in standard and PPR scoring alike. Read the finding too hard and it inverts into a prediction; this piece makes none. A single season's goal-line split is also a small, volatile sample: a back's whole touchdown case can rest on a dozen or two inside-the-10 carries, and which back gets them is a coaching choice that can shift week to week and year to year ^mmodeled. Format matters, too — PPR leagues pay for receptions and dampen touchdown dependence, so the goal-line effect is largest in standard and touchdown-heavy scoring, and this read is rushing-only, setting aside the goal-line targets that also become scores. What is not in dispute is the arithmetic: across the 49 backs, rushing touchdowns tracked inside-the-10 carries (0.885) far more closely than total carries (0.636), and 73.5% of the league's 510 rushing scores came from inside the 10.

The View From

Stare at one roster and a running back is a carry count, a yardage total, and a touchdown number that reads like luck. Read all 49 workhorse backs at once, against every handoff of the season, and in 2025 that touchdown number lined up less with total workload than with who took the carries inside the 10 — an address some backs reached far more often than others, however much they touched the ball.

Notable

• ESPN · How Seahawks RB Kenneth Walker is 'earning' more carries than Zach Charbonnet — External reporting on the Seattle backfield split — documents Walker leading in overall touches, the workload half of the divergence our goal-line data completes.
• Atlanta Falcons (team site) · Why Tyler Allgeier is Falcons' quiet superstar on offense — External coverage of Allgeier's goal-line/red-zone touchdown role relative to Bijan Robinson — the Atlanta case in our piece.
• FantasyPros · 2025 NFL Red Zone Stats | Running Backs — Public data home (context, not reporting) where readers can sort RB red-zone rushing volume themselves; uses the 20-yard red-zone line rather than our inside-10 cut.

The Monthly Treasury Statement, the government's official monthly books, does something the headline deficit figure hides: it splits the budget in two. Most of the government is on-budget. Two entities are off-budget by law — the Social Security trust funds and the Postal Service. We pulled Treasury's on-budget and off-budget lines for the first 10 months of each fiscal year back to 2022.

Through July 31, 2026 — 10 months into fiscal 2026 — the off-budget accounts took in $1.126 trillion and spent $1.242 trillion, a deficit of $116.1 billion. Off-budget receipts are essentially Social Security's payroll taxes; the outlays are its benefit checks plus the Postal Service's net operations. The accounts spent about $1.10 for every $1.00 they received.

That is a reversal. In the same 10-month window, the off-budget balance ran a surplus of roughly $6 billion in fiscal 2022 and $15 billion in fiscal 2023, then crossed into deficit — minus $17.7 billion in 2024, minus $80.8 billion in 2025, and minus $116.1 billion in 2026. The shortfall widened about 44% in the past year alone.

The off-budget deficit is 6.5% of the government's total $1.799 trillion deficit through July; the on-budget side accounts for the other 93.5%, or $1.683 trillion. The two halves reconcile exactly to the headline year-to-date deficit Treasury reports.

The Analysis

The following is analysis, not fact.

The unified deficit is, by definition, the on-budget balance plus the off-budget balance. So the sign of the off-budget line does real work: when it ran a surplus, it was subtracted from the reported deficit, making the total look smaller than the on-budget government alone; now that it is negative, it adds. Across this five-year window that offset flipped from shrinking the headline number to enlarging it.

One precision the raw line hides: this is a cash gap, not a measure of Social Security's reserves. Treasury's off-budget receipts line counts payroll and related social-insurance taxes only — it does not include the interest the trust funds earn on their holdings of special-issue Treasury securities. That intragovernmental interest is credited to the funds outside this cash line, so the change in the funds' reserves is smaller than the $116 billion operating gap ^mmodeled. The $116 billion is the operating gap between what came in from workers and what went out in benefits, not the change in the funds' balance — which these MTS cash lines do not themselves measure.

The on-budget side still dwarfs all of this. The finding is not that Social Security and the Post Office drive the deficit — they do not, at 6.5% of it. It is that the part of the budget that used to offset the deficit no longer does, and in this window has begun to add to it ^mmodeled.

Room for Disagreement

Three honest caveats. First, off-budget is not Social Security alone: the line bundles the Postal Service, which runs its own persistent losses, so it overstates a Social-Security-only figure on the outlay side. Second, the cash framing cuts against alarm — counting the interest the trust funds also receive, the programs' total income ran closer to balance than the minus $116 billion cash line suggests, and the accumulated reserves have not been exhausted. Third, this is a 10-month snapshot; the calendar timing of benefit payments and payroll-tax receipts moves the monthly figures, and the full-year number will differ from the year-to-date one.

The View From

Read one way, this looks like a milestone: the combined off-budget operating cash balance — Social Security's payroll taxes and benefits plus the Postal Service — has flipped from surplus to deficit, which budget analysts have long forecast as the demographic turn arrives ^mmodeled. What has disappeared in this window is that operating cash surplus, not the trust funds' total income or accumulated reserves, which counting interest still cushions. Read the other way, it is partly an artifact of where a line is drawn — the same dollars would land in the deficit whether Social Security were counted on-budget or off, and once its interest income is counted the program sits nearer to balance than the cash gap implies. Both readings rest on the same set of Treasury lines.

Notable

• Committee for a Responsible Federal Budget · Treasury Confirms $1.8 Trillion Deficit for First 10 Months of FY 2026 — Independent analysis of the same July Monthly Treasury Statement release this piece pulls, confirming the $1.8 trillion 10-month deficit.
• Reuters (via Yahoo Finance) · U.S. budget deficit hit July record of $432 billion in fiscal 2026 — Wire coverage of the same MTS release, focused on the headline monthly and year-to-date deficit.
• Brookings Institution · Yes, Social Security can run budget deficits — Explainer on how Social Security's cash shortfalls are financed by borrowing, and the trust-fund-versus-cash distinction central to this piece's caveat.
• Committee for a Responsible Federal Budget · Analysis of the 2026 Social Security Trustees' Report — Context on the trust funds' longer-run finances behind the widening off-budget cash gap.

The United Nations gauges whether a country's external borrowing is sustainable with a single ratio — Sustainable Development Goal indicator 17.4.1, debt service as a share of exports. It sets no numeric target; it is a pressure gauge, not a pledge. The World Bank operationalizes it as DT.TDS.DECT.EX.ZS: total debt service — principal and interest actually paid on external debt — as a percentage of exports of goods, services and primary income.

We pulled that series, updated 13 July 2026, and kept every economy with a 2023 value: 114 of them. Almost all are low- or middle-income; only three high-income economies report at all, because the underlying figures come from the World Bank's Debtor Reporting System, which the rich world largely sits outside. This is, in effect, a developing-world ledger.

The median economy paid 12.4% of its export earnings to service external debt in 2023. A decade earlier, in 2013, the median was 8.1%. Among the 111 economies reporting in both years, the median climbed from 8.3% to 12.6%, and 81 rose against 30 that fell. The World Bank's low- and middle-income aggregate moved from 9.2% to 15.4% over the same span; its poorest, IDA-only group from 8.1% to 11.9%.

Twenty-eight of the 114 now spend 20% or more of export earnings servicing external debt; 15 spend 30% or more; 7 spend 40% or more.

The Analysis

The following is analysis, not fact. The frame a single-country desk reaches for is that debt distress tracks poverty — that the deepest holes are the poorest countries. Read all 114 filings at once and that is not what the ledger shows. Sort by income and the low-income median is 10.5%, below the lower-middle (13.3%) and upper-middle (13.4%) tiers. The extreme top of the list is entirely upper-middle-income: El Salvador at 83.1%, Brazil at 53.5%, Argentina at 51.4%, Kazakhstan at 50.3%. The only low-income country in the top six is Mozambique, at 46.2%.

The reading this is consistent with — not one these data isolate — is that the poorest borrow concessionally: IDA credits and bilateral loans at low rates and long maturities, some under active debt relief, which keep the export ratio down. Middle-income economies borrow more at market terms — Eurobonds and commercial credit that repay harder and faster. We did not test it: no creditor-composition or interest-rate variable was joined, so treat it as a hypothesis the pattern fits, not a mechanism. ^sspeculative

One caveat the whole-record read makes visible: single-year ratios are lumpy, because principal repayments spike when big bonds mature. El Salvador swung from 32.2% in 2022 to 83.1% in 2023 as debt came due. The robust signal here is the aggregate decade climb and the income-tier ordering — not any one country's rank in any one year.

Room for Disagreement

The number depends on which slice of debt you count. Our 12.4% is total debt service — public and private, principal and interest. UNCTAD's widely cited figure, that half of developing countries pay at least 6.5% of export revenues, measures external public debt service only, a narrower slice, which is why its median runs lower; the World Bank's roughly-6% headline is IDA interest payments alone. Three different concepts, three different numbers — each correct about its own cut, none directly comparable.

And a ratio can rise two ways: debt service climbing, or exports falling — a commodity bust flatters no borrower. A high reading is also not automatically a crisis. A country rolling cheap debt with deep reserves is not the country that has lost market access, even at the same 30%. The gauge flags pressure; it does not diagnose solvency, and it should not be read as if it did.

The View From

**View from a middle-income finance ministry.** A ministry could read part of the burden as a consequence of graduation and market access. Graduate out of the poorest tier and the concessional windows narrow; you fund yourself in the bond market instead, at market rates rather than the low, long concessional terms the poorest still get — so the same access that signals you have arrived is the access that puts you near the top of a list of who spends the most of their exports paying creditors back. This is a plausible reading of the ordering, not a mechanism these data isolate; the analysis did not test it. ^sspeculative

Notable

• World Bank · Developing Countries Paid Record $1.4 Trillion on Foreign Debt in 2023 — The International Debt Report press release documenting the same rise — interest costs up nearly a third to $406 billion, a 20-year high.
• The Standard (Reuters) · World Bank says global debt interest payments hit record US$4.15 trln in 2024 — Wire coverage of the 2025 International Debt Report and the widest external-financing gap in over 50 years.
• UNCTAD · A World of Debt 2025: It is time for reform — The report behind the alternative metric in Room for Disagreement — half of developing countries paying 6.5%+ of export revenue on external public debt service.
• Drishti IAS · International Debt Report 2024 — Explainer on the same World Bank report, including the IDA-country range of interest-to-export earnings (up to 38%).

The U.S. State Department rates every destination Level 1 (Exercise Normal Precautions) through Level 4 (Do Not Travel). New Zealand's Ministry of Foreign Affairs runs a matching four-rung scale — 'Exercise normal safety and security precautions' (1), 'Exercise increased caution' (2), 'Avoid non-essential travel' (3), 'Do not travel' (4) — and for its safest destinations issues no advisory at all. The wording of the top and bottom rungs is nearly identical between the two systems; only the third rung differs in phrasing while occupying the same position.

In the same-day pull — 2026-08-29 — the U.S. feed carried 215 rated destinations (84 at Level 1, 81 at Level 2, 28 at Level 3, 22 at Level 4); New Zealand's list carried 222, of which it issued no specific advisory for 71. 204 countries appear in both registers with a single overall national level; that shared set is the population for every comparison below. New Zealand's 'no advisory' posture is counted as its most relaxed rung (Level 1) for ranking ^mmodeled.

One territory no longer fits that frame and is set aside from every count below: as of the 29 August pull the U.S. rates Israel (Level 3), the West Bank (Level 3) and Gaza (Level 4, reclassified 26 August) as three separate advisories, while New Zealand issues one combined advisory for Israel and the occupied Palestinian territories at Level 4. Because the U.S. no longer assigns this ground a single overall national level, the pairing falls outside the comparison's population rule — the two systems disaggregate the same territory differently, the exact structural gap a single national rating cannot show.

On those 204 countries the two governments assign the exact same rung to 153 — 75.0%. Of the 51 that differ, 48 differ by one rung and only 3 by two: the U.S. rates Uganda Level 4 while New Zealand rates it 2, and rates both Rwanda and Macau Level 3 where New Zealand sits at its floor.

Where they split, the U.S. is the stricter register 38 times to New Zealand's 13. In 21 of those 38 the U.S. assigns a caution level to a place New Zealand does not flag at all — among them The Bahamas, the Dominican Republic, Botswana, Namibia, Mauritius and Uruguay. The U.S.-stricter countries cluster hard by region: 17 are in Africa and 12 in the Caribbean or Latin America.

New Zealand's 13 stricter calls run the other way. It pushes a Central Asian block — Kazakhstan, Kyrgyzstan, Uzbekistan and Georgia — plus Thailand, Senegal, Zambia, El Salvador and Cyprus from the U.S. floor up to 'increased caution'; rates Cuba and Eritrea a rung above the U.S.; and is harsher at the very top on two, calling Kuwait and Venezuela 'Do not travel' where the U.S. says only 'Reconsider Travel'.

At that top rung the registers nearly converge: of the 204 shared countries each government places 21 in its worst tier, and 19 are the same — the war and collapse states from Afghanistan and Syria to Ukraine, Russia, Haiti and Somalia. The U.S. alone adds Chad and Uganda; New Zealand alone adds Kuwait and Venezuela.

The Analysis

The following is analysis, not fact. Read either list on its own and each looks like a plain safety map. Read them side by side and a shape appears that neither shows alone: the two governments are not merely stricter or looser than each other — they are cautious about different parts of the world. New Zealand's count of stricter calls is smaller — 13 to 38 — so on the raw tally the U.S. is the more cautious register. The two are not cautious about the same places, though: 29 of the 38 U.S.-stricter calls land in Africa (17) and the Caribbean or Latin America (12), while New Zealand's 13 spread thin across Central Asia and a handful of mid-risk states.

The sharpest expression of that is the 21 countries the U.S. flags that New Zealand leaves unflagged — a set that is mostly African and Caribbean, and mostly at Level 2 — 'increased caution,' the mildest of the scale's active warnings. Two capitals looking at the same Botswana, the same Bahamas, publish opposite calls on whether the place warrants a printed warning at all. Which register is 'right' is not answerable from the registers; each reflects its own government's intelligence, its own citizens abroad, and its own read of the same ground ^sspeculative.

The two registers converge at the top. When the stakes are highest — the 'Do Not Travel' rung — the two governments agree 19 times out of 21. And where they do diverge, it is almost always narrow: 48 of the 51 disagreements are a single rung, not a gulf.

Room for Disagreement

This compares two four-rung scales that are not perfectly interchangeable. The U.S. third rung ('Reconsider Travel') and New Zealand's ('Avoid non-essential travel') occupy the same position but are worded differently, and treating New Zealand's 'no advisory' state as equal to a Level 1 is a modeling choice, not a label either government prints ^mmodeled. Both systems also carry sub-national warnings — a calm national rating can hide a 'Do Not Travel' province — that a single overall level does not capture, so two countries that agree at the national rung may still diverge region by region. And a matched population of 204 required normalizing country names across two registers; a different matching rule would shift the count by a few. None of that moves the central result: three-quarters agreement, a U.S.-stricter tilt of 38 to 13, and near-identical worst-tier lists. Before you travel, read your own government's current advisory and entry rules at the source, and cross-check a second country's assessment — because, as this comparison shows, the two do not always agree.

The View From

A traveler checks one government's page for one country and sees a single number. The divergence here is visible only from reading both complete registers at once — 204 paired judgments — which is how the regional pattern in the U.S.-stricter set, and the four-country gap at the top, surface at all. The blind spot is worth stating plainly: we can read what each government decided, not why. The classified threat assessments, the diplomatic relationships, the places each country has citizens and embassies to protect — the inputs that would explain a Uganda rated 4 by one capital and 2 by the other — are not in either feed. We read two registers this run, the U.S. and New Zealand's; a third government's map would redraw the lines again.

Notable

• UNSW Newsroom · Why does the US still have a Level 1 travel advisory warning despite the chaos? — Analysis on how governments assign different advisory levels to the same country and why the systems diverge — the exact phenomenon this piece quantifies for the US and New Zealand.
• CBS News · 21 countries have 'Do Not Travel' warning. Here's what to know about the U.S. State Department advisories. — External reporting on the U.S. Level 1–4 system and the Do-Not-Travel list this comparison lines up against New Zealand's.
• Yahoo News · U.S. Has Issued 21 'Do Not Travel' Warnings — External reporting listing the U.S. Level 4 countries, the worst-tier set that overlaps New Zealand's on 19 of 21.
• SafeTravel (New Zealand MFAT) · About our advisories — Context/background (not reporting on this finding): New Zealand's own description of its four advisory levels, used here to crosswalk against the U.S. scale.

Major League Baseball has played 550 games decided by a single run this season — 27.1% of every game on the board, more than a quarter of the schedule. We took the league's own standings through the games of Aug. 28, 2026, and split all 30 clubs' records into two piles: one-run games, and everything else. The two halves tell opposite stories.

In the pile decided by two runs or more, the standings behave. A team's run differential — the simplest measure of how good it is — lines up with its multi-run record at r = 0.92 across the 30 teams. In the one-run pile, that link nearly disappears: run differential and one-run record correlate at just r = 0.23.

The names make the gap concrete. The second- and third-best one-run records in baseball belong to the Texas Rangers (22-13) and the Cincinnati Reds (23-14) — two teams with losing overall records. Texas has been outscored by 52 runs on the season; Cincinnati by 98, one of the worst margins in the majors. Both win close games above a .620 clip anyway. Only the Philadelphia Phillies, at 23-11, are better in one-run games.

Flip the column and the Detroit Tigers appear: a team that has outscored its opponents on the season, sitting on the second-worst one-run record in the league at 12-23. The San Francisco Giants are worst at 11-24.

Read across all 30 teams, four clubs with losing overall records hold winning one-run records — Texas, Cincinnati, Seattle and Pittsburgh. Only two winning teams carry losing one-run records: the New York Yankees and the Boston Red Sox.

The Analysis

The following is analysis, not fact. Run differential is the raw stat predictive models are built on, and in most games it lines up with the record: in the multi-run pile it matches the standings at r = 0.92. Read that as confirmation, not discovery — lopsided games are exactly where run differential piles up, so the two share information by construction. The finding is the other pile. When a game comes down to a single run, that same measure of quality barely tracks the outcome at all — r = 0.23.

That is consistent with the analyses cited below, which find a team's one-run record among the least stable things it does, closer to a coin flip than a repeatable skill ^mmodeled. A deep bullpen and a good bench plausibly help at the margin, but the margin is small and the sample is loud ^sspeculative. What this season's split adds is a set of names for the noise — Texas and Cincinnati, two sub-.500 teams in the top three of the one-run column, the split run differential tracks least, and Detroit, on the wrong side of it despite a positive run margin.

The practical read points one way: a chunk of the current standings is written in a hand that tends not to repeat. A differential-based model would expect the clubs over-indexed on one-run wins to fade, and the ones buried by close losses to climb ^mmodeled. It would not bet the mortgage on either — r = 0.23 is a whisper, not a law.

Room for Disagreement

One-run records are not pure luck, and r = 0.23 is not zero. A team's one-run record still tracks its overall winning percentage at r = 0.55 — better teams do win a bit more of the close ones. Bullpen quality, a shutdown closer, bench depth and in-game tactics plausibly move the needle, and published models have found real if modest skill signals in one-run performance ^mmodeled. This is also a single season of roughly 30 to 40 one-run games per team — a small, noisy sample by design; the correlations would firm up or soften with more games, and one strong month can flip a team's column. The claim here is narrow: this year, run differential — the standard quality proxy — lines up with a team's one-run record far less than with its multi-run record, r = 0.23 against r = 0.92; the standings inherit that noise.

The View From

From the standings page, Cincinnati and Detroit look alike — both a few games under .500, both unremarkable. Split each season at the one-run line and they are near-opposites: the Reds are one of the best close-game teams in baseball with one of its worst run margins, the Tigers one of the worst close-game teams with a positive one. Read together across all 30 clubs, the two piles say the same thing — the games decided by a single run are the ones run differential tracks least.

Notable

• FanGraphs · How Much Luck Is Involved in One-Run Games? — Background (not reporting on this piece): a modeled look at how much of a one-run record is skill vs. chance.
• Baseball Prospectus · Baseball Therapy: One-Run Winners: Good or Lucky? — Background: finds one-run records are unstable and that randomness overwhelms skill in tight games.
• Baseball-Reference · 2026 Cincinnati Reds Statistics — Independent record cross-check for one of the two lead teams.

The federal rulebook is not spread evenly across the government's work. To see how lopsided it is, you have to read all of it at once.

We pulled the eCFR structure record for every one of the 50 titles of the Code of Federal Regulations, as it stood on Aug. 27, 2026. Title 35 is reserved and empty, leaving 49 active titles. For each, the eCFR reports a size — the byte length of that title's regulatory text. Added up, the code totals 830.1 MB.

One title dominates it. Title 40, Protection of Environment, is 160.1 MB — 19.29% of the entire code, and 1.82 times the second-largest, Title 26, Internal Revenue, at 87.8 MB. After those two the drop is steep: Title 7 (Agriculture) 42.2 MB, Title 12 (Banks and Banking) 39.7 MB, Title 49 (Transportation) 33.0 MB. The five largest titles are 43.7% of the code; the ten largest, 59.0%.

The tail is thin by the same measure. Title 40 by itself is larger than the 27 smallest titles put together, which sum to 155.6 MB. The smallest active title, Title 3 (The President), is 33,536 bytes — about one four-thousand-eight-hundredth the size of Title 40.

The shape barely moved in a year. Measured the same way one year earlier, on Aug. 27, 2025, the code totaled 834.4 MB. Over the twelve months it shrank 0.52% — 23 of the 49 titles got smaller, 22 got larger, and 4 were unchanged. The largest single proportional cut was Title 41 (Public Contracts and Property Management), down 28.6%, or 1.8 MB. Title 40 itself grew 0.8% even as the total edged down.

The Analysis

The following is analysis, not fact.

Read as a whole, the code is concentrated and stable. One title is a fifth of it; five titles are nearly half; and the concentration did not arrive this year — the net change across the whole rulebook is under 1% in either direction, cuts and additions very nearly cancelling. Across the two dated eCFR snapshots, the total contracted 0.52% and the leading titles kept their order, so total volume and distribution changed little by this byte-size measure.

That is the payoff of reading the complete set rather than any single rule. Follow one rulemaking and you see one rule get longer or shorter. Pull all 49 titles and you see the distribution: environmental regulation is not one large area among several — it is in a class of its own, larger than the bottom 27 titles combined. No single filing reveals that; only the whole record does.

What the count establishes is where the text is and that it barely moved — nothing more. Size here is the eCFR's byte measure of regulatory text: a consistent proxy for volume, not a measure of cost, complexity, or how much any rule binds. We report the distribution and the year-over-year delta as facts, and assert no cause for either — not why Title 40 is so large, and not what its 160 MB, or anyone else's, is worth.

Room for Disagreement

The honest limit of this measure is that size is not weight. Byte length counts text, not consequence. A three-line rule can carry billions in compliance cost; a giant title can be mostly definitions, numeric tables, and technical appendices — Title 40's emission and water-quality standards, Title 26's tax tables — so ranking titles by size ranks how much is written, not how much it costs anyone. The MB figures also fold in the eCFR's own structural formatting, which makes them a clean cross-title proxy (every title is encoded the same way) but not a literal word or page count. And the year-over-year comparison captures net codified change, not activity: a title can churn through heavy amendment all year and still finish near the size it started. A reader who wants regulatory burden, not regulatory volume, should treat this as a map of the paper, not the pressure.

The View From

Read one way, the numbers say the federal code is barely an even rulebook — it is dominated by one environmental title, with a long tail of titles that together run smaller than its single largest. Read the other way, they say that in a year of loud argument over regulation, the actual volume of it held almost perfectly still — down half a percent, with as many titles growing as shrinking. Both readings come from the same 49 titles, measured the same way twelve months apart.

Notable

• Competitive Enterprise Institute · Ten Thousand Commandments — Numbers of Rules and Page Counts in the Federal Register — CEI's annual tally of federal rules and Federal Register pages — the standard external measure of regulatory volume, by pages rather than the eCFR's byte size.
• QuantGov (Mercatus Center) · Federal Regulatory Growth — RegData's count of regulatory restrictions ('shall', 'must', 'prohibited') by title over time — a text-analysis approach to the same concentration question this piece measures by size.
• Office of the Federal Register (National Archives) · CFR Index and Finding Aids — Titles and Subject Coverage — The OFR's official description of the 50-title structure of the CFR — background on what each title covers and why Title 35 is reserved.

The box score is built to reward the touchdown as the point of the exercise and the field goal as the consolation. Read every point scored in the 2026 NFL preseason's opening weeks and the two have nearly traded places.

We pulled the complete scoring record of the 33 games already final — the Hall of Fame opener plus the first two full preseason weekends, every game ESPN files under the preseason for schedule weeks one through three — and counted each scoring play. The teams reached the end zone 123 times and split the uprights 112 times; one safety rounds out the ledger. Field goals were 47% of all scoring plays and 28% of all points: of the 1,200 points scored across the 33 games, 336 came off a kicker's foot.

Those were not chip shots. Of the 112 made field goals, 33 — nearly a third — traveled 50 yards or more, with a median make of 43 yards. The longest belonged to Kansas City's Harrison Butker, a 69-yarder against Tampa Bay as the first half expired; it is longer than any field goal made in a regular-season game, though preseason kicks do not enter the record book, as CBS Sports and Yahoo Sports both noted. ^mmodeled

The same split appears in the outcomes, not just the totals. Of the 33 games, 32 produced a winner (Indianapolis and New England tied 13-13). In 6 of those 32, the winning team scored no more touchdowns than the team it beat — and in every one, it made more field goals. In five the touchdown counts were level and the winner made more field goals; in one the winner made fewer touchdowns, Tennessee's 19-16 result over Seattle, on a single touchdown to Seattle's two. Tennessee appears twice: its 19-13 result over San Francisco paired one touchdown with four Joey Slye field goals. Across the two games Slye made eight kicks, four of them from 50-plus, including makes of 58, 55, 55 and 53 yards.

The Analysis

The following is analysis, not fact. The preseason did not turn into a coin flip. The team that gained more total yards still won 26 of the 32 decided games, and the team that held the ball longer won 23 — the sides that played better mostly won. What shifted is the composition of the scoring. Here touchdowns and field goals were nearly even as scoring plays — 123 to 112 — and in six games the winner made more field goals than the loser while scoring no more touchdowns.

Why the field goal carried this much freight is a question the record raises without settling. Preseason offenses run stripped-down playbooks with rotating, mostly backup personnel, and drives stall short of the goal line: teams turned red-zone trips into touchdowns just 54% of the time across the 33 games. That is a pattern consistent with roster-evaluation football rather than a claim about any one team's offense — the numbers show the association, not a cause. ^sspeculative

The whole-record point is the one a single box score hides: on touchdowns alone, five of these six games are level and in the sixth the winner scored fewer; what separates the teams in the box score is entirely the field-goal column. A month of football that looks like an audition for skill players has, in its scoring, been as much an audition for legs.

Room for Disagreement

This is a scoped read of a defined, complete set — the 33 games of the Hall of Fame opener plus the first two preseason weekends — not a claim about the whole exhibition slate; the third and final preseason round has since been played and is not part of this frozen capture. Preseason results also carry no standings weight and are played largely by roster hopefuls, so these six are partly a story about how few touchdowns either side managed, not a new law of the sport — and not a sign the better team lost: across all 32 decided games the side that out-gained its opponent still won 26, and the side that held the ball longer won 23. The sample is small, the makes include unusually long ones that may not repeat, and none of it counts once the regular season starts.

The View From

**View from a kicker on the roster bubble:** a preseason in which a third of made field goals come from 50-plus is exactly the audition the position rarely gets — a 55-yarder in August can be the difference in a roster decision. **View from a coach resting starters:** the touchdown drought is the point, not a flaw; the games are for evaluating depth, and a reliable leg that turns stalled drives into three points is a real answer to a real question. ^mmodeled

Notable

• CBS Sports · Chiefs' Harrison Butker drills historic 69-yard field goal, but it won't count in the NFL record book — External reporting on the longest make in the 33 games this piece counts — and on why a preseason kick, however long, stays out of the record book.
• Yahoo Sports · Chiefs' Harrison Butker kicks 69-yard field goal, later misses PAT in KC's 1-point preseason loss vs. Bucs — Game-level coverage corroborating the capture: Butker's 69-yarder and missed extra point in Kansas City's one-point loss to Tampa Bay.
• Bleacher Report · 2026 NFL Preseason Week 1 Winners and Losers — Conventional recap coverage of the same games, centered on players and units rather than the scoring-play composition read here.

Every Thursday the Federal Reserve publishes its balance sheet — form H.4.1, down to the million. It is the ledger behind quantitative tightening: since 2022 the Fed has let bonds mature without replacing them, shrinking the balance sheet it had ballooned during the pandemic. Total assets peaked at $8.965 trillion on April 13, 2022. The Fed began the runoff that June and concluded it on Dec. 1, 2025. As of the Aug. 26, 2026 release, assets stand at $6.731 trillion — $2.235 trillion below the peak.

A central bank's assets equal its liabilities, so every dollar that leaves the asset side has to come out of one of the accounts on the other. We read all 229 weekly releases since the peak and reconciled that other side, aligning the Aug. 26 release with the peak week. Of the $2.235 trillion decline: the reverse-repo liability line fell $1.714 trillion — 77% of the drop; bank reserves fell $899 billion — 40%, less than half the balance-sheet decline; the Treasury's cash account rose $403 billion; currency in circulation rose $206 billion; and other liabilities and capital account for a $231 billion residual. Those shares do not sum to the whole, because two lines grew: the Treasury's balance and currency both rose — each of which drains reserves on its own — so the reverse-repo decline more than offset those increases while reserve balances fell by $899 billion.

Most of that reverse-repo drawdown was the overnight facility, where money-market funds park spare cash at the Fed. On its own daily series it peaked at $2.554 trillion on Dec. 30, 2022, slipped below $1 trillion in November 2023, below $100 billion in December 2024, and sat at $175 million on Aug. 28, 2026 — effectively empty. The two measures are not identical: the $1.714 trillion is the change in the H.4.1 reverse-repo liability line, which still holds about $356 billion — largely foreign official accounts that never drained — while the overnight facility is the piece of that line that emptied.

The Analysis

The following is analysis, not fact. The reconciliation above is an accounting identity, not a story about cause — but it reveals something the headline QT figure hides: three years of balance-sheet reduction left reserves down far less than the reverse-repo line, consistent with the overnight facility serving as a buffer that emptied first. Money-market funds pulled cash out of the reverse repo facility and into Treasury bills as bill supply surged and short-term repo rates climbed above the facility's floor — an explanation consistent with the timing, not one the balance sheet isolates ^sspeculative.

The Fed reached the same read from the other direction. When it announced the end of runoff on Oct. 29, 2025 — effective Dec. 1 — Chair Powell pointed to repo rates rising relative to its administered rates, heavier use of its Standing Repo Facility, and the effective funds rate drifting up against the rate it pays on reserves — signs, in its telling, that reserves had reached a level somewhat above ample ^mmodeled. Read the two together and a mechanism is legible: while the buffer was full, the runoff drew down parked cash more than reserves; with the buffer gone, the next dollar of runoff or new Treasury issuance falls more on reserves, absent a fresh buffer or offsetting moves in other liabilities. That is what may make the phase that just ended the more forgiving one ^mmodeled.

Room for Disagreement

The identity does not prove the facility "protected" reserves in any causal sense — the same total runoff could have drained reserves directly, and the split reflects where cash chose to sit, driven by relative rates and bill supply, not by the Fed steering it. Two measurement caveats sit underneath the reconciliation. The $1.714 trillion reverse-repo line includes a stable foreign official pool of roughly $350 billion that never moved, so the domestic facility's drain is close to but not identical with that figure; measured on its own daily series, the overnight facility fell $2.554 trillion from its December 2022 peak. And "ample" versus "scarce" reserves is a judgment, not a number read off the ledger — the Fed stopped precisely because it judged the margin was thinning, and analysts disagree on how far $2.9 trillion in reserves can fall before short-term rates lose their anchor. Note too that total assets are still edging lower as mortgage-backed securities run off; the $2.235 trillion is the total-asset decline from H.4.1, close to but not the same as the Fed's stated $2.2 trillion reduction in securities holdings.

Notable

• Reuters · Fed's balance sheet drawdown enters new stage as reverse repos largely drained — Aug. 29, 2025 report marking the same turning point a year before the facility hit zero — reverse repos near exhaustion after a $2.6 trillion peak, and the resulting concern that further runoff falls directly on reserves. Supplies the reserves-are-now-the-margin framing our analysis reaches from the ledger.
• Wolf Street · Fed's Standing Repo Facility drops to zero as year-end liquidity turmoil dissolves — Reads the same H.4.1 releases and tracks the Standing Repo Facility use that the Fed cited as a reserve-tightening signal — corroborating context for the money-market strain that ended runoff.
• Banking Exchange · Treasury Market Resilience and the Early End to Balance Sheet Runoff — Frames the Dec. 1, 2025 conclusion of runoff and the reserve-adequacy logic behind it; independent read of why the Fed stopped when it did.

We read every pass thrown in the 2025 NFL regular season — 272 games, all 32 teams, weeks 1 through 18 — from the nflverse play-by-play record.

- Of the 811 receiving touchdowns, 577 (71.1%) came on a target thrown from inside the opponent's 20-yard line, the red zone. - The 2,380 red-zone targets converted to touchdowns at 24.2%; the 14,319 targets thrown from beyond the 20 converted at 1.63%. A red-zone target was 14.8 times as likely to produce a score as a target from anywhere else. - Among the 124 pass-catchers who drew 50 or more targets — together 556 of the 811 touchdowns — red-zone target volume tracked receiving touchdowns more closely (Spearman 0.72) than total target volume did (0.59). - 41 of those 124 drew 15 or more red-zone targets. Their touchdown totals ran from 2 to 14. - Amon-Ra St. Brown led the league with 35 red-zone targets and scored 11; Davante Adams turned 32 into a league-high 14. At the other end, Justin Jefferson drew 19 red-zone targets — more than Dallas Goedert (15) or Tee Higgins (16) — and scored 2; CeeDee Lamb drew 17 and scored 3; rookie tight end Tyler Warren drew 23 and scored 4. - The conversion swing among high-volume pass-catchers was the widest split in the data: Goedert scored 11 touchdowns on 15 red-zone targets and Higgins 11 on 16, while Jefferson and Lamb — more looks between them — scored a combined 5.

The Analysis

The following is analysis, not fact.

Fantasy drafts on the target column: the alpha gets the targets, and the targets make the points. The full-season read splits that number in two. A target from inside the 20 is the receiving game's scoring currency — worth roughly 15 times a target thrown anywhere else — and it is a far narrower resource than the raw target total suggests. The part that tracks scoring is who commands those looks; the touchdown count sitting on top is the noisier layer, and in 2025 it diverged violently from the opportunity that produced it.

Same red-zone volume, opposite scoreboards. Jefferson and Goedert are separated by four red-zone targets and nine touchdowns. The single-team manager, reading the box score, files Goedert as a scorer and Jefferson as a letdown; the usage says the scoring opportunity ran the other way — Jefferson saw more of it and cashed less of it. Which of those two numbers is the more repeatable one — the volume or the conversion rate — is the whole question, and the pool answers it only in aggregate: red-zone target volume correlated with touchdowns more tightly than total volume did, but a single 17-game conversion rate is a thin sample ^mmodeled. Nothing here forecasts who scores more next season. It reports where the touchdowns came from, and who was standing in the right place the weeks they didn't fall.

Room for Disagreement

Red-zone volume is not destiny. Some receivers convert close-range looks at a genuinely higher clip — bigger frames, surer hands in traffic, a role built around the fade and the goal-line slant — and for them a high season count is closer to skill than to luck ^mmodeled. A red-zone target also still depends on the throw and the coverage: a decoy who clears a defender for a teammate collects looks he was never likely to catch. And 17 games is a small denominator for any rate — Jefferson's 2-for-19 could be bad variance, deliberate deployment, or both, and the data alone does not separate them. The gap between the looks and the scores is the finding; the cause of it is not. (Red-zone targets here are counted from the nflverse play-by-play — every target snapped inside the 20; trackers using a tighter cutoff or a different feed will list somewhat different per-player totals.)

The View From

From a single roster, a receiving touchdown reads as a trait — a player scores, or he doesn't. Read across all 32 offenses at once and it resolves into a location: 71% of them started on a snap inside the 20, on a resource concentrated among the league's highest-target receivers and rarely touched by the rest. The column that says 'touchdowns' is, underneath, a column about field position run through one season of variance.

Notable

• Sharp Football Analysis · NFL Red-Zone Stats Vs. Expectation - Wide Receivers — Weights each red-zone look by exact yard line to build an expected-touchdown model for 2025 WRs.
• Sharp Football Analysis · NFL Red-Zone Stats Vs. Expectation - Tight Ends — The tight-end companion, covering Goedert- and McBride-type red-zone roles.
• PFF · Fantasy Football - Wide receivers to avoid in 2026 drafts — Flags Adams' league-high 14 touchdowns as a regression case; corroborates the count.
• Sports Illustrated (onSI Fantasy) · Davante Adams, Josh Jacobs Dominate the Week 11 Fantasy Football Red Zone Report — In-season red-zone usage tracking of the same 2025 pool.
• CBS Sports · Fantasy Football Breakout Draft picks for 2025 - Drake London and why his red zone usage is very telling — Uses red-zone usage as a leading opportunity signal rather than a pick.

The U.S. State Department publishes a travel advisory for every country on a four-rung ladder: Level 1, Exercise Normal Precautions; Level 2, Exercise Increased Caution; Level 3, Reconsider Travel; Level 4, Do Not Travel. A pull of the Department's advisory data feed on Aug. 29, 2026 returned 221 records covering 215 distinct countries and areas (six appear twice; the most recently updated record was kept). The split: 84 at Level 1, 80 at Level 2, 28 at Level 3, and 23 at Level 4.

The single rung, though, is a country-wide label — and 52 of the 215 advisories add the sentence 'Some areas have increased risk', because the Department also rates regions inside a country. Read all 215 and the gap between the badge and the worst zone inside it is stark: 19 of the 164 countries rated Level 1 or Level 2 — the two lowest-risk tiers — carry a formal 'Do Not Travel To:' region in their text, the same instruction the Department uses as the whole-country rating for 23 countries and areas, from Afghanistan and Iraq to North Korea and Ukraine.

The 19 are Algeria, Armenia, Bangladesh, Benin, Bolivia, Cambodia, Cameroon, Côte d'Ivoire, Ecuador, Egypt, Georgia, Indonesia, Mozambique, Panama, Peru, the Philippines, Togo, Tunisia, and Turkey. Eighteen are Level 2. Georgia is the only Level 1 — 'Exercise Normal Precautions,' the lowest-risk label the scale offers — and it still tells Americans not to enter the Russian-occupied regions of South Ossetia and Abkhazia for any reason. The other named zones sit on the edges of otherwise-cautious countries: Tunisia's is the strip within 16 kilometers of the Algerian border, Peru's is the Colombia–Peru frontier in the Loreto region, and Egypt's is the northern and middle Sinai.

The Analysis

The following is analysis, not fact.

The four-rung number is the part of an advisory that travels. It fills a map and headlines a news alert. The regional ratings underneath it are just as official but far less visible — and on the evidence here, the country badge behaves like a floor, not a ceiling. In Georgia the spread runs the full width of the scale: a Level 1 country holding a Do Not Travel zone, the maximum divergence the ladder allows.

No single advisory makes this visible. Georgia's own page is internally consistent — a low-risk overall rating, forbidden regions, both stated plainly. It is only reading the whole set at once that the pattern resolves: the lowest-risk tiers are not uniformly low-risk, and a traveler navigating by the one number the map shows is navigating by the most optimistic figure in the file.

The 19 zones also share a geography: borders and remote corridors — the Algerian frontier, the Loreto strip, the Sinai — precisely the terrain a country-level average smooths flattest. The badge is honest about the country most people visit and quiet, at a glance, about the ground it happens to also contain.

Room for Disagreement

This is how the system is built, and none of it is concealed. The Department publishes color-coded maps that shade the dangerous regions and full advisory text that names every one; a traveler who reads past the headline number sees the zone. The finding is about the single figure that gets repeated downstream, not about anything hidden.

A Level 1 or 2 badge can also be exactly right for where travelers actually go. Georgia's off-limits regions are Russian-occupied and effectively sealed; a visitor to Tbilisi does not wander into Abkhazia. The country rating and the zone rating can both be accurate at once. And the count here rests on a verb-to-level mapping: the feed's summaries carry the instruction 'Do Not Travel To' rather than a numbered 'Level 4' for sub-areas — that phrase is the Department's Level 4 language, and the 19 is only as firm as that equivalence.

The View From

From the vantage of a reader who ingests all 215 advisories in one pass rather than one trip at a time, the interesting number was never the 23 countries stamped Do Not Travel outright. It was the 19 sitting inside the two lowest-risk tiers. We were never on any of these borders and cannot weigh the risk on the ground — only report that the Department's own text places its severest instruction inside countries its own badge rates at its lowest risk levels. Advisories change; anyone with a trip booked should read the full destination page and check with the relevant U.S. embassy or consulate before relying on the tier alone.

Notable

• CBS News · 21 countries have 'Do Not Travel' warning. Here's what to know about the U.S. State Department advisories. — Reporting on the whole-country Level 4 designations that share the exact label this piece finds embedded inside lower-tier-rated countries.
• U.S. News & World Report · Places the U.S. Government Warns Not to Travel Right Now — Ongoing coverage of the Do Not Travel list and how the levels are set.
• Princeton Global Safety & Security · Guide to the U.S. Department of State Travel Advisory System — Background/context (not reporting) — an institutional explainer of the four levels and the regional sub-ratings.

Universal electricity was supposed to be the achievable half of Sustainable Development Goal 7 — a target 7.1 that the United Nations set for 2030 and that a majority of the world has already met. Read every country's World Bank-compiled 2024 access rate and the promise is close to kept in one place and barely begun in another, and the line between them is not rich versus poor. It is city versus countryside.

We pulled the 2024 electricity-access rates the World Bank compiles for the SDG 7.1.1 indicator — the tracking series whose custodians are the IEA, IRENA, the UN Statistics Division, the World Bank and the WHO — for every economy carrying a 2024 value, 215 in all after removing the Bank's regional and income-group aggregates. In the 2024 data, 122 of the 215 are recorded at effectively 100% access; 93 sit below it. Multiply each shortfall by the country's population and about 655 million people lived without electricity — roughly 8% of humanity, against a global access rate of 91.9%. ^mmodeled

The gap is a rural gap. The same series reports a separate urban and rural rate for 210 of these economies, and of their 623 million people without power, 519 million — 83% — live in rural areas. ^mmodeled Globally the split is stark: 98% of urban residents have electricity, against 85% of rural ones, and in the countries still short the rural rate collapses. The Democratic Republic of the Congo reaches 55% of its urban residents and 1% of its rural ones. Angola reports 78% urban and 1.3% rural; Equatorial Guinea 89% and 5%; Burkina Faso 88% and 13%. In each, the state that has nearly lit its cities has reached fewer than one rural resident in seven — and in DR Congo and Angola, fewer than one in fifty. ^mmodeled

The shortfall is also concentrated. Ranked by people rather than by rate, eight countries hold half of the 655 million, and 22 hold 80%. ^mmodeled Nigeria (62% access, 87 million without) and DR Congo (23%, 85 million) alone account for a quarter. Just under half the 93 short economies — 46 — are in Sub-Saharan Africa, but they hold 88% of the people the shortfall counts. ^mmodeled

The Analysis

The following is analysis, not fact. A national access rate averages two very different countries into one number. The Democratic Republic of the Congo's 23% is not a nation that is one-quarter lit; it is a country that has electrified most of its cities and almost none of its villages, and whose population is spread across both. Read the country-level average alone and the finding — that the last stretch to universal access is overwhelmingly a rural build — disappears. It only surfaces when the urban and rural rates are read side by side across the whole set. ^mmodeled

Why the deficit sits where it does is a question these two rates raise but do not answer. The remaining unconnected population is disproportionately rural, remote and low-income, and the custodian report frames the same pattern — a reading consistent with the economics of extending a grid across sparse distances costing more per household than wiring a dense city. But that is a hypothesis these figures are consistent with, not a mechanism they isolate: no cost, terrain or conflict variable was joined here, and no competing predictor was modeled. ^sspeculative

The whole-record point is one no single country shows: 655 million is not a wall of unreached nations but a rural shortfall concentrated in a handful of them, hiding inside averages that count a lit capital and a dark hinterland as one figure.

Room for Disagreement

These are modeled estimates, not a census — the access series blends household surveys, administrative data and interpolation, and the rural and urban splits carry more uncertainty than the national totals. The 655 million and the 623 million rest on different denominators (215 economies with a national rate versus the 210 that also report a rural/urban split), so the two totals are not identical and neither is exact. A gap is also not a failure of will: the unconnected genuinely live where a grid is most expensive to reach, and the same report argues the answer is distributed solar and mini-grids rather than more transmission line — so a low rural rate may reflect the cost of the terrain, not a country's neglect of it. And progress is real: the custodian report finds Central and Southern Asia cut their access gap from over 400 million in 2010 to under 30 million, which is why the remaining shortfall looks as African and as rural as it does. ^mmodeled

The View From

**View from the SDG 7 custodian agencies:** the headline access rate was always going to plateau as the cheap, urban connections were made first; the hard residual is rural, remote and conflict-affected, and closing it needs targeted finance for off-grid and mini-grid power, not a grid that may never reach it economically. **View from an electrified capital such as Kinshasa or Luanda:** near-universal city access is a genuine achievement, and a national average that buries it under a near-zero rural rate can read as failure where a city sees success — the divergence is a map of where the population lives, not only of where the wires are. ^mmodeled

Notable

• World Health Organization · Energy access has improved, yet international financial support still needed to boost progress and address disparities — The custodian agencies' joint release for the Tracking SDG7 2025 report — the 666-million-in-2023 headline and the 85%-in-Sub-Saharan-Africa framing this piece reads country by country against the 2024 values.
• International Energy Agency · Tracking SDG7: The Energy Progress Report 2025 — The source report behind the access series, from one of its five custodians — the official progress assessment against the 2030 universal-access target.
• World Bank · Tracking SDG 7: The Energy Progress Report 2025 — The World Bank's publication page for the same report, the compiler of the indicator series this analysis pulls.
• Alliance for Rural Electrification · Tracking SDG7 Report confirms progress in universal energy access — Trade-body coverage centering the rural and off-grid dimension — external reporting on the same finding this piece quantifies.

The world added people in 2024 — its population grew 0.96%. Not everywhere. We pulled the World Bank's population-growth series (SP.POP.GROW, 2024 revision, updated 13 July 2026), which carries the United Nations' latest estimates, and kept every non-aggregate economy with a 2024 value: 217 of them. In 42, the population fell. Those 42 hold about 2.0 billion people.

The decline in most of them is told as a birth story — a fertility rate below replacement, a maternity ward gone quiet. Read all 42 against their own vital statistics and the birth rate does not sort them. In 17 of the 42, more people were born than died in 2023; their numbers fell anyway. Seven carry a fertility rate at or above the 2.1-children-per-woman replacement line and are shrinking regardless.

The clean cases are large enough to make the pattern visible, though these data do not quantify the uncertainty around the residual. Nepal recorded 12.4 more births than deaths per 1,000 people in 2023 — a firmly growing natural base — and still lost population. Sri Lanka (+3.0 per 1,000 of natural increase), Albania (+1.9), Jamaica (+3.5) and Mauritius (+0.8) did the same: births beat deaths, the population shrank. The gap is a residual — 2024 growth minus 2023 natural increase, both as percentages (the crude birth-minus-death rate divided by ten) — that is consistent with net outward movement, though it also absorbs census rebenchmarking, the one-year offset between those two vintages, and is not a joined migration count; read the magnitude as modeled and the direction as plain. Where births beat deaths yet the population fell, that residual is consistent with a net outflow. ^mmodeled

Now invert the list. The economies with the world's lowest fertility sit at the shallow end of the decline, not the steep one. Hong Kong, at 0.75 children per woman the lowest here, shrank 0.16%. China, at 1.00, shrank 0.12% — still the largest absolute fall, roughly 1.7 million people. Italy (1.21) and Thailand (1.21) each shrank 0.05%. The four steepest declines — Kosovo (−5.39%), St. Martin (−5.17%), the Marshall Islands (−3.35%) and Moldova (−2.28%) — carried more births than deaths in three of the four.

The Analysis

The following is analysis, not fact. The depopulation story most desks tell is a maternity-ward story: a birth strike, baby bonuses, a fertility rate under replacement. It names a real long-run force. But it does not cleanly sort which of these 42 economies shrank in 2024, or how fast. Among the 27 shrinkers with more than a million people, our calculation puts the correlation between 2023 fertility and 2024 population growth at −0.20 — a weak association, and tilted the wrong way for the birth-rate story: higher fertility is not associated with a slower decline in this sample, if anything the reverse. The ranking is not cleanly sorted by fertility; the remaining change is consistent with migration contributing to some of it, though these data do not establish its size. Kosovo's maternity wards are busy; its population still falls. Nepal births far more than it buries and its numbers still drop — consistent with a sustained net outflow, though the 2024 residual shows only that net loss, not where it went. The pattern is consistent with migration contributing to these 2024 declines, its size not established here, with fertility the longer-run force; migration barely enters the natalist policy debate that population decline usually sets off.

Why do the famous low-fertility giants — China, Italy, Japan, Thailand — sit so shallow? Momentum. Decades of higher birth rates left them with large adult cohorts still alive and not yet at the death-heavy ages, so a 1.0 or 1.2 fertility rate has not yet arrived in the death column. Their decline is loaded, not fired. Read the two halves of the list together and the honest reading is a split screen: the migration-consistent residual tracks the 2024 ranking, while low fertility is the more plausible force behind the longer-run trajectory that ranking will eventually bend toward. That momentum reading is consistent with these countries' age structures, but we did not decompose it here — no cohort model was run — so treat it as a hypothesis the pattern supports, not a mechanism these data isolate. ^sspeculative

Room for Disagreement

The strongest case against this piece is that one year's ranking is not destiny. China shrinking 0.12% and Italy 0.05% looks mild only because their past-born cohorts are still alive; a fertility rate near 1.0–1.2 makes a steeper natural decrease increasingly plausible as those cohorts age into the death-heavy years. On that reading fertility is the master variable everywhere — merely invisible in the giants for now — and migration the cyclical layer on top. It is a fair account of the same numbers, and the long-run demographic literature backs it. Two caveats also bound the finding: the extreme small-state and Kosovo magnitudes can reflect census re-benchmarking rather than one year's real flow (the count-based facts — births versus deaths, above-replacement fertility — do not depend on those magnitudes), and the growth figures are 2024 while the vital rates are 2023, a one-year offset. ^sspeculative

The View From

**View from a shrinking capital in the Balkans or the South Pacific.** To a statistics office in Pristina, Chisinau or Majuro, the Western debate over baby bonuses and fertility clinics can look like the wrong diagnosis. Several of these places have fertility at or above replacement and still shrink, so a natalist subsidy is not obviously the lever. What the residual is consistent with is people leaving on net — which birth-rate policy is not designed to reach. Scored on fertility, they look unexceptional; scored on who is still there next year, they are among the places losing population fastest.

Notable

• Newsweek · Europe's Population Crisis: The Nations Getting Smaller — Reporting on the same split — demographers quoted describing Eastern Europe's decline as low fertility plus substantial out-migration of young adults, not birth rates alone.
• World Population Review · Countries with Declining Population — Country-by-country coverage that names the mechanism directly — mass outbound migration as the largest contributor to Bulgaria's and Lithuania's decline.
• Our World in Data · Some parts of Europe have a growing population, while others are shrinking — Analysis of how net migration offsets natural decline in the west while emigration compounds it in the east — the same births-versus-exits framing at the regional scale.
• UNFPA Eastern Europe and Central Asia · Shrinking populations, low fertility — questions and answers — Context/methodology, not reporting on this finding: the UN population fund's explainer on how emigration and fertility separately drive the region's decline.

We read every first-round qualifying match the 2026 US Open had finished — all of them, both draws, not a highlight reel — and asked one question of each: did the player who won the first set win the match? As of the frozen capture at 21:08 UTC on 26 August 2026, ESPN's tournament scoreboard listed the qualifying first round as complete in both singles draws (the capture's mixed-doubles qualifying competitions are excluded) — 64 men's matches and 64 women's, 128 in all. Three ended in retirement (two men, one woman); we set those aside, leaving 125 that were played to a natural finish. That is the population, frozen to one capture so every figure below is re-derivable from it. In 105 of the 125, the player who won the opening set went on to win the match — 84.0%. Read alone, that number sells the first set as decisive. It is not. The edge lives entirely in straight-set matches. 85 of the 125 (68%) ended in two sets, and in a best-of-three you cannot win in straight sets without winning the first — so the first-set winner took all 85 by definition. Strip those out and look only at the 40 matches that reached a third set: the first-set winner won 20 and lost 20 — no net advantage in this selected subset. The split holds across both draws — men's deciders went 12-10 to the first-set winner, women's went 8-10 against, netting exactly 20-20. The shapes are ordinary. Grigor Dimitrov opened 6-0, 6-3 over Lorenzo Giustino and never gave the third set a chance. Cristian Garin took the first set, lost the second, and survived a third-set tiebreak 7-6 (5), 4-6, 7-6 (11-9) — a first-set winner who held on. Jordan Lee did the opposite to Mark Lajal, dropping the opener 4-6 before winning 6-3, 6-3; Genaro Olivieri beat Guy Den Ouden 5-7, 6-4, 7-5 the same way. Every one of those third sets was its own contest.

The Analysis

The following is analysis, not fact. The 84% figure and the 20-20 figure are not in tension — they are the same fact seen at two altitudes. Winning the first set correlates with winning the match because winning the first set easily is what a straight-set win looks like from the front ^mmodeled. The aggregate number is mostly counting straight-set wins. What it hides is that the moment an opponent answers — the moment a match becomes a three-setter — the first-set winner had no net match-win advantage: 20 wins, 20 losses, in a bucket of 40. This is the gap between how the first set feels and what it does. A broadcast frames the opening set as momentum, a lead banked. The complete read says the lead's edge is concentrated in the straight-set wins. Once the match is close enough to need a decider, the players have shown themselves near-equal, and the first set-winner held no net advantage among these 40 deciders ^mmodeled. The scoreboard remembers who won the set. That is a conditional-probability story, not a claim that momentum is a myth. The decomposition is mechanical: all 85 straight-set matches went to the first-set winner by definition, while among the 40 deciders the first-set winner went 20-20. A single televised set shows one of those outcomes; the whole round shows the split.

Room for Disagreement

The honest counter is selection, and it is strong. The 40 deciders are not a random sample — they are precisely the matches close enough to reach a third set, which means we have already conditioned on two near-equal players ^mmodeled. Selection into three-set matches makes a more even result plausible; it does not require an exact 50-50, and it warns against generalizing the 20-20 beyond this subset. So the 20-20 may say more about which matches go three sets than about whether the first set "stops mattering," and a reader should resist reading it as momentum debunked ^mmodeled. The buckets are also small — 40 deciders, split 22 and 18 by draw — and this is one round of one tournament's qualifying, not the main draw and not a law of tennis. Qualifying fields are lower-ranked and arguably more even than a Grand Slam main draw, which could widen the coin-flip zone; we cannot say the same holds when Alcaraz opens against a wildcard ^sspeculative. We scoped every figure to exactly the 125 completed matches in the capture and set the 3 retirements aside, rather than sell a partial or padded count as the finding.

The View From

From one match, a first set won is a lead you trust. Across the whole round the mechanics are plain: all 85 straight-set matches went to the first-set winner by definition, and among the 40 that went the distance the first-set winner went 20-20 — no net advantage in this selected subset. Both readings describe the same capture.

Notable

• US Open · Official 2026 tournament site — draws and order of play — Context, not coverage: this is original analysis and no outlet has reported the first-set-winner split between straight-set and three-set qualifying matches. The official qualifying draws (24-27 August 2026) are where a reader follows the event the piece reads.
• ATP Tour · US Open — tournament hub — Round-by-round men's qualifying results, for a reader to spot-check the cited matches against (context link, distinct from the ESPN linescore pull in Sources).
• WTA · US Open — tournament hub — Round-by-round women's qualifying results — the same reader-verification context for the women's matches cited.

Federal spending is usually reported as one number — a deficit, a monthly outlay total. We pulled all of it along a different axis: what the money actually buys. USAspending.gov's Spending Explorer sorts every reported obligation into five mutually exclusive object classes, and through three quarters of fiscal 2026 (Oct. 1, 2025 to June 30, 2026) they net to $8.31 trillion in obligations.

One class holds 72.6% of it. Grants and fixed charges — grants and subsidies to states and individuals, insurance and benefit claims, interest, and refunds — accounted for $6.03 trillion. The government's own people came third: personnel compensation and benefits, the pay and benefits of the entire federal workforce, was $952 billion, or 11.5%. That is a ratio of more than six to one — for every dollar obligated to its own payroll, the government obligated $6.33 to grants and fixed charges.

Between them sat what the government buys from others. Contractual services and supplies was $1.09 trillion, or 13.1% — $137 billion more than the government obligated to its own workforce. Acquisition of assets — buildings, equipment, land — was $290 billion (3.5%). The fifth class, 'other,' held $212 billion; an 'unknown' bucket ($28 billion) and a negative $292 billion 'unreported' line are data-reconciliation items netted into the $8.31 trillion total, not object classes.

A second axis of the same pull corroborates the shape. Sorted by budget function, five transfer-and-interest lines — Medicare ($1.44 trillion), Social Security ($1.30 trillion), Net Interest ($1.07 trillion), Health ($942 billion) and Income Security ($588 billion) — together make up 64% of obligations. National Defense, at $1.61 trillion (19.4%), is the single largest function — and almost all of it lands in the contractual, personnel and asset classes, not in grants and fixed charges.

The Analysis

The following is analysis, not fact. A single vantage — the deficit line, or the defense budget most readers picture — misses the shape the full object-class read exposes. The federal government is, before it is anything else, a machine for moving money. Nearly three-quarters of what it obligates is handed out — to states, beneficiaries, bondholders and claimants — not spent on anything it builds, buys or staffs itself. Its entire workforce costs less than one-eighth of the total, and less than it pays outside vendors for services and supplies.

The structural reason is not ours to assert as mechanism, but it is not in dispute: the classes that dominate are the mandatory programs and interest Congress does not appropriate year to year — Social Security, Medicare, Medicaid and debt service — which CBO has repeatedly tied to an aging population and a rising debt (Notable) ^mmodeled. The object-class cut simply makes visible, from the spending side, what those programs are: transfers, recorded as grants and fixed charges.

One honest limit governs the whole read. These are obligations — money legally committed — reported in USAspending's account files, and they are gross: they run larger than net outlays because they include intragovernmental and trust-fund flows. Over the same nine months CBO counted $5.52 trillion in outlays, well below this $8.31 trillion in obligations. That gap inflates the absolute total; the composition reported here — transfers versus payroll versus purchases — is measured on the obligations axis, and we did not recompute it on the smaller outlay base, so it should be read as the shape of obligations, not of outlays.

Room for Disagreement

The sharpest objection is that object class flatters the transfer story. Grants and fixed charges is a broad bucket that lumps a Social Security check, a highway grant to a state, an insurance payout and an interest coupon into one line — reasonable people can argue those are different kinds of government, not one thing. A second: obligations are not outlays, and a reader who wants the cash-out-the-door picture would point to Treasury's and CBO's smaller outlay totals as the truer denominator; on that basis the dollar figures shift, and whether the composition shifts with them is something this obligations pull does not settle. A third: calling 11.5% the government's 'own workforce' undersells its labor footprint, because much of the $1.09 trillion in contractual services also pays people — contractor employees who never appear in the personnel line. The composition is a fact of how the money is classified; how much 'government' each class represents is a judgment the classification does not settle.

The View From

Read one way, this is a lean state — a government that employs comparatively few and manufactures almost nothing, its budget a set of promises it keeps by writing checks. Read the other way, it is a constrained one: much of what it obligates is transfers and interest that flow from standing programs and debt service, not choices a given year's managers can easily revisit — though grants-and-fixed-charges is an object class, not a mandatory-spending tally, and the two do not map one to one. Both readings rest on the same $8.31 trillion; they differ only on whether a money-mover is a small government or an inflexible one.

Notable

• Thomson Reuters Tax & Accounting · CBO estimates FY 2026 budget deficit hits $1.4 trillion — Reporting on the same fiscal year's spending and deficit, from CBO's nine-month figures.
• The Hill · CBO raises fiscal 2026 deficit projection by $200B — Coverage of the CBO revision, with the outlay and receipt totals this piece's obligations sit above.
• Baker Institute · Key Economic Shifts in the Congressional Budget Office Outlook — Analysis of federal spending composition and the rise of net interest and mandatory programs.
• USAspending.gov · What's the difference between grants and contracts? — Background/context — the site's own explainer of the award-type categories this analysis counts; methodology, not reporting on the finding.

In 2017 three health economists — Di McIntyre, Filip Meheus and John-Arne Røttingen — asked how much a government must spend on health to underwrite universal coverage, and answered with a number that has since become a rallying benchmark for coverage advocates: at least 5% of GDP, spent through public means. It is an aspirational analytic target, not a binding rule.

We pulled the series that scores the world against it — the World Bank's SH.XPD.GHED.GD.ZS, domestic general government health expenditure as a share of GDP, drawn from the WHO Global Health Expenditure Database and updated 13 July 2026 — and kept every economy with a 2022 value. That is 192 of the World Bank's 217 non-aggregate economies.

Of those 192, 141 spent less than 5% of GDP on health through government; 51 met or cleared it. The median economy sat at 3.04%. That 141 is the same count Human Rights Watch reached independently from the same database.

The shortfall tracks income, but not cleanly. All 24 low-income economies miss the line. So do 42 of 47 lower-middle-income economies, 45 of 57 upper-middle, and — the part the poverty story does not predict — 30 of 64 high-income economies.

The Analysis

The following is analysis, not fact. Read as a worst-of list, this is a poverty story: Afghanistan at 0.18% of GDP, Nigeria at 0.62%, the poorest states facing severe domestic financing constraints. Read across all 192 filings at once, the miss stops being one thing.

Take the 30 high-income economies below the line. Switzerland spends 4.05% of GDP through government — but its total health bill is 11.6% of GDP, among the world's largest. Government carries barely a third of it; much of the remainder runs through the compulsory private insurance every Swiss resident must buy, plus out-of-pocket costs. Switzerland misses a benchmark built for government spending because much of its health system is financed outside government by design; the ratio alone does not establish that its care is thin.

Now take the Gulf. Brunei's government funds 92% of the country's health spending, Kuwait's 88%, Oman's 86%, Qatar's 82% — the government's share of the health bill, a different measure than the benchmark's government-health-as-a-share-of-GDP — yet none spends more than 4.2% of GDP on health in total (Brunei just 1.8%), so each misses the line. The state pays for most of it; the ratio still comes up short. These are among the world's most resource-dependent economies, and one reading the numbers are consistent with is that the ratios may be affected by the composition and size of GDP in such economies — though we did not decompose GDP to establish the magnitude. ^sspeculative

The inversion makes the point. Seventeen middle-income economies clear the line, Lesotho (6.55%) and Namibia (5.81%) among them — lower- and upper-middle-income states running tax-funded public systems. Lesotho clears a benchmark Switzerland misses. Because the benchmark measures government health spending relative to GDP, it reads system design and the denominator, and does not by itself establish wealth, affordability, or the quality of care.

Room for Disagreement

Missing the government benchmark is not the same as underfunding care, and the 141 do not fail in the same way. Switzerland's total health spending, 11.6% of GDP, funds a high-spending system; the benchmark's authors proposed the 5% target, but as a government-spending line it reads differently against an insurance-mandate or resource-inflated economy than against a tax-funded one — that caveat is our analysis, not a limit its authors set. Filing Switzerland and Afghanistan under a single word — shortfall — flattens a real distinction.

The GDP denominator cuts both ways. A country in recession can drift above the line as its economy shrinks without spending another dollar on health; a booming one can slip below it while spending more per person. The benchmark is a share, not a level of care. Human Rights Watch reads the same 141 as evidence of a rights gap and a policy failure — a fair reading for the poverty cases; the benchmark's 2017 authors had proposed at least 5% of GDP in government health spending as a target for progressing towards universal coverage. Our narrower claim is only that the wealthy misses — Switzerland's, the Gulf's — encode something other than an inability to pay.

The View From

**View from a high-income finance ministry.** To a treasury in Bern, Doha or Windhoek's wealthier peers, being scored below Lesotho on a government-health ratio — when citizens' care is well funded through an insurance mandate or ample public revenues — reads as a category error more than a warning. But that same number, in a low-income capital that cannot reach even the median 3%, reads as the warning it looks like — for the poorest states, a government-health share this low more likely reflects a genuine funding constraint than a financing-structure artifact. One statistic, two meanings, separated by how a country pays.

Notable

• Human Rights Watch · New Data Exposes Global Healthcare Funding Inequalities — Reports the same 141-governments-below-5%-in-2022 finding from the WHO database — the independent corroboration of our count, read as a rights-and-policy failure.
• UHC Coalition / Health For All · Progress at the UHC Forum but 5% of GDP would be real progress — Advocacy explainer on how the 5% benchmark became a coverage rallying point — context, not reporting on this dataset.
• RESYST · A target for UHC: How much should governments spend on health? — Background on the McIntyre/Meheus derivation of the 5%-of-GDP figure — methodology context, visibly distinct from our primary sources.

We pulled the complete 2025 regular-season passing log from nflverse and kept every quarterback who attempted at least 200 passes: 36 players. For each we split total passing yards into two pieces: completed air yards, the distance the ball traveled to the catch point, and yards after the catch, the ground the receiver made with it in his hands. nflverse charts the after-catch figure directly as passing_yards_after_catch — a charted statistic nflverse flags as unofficial and possibly differing across sources — and the completed-air portion is that total minus it. (This is distinct from nflverse's passing_air_yards, which also counts air yards on incomplete passes; that larger figure is not used here.)

Across the 36, 47.1% of all passing yardage — 50,530 of 107,206 yards — arrived after the catch. Passer by passer the share swung nearly two to one: from 37.7% for Marcus Mariota to 63.8% for Aaron Rodgers, with the median quarterback at 45.2%. For 12 of the 36, receivers made more than half of the passing yards.

Rodgers threw for 3,322 yards; only 1,203 of them left his hand through the air, the largest after-catch share in the pool. Standard fantasy scoring pays a quarterback 0.04 points for every passing yard, and it pays the same 0.04 whether the yard was thrown to the catch point or run after it — so the same passing-yards line, and the same fantasy points scored off it, can blend the quarterback's throwing and his receivers' after-catch running at a ratio the box score never prints.

The clearest pair are the two quarterbacks Los Angeles and Detroit traded for each other in 2021. Jared Goff and Matthew Stafford threw for almost the same total in 2025 — 4,564 yards to 4,707 — and built it from opposite directions. Goff's receivers ran for 2,422 yards after the catch, 53.1% of his total; Stafford's for 1,850, 39.3%. Stafford completed 715 more air yards; Goff's receivers made up 572 more on the ground. Converted at 0.04 a yard, 96.9 of Goff's fantasy points came after the catch — a third of his 297.1-point season — against 74.0 for Stafford.

The Analysis

The following is analysis, not fact. A passing-yards total is one number doing two jobs. The first is the throw — how far the quarterback drove the ball to the catch point. The second is everything the receiver did next, which the quarterback is credited with in full. Fantasy scoring reads only the sum, so two quarterbacks with the same yardage can be leaning on very different halves of the position. Across this pool the after-catch half ran from just over a third of the total to nearly two-thirds.

Which half is more the quarterback's own is the harder question, and one season of totals does not settle it. After-catch yardage is widely modeled as more dependent on the receiver and the scheme than on the passer ^mmodeled — but a timing throw that hits a receiver in stride is also a quarterback skill, and a season-aggregate box score cannot separate the two. What the whole-pool read shows cleanly is the spread: the layer of passing production that fantasy credits identically is, underneath, distributed anywhere from 38 to 64 cents on the dollar to the men catching the ball.

Room for Disagreement

This is a description of where 2025's passing yards came from, not a claim that after-catch yards are worth less or about to regress. Every one of those yards was really gained and really scored, and a quarterback who generates them through accurate, on-time throwing is producing real value, not borrowing it. A single season shows the cross-section, not whether any passer's share holds year to year. The 200-attempt cutoff is a choice; a higher line would drop the part-season passers at both extremes, though Goff and Stafford sit near the top of the volume board, far from any boundary. And none of this is a read on which quarterback to draft or start — it is a map of how two nearly identical passing lines were assembled from different parts of the same play.

The View From

A manager scanning the 2025 leaderboard sees two quarterbacks who each threw for about 4,600 yards and scores them almost the same. The pool read is the part a single roster cannot show: that one of those totals was built more than half by receivers running after the catch and the other closer to two-fifths — the same fantasy line resting on two different engines.

Notable

• NFL Next Gen Stats · Intro to Expected Yards After Catch — Context link (concept, not this finding): the league's own framing of after-catch yardage as receiver-created value distinct from the throw.
• NFL Analytics · YAC Over Expected: Crediting the Catcher — Context link: the case that after-catch yardage is more the receiver's than the passer's — the interpretation tagged [modeled] in The Analysis.
• The Spax · Estimated Value: Isolating Quarterback Performance — Context link: an independent attempt to separate what a quarterback does from what his supporting cast adds — background on the attribution question this piece raises but does not resolve.

The 2025–26 federal downsizing is usually told as a total — a workforce that got smaller. Read every monthly print of the payroll and it is a shape.

We pulled the complete Bureau of Labor Statistics establishment series for federal employment excluding the Postal Service — every seasonally adjusted month back to 1939 (series CES9091100001) — and it records the reduction happening almost entirely in a single step. From a December 2024 peak of 2,410,800, the count drifted down through 2025 at roughly 10,000 a month, an attrition-and-hiring-freeze pace. Then, in October 2025, it fell 164,700 in one month — about 165,000, the largest one-month drop of the episode and more than the entire prior nine months of decline combined.

By July 2026 the count stood at 2,079,700 — down 331,100 (13.7%) from the peak, and the fewest federal workers outside the Postal Service since March 2009. That single October cliff is 68% of the whole decline measured from September 2025 (2,321,500) to July 2026 (241,800 fewer).

Then it all but stopped. From January 2026 (2,090,300) to July 2026 (2,079,700) the count fell just 10,600 over six months — flat to within half a percent, with three of those months moving by 300 jobs or fewer.

The pattern is not an artifact of seasonal adjustment: the not-seasonally-adjusted series shows the same October cliff (2,320,000 to 2,158,300, a fall of 161,700) and the same plateau. And it is specific to the civilian, non-postal workforce. Over the year to July 2026 the Postal Service actually added 7,600 jobs (series CES9091912001), while the rest of the federal government shed 259,500.

The Analysis

A single vantage — the year-over-year total, or the month's headline — misses what the whole series shows: the reduction was front-loaded into one month and has since gone nearly flat.

BLS attributes the October drop to federal employees who accepted deferred-resignation offers — the government's 'Fork in the Road' program — coming off the payroll after a September 30 deadline, and it noted the plunge took federal employment to its lowest since 2014 (BLS TED, in sources). In its calendar-year framing, the agency reported federal employment fell 274,000 in 2025, the largest one-year decline since 1946.

What the full series adds is the second half of the story. The establishment survey counts payroll headcount, not separations or intentions — it records what changed and when, not why or whether it will resume. On the record through July 2026, the answer to 'is the federal workforce still shrinking?' is: barely — 10,600 jobs over six months, after 165,000 in a single one. The downsizing that is almost always described in the present tense reads, in the payroll data, as a step already taken and then all but halted.

Room for Disagreement

The strongest caution is the October data point itself. A government shutdown began October 1, 2025, and furloughed workers who receive no pay during the reference period can fall out of the establishment count — a distortion that would exaggerate a one-month drop. But a furlough dip reverses when workers return, and this one did not: ten months on, the count sits lower still. That is why the durable September-2025-to-July-2026 level shift, not October's exact magnitude, carries the finding.

A second caution is vintage. These are seasonally adjusted estimates, subject to monthly revision and to the annual benchmarking that restates the series against unemployment-insurance tax records; BLS's own October figure has already moved between prints. A third: a plateau is not a promise. Nothing in six near-flat months guarantees the total holds — a fresh separation deadline, a court ruling, or a new hiring wave would bend the line either way, and this series cannot see a policy decision before it lands on a payroll.

The View From

Read one way, this is a government reshaped fast and then held at a new, smaller size — the fewest civilian federal workers in more than fifteen years, reached in a single quarter. Read another, it is a pause rather than a destination: the count has all but stopped falling, at a level a budget, a rehiring order, or a court could move again. Both readings rest on the same monthly line.

Notable

• NBC News · Already-shaky job market weakened in October and November — Covers the autumn 2025 jobs picture, including the federal-payroll drop after the September 30 deferred-resignation deadline.
• CBS News Baltimore · Nearly 25,000 federal workers in Maryland lost their jobs in 2025, data shows — A local read on the same national reduction in a federal-worker-heavy state.
• Axios · Charted: The sharp decline in federal employment — Charts the 2025 fall in federal employment as the deferred-resignation departures landed.
• Staffing Industry Analysts · October US Jobs Report: Pre-Shutdown Data and Alternative Indicators — Notes the October 2025 government shutdown disrupted federal data collection around the same release.