Debate over METR capability acceleration: full-data fit yields 1.6x

Toby Ord and @austinc3301 debated how to fit METR's frontier capability trend, with Ord initially replicating a 1.4x speedup but agreeing on 1.6x when all 16 model snapshots are used; sensitivity analysis shows estimates ranging from 0.99x to 1.90x depending on method.

2026-11-10 ~ 2026-11-10 · 3 related posts

• Refitting all of METR's data yields ~1.6x AI capability acceleration, above Toby Ord's 1.4x — austinc3301 · 2026-11-10
• Toby Ord revisits METR capability trends: full-data fit yields 1.6x acceleration, not 1.4x — tobyordoxford · 2026-11-10
• METR capability acceleration estimates range 0.99x–1.90x depending on methodology — tobyordoxford · 2026-11-10