Falcons beat the 49ers: the 29-point box score is missing a safety

I spent longer with the Atlanta Falcons–San Francisco 49ers box score than the game itself lasted, to do the one thing almost nobody does after a slate of games: add it up.

Michael Penix Jr. threw three touchdown passes. Three times seven is 21. Nick Folk made a 31-yard field goal. That is 24. The report says Atlanta led 26 before that kick, and finished with 29.

In the NFL there is exactly one legal scoring path to two points: a safety. No safety appears anywhere in the report's twenty-one information points — nobody was tackled in his own end zone, nobody was run down behind the goal line, no such sequence is described.

Two possibilities, only one true: either the report omitted a scoring play, or its numbers do not reconcile. Both lead to the same place. This box score needs cross-checking before anyone writes another word about it.

Context: an NFL game filed under the wrong label

The source record labels the piece “football.” Every proper noun in it — Penix, Brock Purdy, Christian McCaffrey, George Kittle, Kyle Pitts, Drake London, Kyle Juszczyk — belongs to American football. Soccer's instruments are useless here: xG, PPDA, release clauses, salary caps. The substitutes are EPA per play, success rate, and the hard cap.

San Francisco arrived unbeaten, lost for the first time, and sits at 6-1. Atlanta is 3-4. McCaffrey left with a thigh injury after most of the first half, in a receiver room already thin before kickoff. Penix went 25-of-34 for 213 yards, three touchdowns, no interceptions. Purdy went 27-of-40 for 288 yards, one touchdown, two interceptions. Atlanta took four turnovers and turned them into 13 points. A 23-yard touchdown was erased by replay review; the next snap was a fourth-down sack by Za'Darius Smith. Eddy Pineiro missed his fourth extra point of the season. A failed Atlanta punt handed San Francisco the ball at its own 47 with about four minutes left.

The report carries no byline, no sourcing, no citation to an official gamebook. Everything downstream starts from that gap.

Core: 45% of the scoring did not come from the offense

Thirteen of twenty-nine points — 44.8% of Atlanta's total arrived directly from turnovers. That is the number that decided the game, and the number least likely to repeat in professional football. An offense that manufactures 16 points from its own possessions across 60 minutes did not win the game; it was carried by a defense paying off a debt.

Penix's passing was efficient but vertically compressed: 6.26 yards per attempt, 8.52 per completion. The gap between those figures describes the whole design — short reads, quick throws, let the receiver run. Two of three touchdowns went to tight ends, the signature of multi-tight-end personnel, play-action and bootleg: a system that protects a young quarterback by never asking him to throw deep. He threw no interceptions, and that came from design, not luck.

San Francisco lost the opposite way. 7.2 yards per attempt, 288 yards, 75 more than Atlanta, and a nine-point defeat. They won the yardage battle and lost the scoreboard because four low-frequency events all broke against them: the erased touchdown, the fourth-down sack on the next snap, the fourth missed extra point of the season, and an interception with no time left to correct it.

The fourth-down decision chain was the real hinge. After the touchdown came off the board, San Francisco went for it and Purdy was sacked. A red-zone trip became zero points in a game lost by nine. No schematic adjustment in this game carried the weight of that one snap.

The conditional tree behind the missing two points. If a safety occurred, everything reconciles: 21 + 3 + 2 = 26, and the later score takes it to 29. If no safety occurred, then either the 26 or the 29 is wrong, and every inference about game flow tilts with it. Without the official gamebook I take the first branch, because it is the only scoring path that does not require a second, separate omission.

Four findings that do not reconcile. The report claims four turnovers; only three are described — an interception, a Kittle fumble recovered by Kendal Daniels, and at most a second interception. The fourth is never identified. The report also describes Purdy's “second interception” twice as if it were two events; read literally that is three picks, contradicting the stat line of two. And the personnel attributions need verification rather than assertion: Pineiro assigned to San Francisco, Folk to Atlanta, Jake Bailey as Atlanta's punter — three assignments inconsistent with the most commonly documented 2026 rosters. I am not saying who is right. I am saying nobody checked. A wrong name does not bring a sport down. It brings down trust in the person who wrote it.

Contrarian: the story that will be told

For the next 48 hours the story will be: rookie quarterback topples the unbeaten team. Tidy, with a protagonist, and wrong about causation.

Atlanta did not win by being better. They won by receiving four gifts and spending all four. That is a variance profile, and variance has no season. Any team that walks into next week intending to reproduce the 45% model will lose.

The other side is the mirror. San Francisco left with a process-positive, results-negative profile — the kind forecasting models like and standings punish. Six turnovers in the first six games, then four in one afternoon, is a statistical anomaly, not a trend. And one variable went missing from every bulletin: McCaffrey played only most of the first half. An offense that loses its primary running back mid-game does not lose 75 yards; it loses its rhythm. Purdy's 288 yards are more impressive than they look.

The biggest blind spot this week is not either team. It is the report. An account with no byline, no sourcing, missing a safety and contradicting itself on turnovers will be copied by dozens of outlets. The error will not stay in one article; it will seep into every analysis built on top of it.

Data is never scarce in football. What is scarce is the habit of asking where the data came from.

Takeaway: the gamebook is the NFL's VAR

Before 2026 I trusted memory. After 2026 I trust three checks. 2026 added a rule: never explain a law without the document in front of you. The NFL has its own VAR — the official play-by-play, every snap numbered. Until it lands, every conclusion about this game, mine included, carries a discount.

The task for next week is concrete: when the official box score is published, who will be the first to correct the number 26?