The 40 Days. What A World Cup Does To The World.

Each point is the mean tone of every news event GDELT recorded that day. The shaded band is June 11 – July 20, 2026. Solid lime segments are window means; dashed cyan lines are the means of the same 40 calendar days in 2024 and 2025. Higher is less negative.

Δ is the difference in mean daily tone (World Cup window minus comparator; positive = less negative). p is from Welch's t-test on daily values; d is Cohen's effect size (0.5 = medium, 0.8 = large). The 40-day window is June 11 – July 20 inclusive; the companion 39-day analysis (June 11 – July 19) in the operations research returns the same conclusion with slightly larger effects.

Where it was strongest: the hosts

News datelined in the United States, Canada and Mexico read less negative during the 40 days than in every comparator — d = 0.64 vs before, 1.19 vs after, 1.51 vs 2025, 1.23 vs 2024 (all p < 0.006). A home tournament changes the country that holds it.

Who it reached: everyone

Splitting every event by whether its primary actor was one of the 48 participating nations: participants and non-participants improved by statistically indistinguishable amounts (difference-in-differences +0.115 to −0.034 across the four comparisons, none significant). The whole world's news cycle lifted, not just the countries on the pitch.

When it was strongest: while every flag was alive

The calm was the group stage and the round of sixteen; it faded as nations went out. Phase means with one or two days are descriptive only. Read as a design instruction: a season-long nations table keeps every flag alive for nine weekends.

• Source. GDELT 1.0 daily event files (data.gdeltproject.org/events/YYYYMMDD.export.CSV.zip), the public database that codes every news event it can parse worldwide, ~100,000 events a day. Fields used: AvgTone (−100…+100), GoldsteinScale (−10…+10, cooperation vs conflict), QuadClass (4 = material conflict), Actor1CountryCode, ActionGeo_CountryCode. Column positions verified empirically against real rows, not the codebook alone.
• Population, not a sample. Every day in five windows: 195 day-files attempted, 177 available (GDELT's own archive is missing June 14 – July 1, 2025 — verified against its master index, not a download failure). 18,604,026 events.
• Statistics. Welch's t-test on daily means; Cohen's d on pooled daily SD. Weekday adjustment: each day's tone minus the mean for that weekday across all non-tournament days (Sundays read ~0.18 more negative than Thursdays — the same order as the effect, so this check matters). Every conclusion survives it.
• Participant split. The 48 qualified nations mapped to GDELT country codes (England and Scotland share GBR; 27 of 29 ambiguous codes confirmed against real rows). Difference-in-differences on participant vs non-participant daily tone.
• Second instrument. GDELT DOC 2.0 — a full-text search over articles, a different pipeline from event coding — tone of coverage matching war OR conflict OR attack, daily, same windows. Less negative in the tournament window vs all four comparators (p = 0.028, 0.0000009, 0.00006, 0.019).
• Reproduce it. The per-day table is downloadable above. The fetch and analysis scripts are available on request; the whole run is ~175 public files and one afternoon.

• Tone is not violence. These are measures of how the world's coverage read, not of casualties. Material-conflict share fell against the after, 2024 and 2025 windows but not against the immediately preceding weeks. The independent fatality database (UCDP) now requires an access token and was not checked.
• Coverage of the tournament itself was not the driver. Articles matching "World Cup" were slightly more negative during the tournament than before it (referees, injuries, ticket pricing). The lift came from the rest of the news.
• Before and after are not clean controls. Both contain whatever else happened in the news; the same-window 2024 and 2025 comparisons are the cleaner tests, and they agree.
• GDELT skews toward English-language and wire-service sources. This is a property of the instrument.
• Cadence is a hypothesis. No study has tested whether an annual nations event sustains this effect; hedonic adaptation predicts it fades, the London 2012 wellbeing literature found the hosting effect gone within a year. The championship intends to publish this measurement after every season.

Every other page in the data room prices the championship as a property: purse, rights, weekends, franchise comparables. This page prices the reason anyone will watch. The audience evidence behind a nations format is on the dispatch and the nations play; the format that keeps every flag alive is the operations pack; the rulebook written for the broadcast is the No-Touch Standard. This page is the number underneath all of it — and the commitment to keep measuring.