Skip to main content
AI23 July 2026

Testing and documentation

5 tasks. July 23, 2026 closed at 19.2x weighted leverage across 168.0 human-equivalent hours in 525 minutes of wall-clock time. Supervisory leverage came in at 336.0x.

That is 4.2 weeks of human-equivalent throughput in 8.8 hours. The ceiling was 28.8x; the floor was 10.1x. 5 of the 5 entries came from a single project.

Task Log

#TaskHuman Est.ClaudeSup.Factor
1Feedback batch 1: in-flow header vertical tag + DS 0.2.4/0.2.5 (Brand.subtitle + Button nowrap) + engine math study-plan variety + calibration skip discount + unit descriptions (snapshot v15) + unicode-math…24.0h50m5m28.8x
27-lane Opus orchestration (17-item feedback batch): session auto-open+Convoy+math wiring, lesson viewer+math rendering+sim modal, dashboard/course-detail 10-defect sweep, sim chrome gating, autopilot date…80.0h170m10m28.2x
3Zero-floor evening: silent no-op regen defect root-caused+fixed (hub item_refs nesting + worker guard), scenario_rubric_fields generator, criticals spec-sync fix both sides, 9 repair chains ->…28.0h105m4m16.0x
4Design and frontend20.0h105m6m11.4x
5Content burn-downs to 0 HIGH: CHEM1B (17H→0, 88M→58M, promoted), Algebra II anaphoric stems (promoted), AP Macro+Micro full chains (promoted) + embedder-seam and regen-kind-inference platform fixes16.0h95m5m10.1x

Aggregate Statistics

MetricValue
Total tasks5
Total human-equivalent hours168.0
Total Claude minutes525
Total supervisory minutes30
Total tokens2,925,000
Weighted average leverage factor19.2x
Weighted average supervisory leverage factor336.0x
Human-equivalent weeks4.2

Analysis

The highest factor of the day came in at 28.8x and the lowest at 10.1x, a spread of 2.9 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 24.0 of the 168.0 human-equivalent hours, or 14 percent of the day. No single task dominated the total, so the weighted average is representative.

Supervisory time was 30 minutes against 525 minutes of execution, a ratio of about 1 to 18. Supervisory leverage of 336.0x 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.