Skip to main content
AI31 May 2026

Mostly testing

4 tasks. May 31, 2026 closed at 14.9x weighted leverage across 90.0 human-equivalent hours in 362 minutes of wall-clock time. Supervisory leverage came in at 450.0x.

That is 2.2 weeks of human-equivalent throughput in 6.0 hours. The ceiling was 16.0x; the floor was 8.0x. 3 of the 4 entries came from a single project.

Task Log

#TaskHuman Est.ClaudeSup.Factor
1Accessibility audit remediation — all 71 findings (5 blocker/30 serious/24 moderate/12 minor) fixed across web/iOS/Android/Electron + design-system; 4 waves; per-client build/test-gated commits80.0h300m8m16.0x
2Promoted 13 alpha packages to beta (Tier A+B): reweighted 12 goal_weights, generated 621 CCSP recall questions + 17 CISM pairs, verified all 27 beta packages pristine (zero findings); produced AP Precalc +…5.0h25m2m12.0x
3Brought AP Precalculus to pristine + beta: generated 3,359 questions from 0 (fixed a fresh-bank write bug mid-run), reweighted goal_weights (5 criticals), +1 pair, re-stamped, verified all 28 beta packages…3.0h22m1m8.2x
4Coding2.0h15m1m8.0x

Aggregate Statistics

MetricValue
Total tasks4
Total human-equivalent hours90.0
Total Claude minutes362
Total supervisory minutes12
Total tokens810,000
Weighted average leverage factor14.9x
Weighted average supervisory leverage factor450.0x
Human-equivalent weeks2.2

Analysis

The highest factor of the day came in at 16.0x and the lowest at 8.0x, a spread of 2.0 times between the two. That is a narrow range, which tends to happen when a day stays inside one kind of work.

The largest single entry accounted for 80.0 of the 90.0 human-equivalent hours, or 89 percent of the day. One task carrying that much of the total is worth noting; the average is doing less work than it appears to.

Supervisory time was 12 minutes against 362 minutes of execution, a ratio of about 1 to 30. Supervisory leverage of 450.0x is the figure I find most honest, because it measures the hours I actually spent rather than the hours a machine spent on my behalf.

Every figure here is recorded at the time the work is done rather than reconstructed afterwards. The human estimate is my own judgement and carries the uncertainty that implies; the minutes and tokens are measured. The full dataset, including this day, is available for download.