9 tasks. June 3, 2026 closed at 40.9x weighted leverage across 617.5 human-equivalent hours in 905 minutes of wall-clock time. Supervisory leverage came in at 661.6x.
That is 15.4 weeks of human-equivalent throughput in 15.1 hours. The ceiling was 160.0x; the floor was 5.0x. 8 of the 9 entries came from a single project.
Task Log
| # | Task | Human Est. | Claude | Sup. | Factor |
|---|---|---|---|---|---|
| 1 | Design and frontend | 240.0h | 90m | 10m | 160.0x |
| 2 | Coding | 80.0h | 75m | 4m | 64.0x |
| 3 | Harden 8 CIP [ip] drafts (Z/AA/BB/CC/DD/EE/FF/GG) per review: soften bitwise/RAG/[engine subsystem]/guarantee overclaims to calibrated tolerances + answer-support verifier; fix stem-type count; worked-example… | 12.0h | 12m | 4m | 60.0x |
| 4 | Infrastructure | 40.0h | 42m | 4m | 57.1x |
| 5 | Documentation | 7.0h | 10m | 5m | 42.0x |
| 6 | Full a11y audit + fix of all 10 accelastudy.ai sites (main + 9 sisters): built axe harness, diagnosed FOUC/scroll-reveal/dark-mode/redirect/gate-timing false positives, fixed contrast tokens (light+dark),… | 100.0h | 200m | 3m | 30.0x |
| 7 | Audit and review | 120.0h | 300m | 20m | 24.0x |
| 8 | Implement 12 unimplemented [ip] claims to reduction-to-practice (BB-9/12/18/19/26, DD-17/24/25/26, Z-7/17/18) with tests+endpoints+config; prove 0% exam-bug fix with 7-test regression; App Z cleared to 0… | 18.0h | 170m | 3m | 6.4x |
| 9 | Coding | 0.5h | 6m | 3m | 5.0x |
Aggregate Statistics
| Metric | Value |
|---|---|
| Total tasks | 9 |
| Total human-equivalent hours | 617.5 |
| Total Claude minutes | 905 |
| Total supervisory minutes | 56 |
| Total tokens | 12,118,000 |
| Weighted average leverage factor | 40.9x |
| Weighted average supervisory leverage factor | 661.6x |
| Human-equivalent weeks | 15.4 |
Analysis
The highest factor of the day came in at 160.0x and the lowest at 5.0x, a spread of 32.0 times between the two. That is a wide range for a single day, and it usually means the day mixed mechanical work with work that needed real judgement.
The largest single entry accounted for 240.0 of the 617.5 human-equivalent hours, or 39 percent of the day. No single task dominated the total, so the weighted average is representative.
Supervisory time was 56 minutes against 905 minutes of execution, a ratio of about 1 to 16. Supervisory leverage of 661.6x 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.