HOT RADAR
llmHEAT 16.7°

AI CURATED HOT STORY · 11/23/2026

Harvard Study Finds Top Model Outage Barely Moved Developer Merge Rates

1 reports archived1 independent sourcesupdated 11/23/2026, 00:51:03
Synthesis & Latest Updates
1 Sources Cross-Validated

A Harvard paper exploits the 19-day suspension of Claude Fable 5 under a U.S. export control directive in June 2026, comparing 297 affected users with 8,908 developers who stayed on Opus 4.8. Access to the stronger model raised the 7-day pull request merge rate by only 1.4 points, within noise, and five other model upgrades showed no merge-rate gains either, with 94.5% of PRs already merged beforehand.

LATEST/A Harvard paper exploits the 19-day suspension of Claude Fable 5 under a U.S. export control directive in June 2026, comparing 297 affected users with 8,908 developers who stayed on Opus 4.8. Access to the stronger model raised the 7-day pull request merge rate by only 1.4 points, within noise, and five other model upgrades showed no merge-rate gains either, with 94.5% of PRs already merged beforehand.

Hourly heat curve

2 fully observed hours
Fewer than 3 fully observed hours; trend not plotted yet.

Coverage timeline

Total 1 reports · Latest first
  1. Rohan Paul (@rohanpaul_ai)T2·72 pts
    • During the 19-day Fable 5 suspension, users on weaker Opus 4.8 saw merge rates differ by just 1.4 points, within noise
    • The study compares 297 affected users against 8,908 unaffected developers using GitHub pull request data
    • Five other model upgrades also failed to lift merge rates, and a 94.5% baseline merge rate suggests a strong ceiling effect