diff --git a/docs/decisions.md b/docs/decisions.md index e070920a..6dc91432 100644 --- a/docs/decisions.md +++ b/docs/decisions.md @@ -825,7 +825,7 @@ Valid while Claude Code loads `~/.claude/rules/*.md` in every project, a path-sc **Defaults stay 5 bugs and 5 runs.** A sixth run per side cuts the spread of the mean of the bugs' cache-read changes from 16% to 13% for 10 more runs, about 35 minutes. -**A chunk-sized change doesn't rise above the noise.** PR #490 moved the medians by 5% to 9%. The spread at 5 × 6, the largest count measured, is 11% to 13%. It falls with the square root of the runs, so seeing a 9% change takes about 13 runs per side and bug: 130 runs, 8 hours and 70M cache-read tokens. No chunk waits that long. The bench shows a change of more than about 15% in tokens. Its All change is the mean of the bugs' own changes (item 514); the change of the pooled medians, which it printed before, needed more than 31% in output tokens and 50% in cache-read tokens. What to try is filed as items 515 (plant bugs Haiku fails at the base, so the pass rate moves) and 516 (measure the spread of the reads and the turns, which are counts, not token sums). +**A chunk-sized change doesn't rise above the noise.** PR #490 moved the medians by 5% to 9%. The spread at 5 × 6, the largest count measured, is 11% to 13%. It falls with the square root of the runs, so seeing a 9% change takes about 13 runs per side and bug: 130 runs, 8 hours and 70M cache-read tokens. No chunk waits that long. The bench shows a change of more than about 15% in tokens, since its All change is the mean of the bugs' own changes (item 514). Before that, the change of the pooled medians needed more than 31% in output tokens and 50% in cache-read tokens. What to try is filed as items 515 (plant bugs Haiku fails at the base, so the pass rate moves) and 516 (measure the spread of the reads and the turns, which are counts, not token sums). **Pilot first.** A new setup, or a project the bench hasn't run on, runs 2 bugs × 2 runs before a full bench (lilleman, 2026-11-21). A pilot takes about 40 minutes. The first pilot of this calibration passed no run, because the API refused a token that held a line break. It showed that within 15 minutes, where a full bench would have planted for 25 minutes first and printed the same. @@ -842,7 +842,7 @@ Measured on `fejkdata` (Go, 5 bugs planted in the files `larv-review places` nam | 5 × 6 | 28 / 11 | 43 / 13 | 19 / 8.5 | | 3 × 6 | 28 / 14 | 40 / 17 | 16 / 12 | - The All rows pool every run, so their median lands in whichever bug's runs sit in the middle of the pooled costs, and a bug's cost differs from another's by up to 8×. In tokens, the mean of the bugs' changes has a third to half of the pooled change's spread. The bench keeps the pooled All rows per side, which cost nothing, and prints the mean as the All change. In wall time it gains only from 5 × 5. The 86 at 5 × 3 wasn't traced. For the mean of the bugs' token changes, 5 × 4 and 3 × 6 give the same spread (14 and 14, 18 and 17) for 40 and 36 runs. + The All rows pool every run, so their median lands in whichever bug's runs sit in the middle of the pooled costs, and a bug's cost differs from another's by up to 8×. In tokens, the mean of the bugs' changes has a third to half of the pooled change's spread. In wall time, the mean gains only from 5 × 5. The 86 at 5 × 3 wasn't traced. For the mean of the bugs' token changes, 5 × 4 and 3 × 6 give the same spread (14 and 14, 18 and 17) for 40 and 36 runs. The bench still prints the pooled All rows per side, and the mean as the All change. - **One finished chunk**: `fejkdata`'s PR #490, base `33b7fa80`, head `5b9c2a14`, which the comprehension panel ruled better on Navigation, Locality and Shape and the same on Self-sufficiency. The bench planted 5 bugs and ran 38 of 50 runs before the machine restarted. The first 3 runs per bug and side, 30 runs, are the ones counted, and all passed. The mean of the bugs' changes was +4.6% in output tokens, +8.9% in cache-read tokens and +3.8% in turns, with the bugs' own changes in cache-read tokens from -41% to +47%. The two token figures sit inside the spread at 5 × 3, 17% and 22%; the spread of the turns wasn't measured. The mean of the wall times' changes, +105%, lies above the 86% of that row, from one bug whose head runs took 6.8 times as long at the median. Nobody read its streams to find out why. The head cost slightly more tokens where the panel finds it easier to read, and the wall time can't be judged, so the bench neither confirmed nor contradicted the panel. - **Cost**: 5 bugs × 6 runs took 3.9 hours and 34M cache-read, 1.2M cache-write and 0.3M output tokens; a run takes about 3.5 minutes, and planting 16 to 25 minutes. A 5 × 5 bench is about 3.3 hours and 28M cache-read tokens. Gremlins ran out of 8 GB at the base of PR #490, so the help's example takes 16 GB. diff --git a/todo.md b/todo.md index f1d78ffe..7db19de2 100644 --- a/todo.md +++ b/todo.md @@ -124,7 +124,7 @@ Touches: **Record in a decision the 95th-percentile spread, at 5 bugs × 5 runs, of the reads before the first open of the bug's file and of the turns, beside the tokens'.** -Item 503 found no affordable count that lifts a chunk-sized change above the tokens' noise (`docs/decisions.md` → `task-bench: a bench takes 5 bugs and 5 runs per side, after a 2 × 2 pilot on a new setup, and at no affordable count does a chunk-sized change rise above the noise`). Reads and turns are counts, not token sums, so they may be steadier. Run `task-bench` with the head against itself (`--base` the head's own commit) on a Go project, 5 bugs × 5 runs after a 2 × 2 pilot. Split each bug's runs at random into two sides, as the decision describes, and record each figure's 95th-percentile spread, for the All change and for the mean of the bugs' changes. Say in the decision whether either count can see a change the tokens can't. It needs a Claude token, given as `task-bench --help` says, and about 4 hours of wall time. +Item 503 found no affordable count that lifts a chunk-sized change above the tokens' noise (`docs/decisions.md` → `task-bench: a bench takes 5 bugs and 5 runs per side, after a 2 × 2 pilot on a new setup, and at no affordable count does a chunk-sized change rise above the noise`). Reads and turns are counts, not token sums, so they may be steadier. Run `task-bench` with the head against itself (`--base` the head's own commit) on a Go project, 5 bugs × 5 runs after a 2 × 2 pilot. Split each bug's runs at random into two sides, as the decision describes, and record each figure's 95th-percentile spread, for the change of the pooled medians, which the bench printed before item 514, and for the mean of the bugs' own changes, which it prints now. Say in the decision whether either count can see a change the tokens can't. It needs a Claude token, given as `task-bench --help` says, and about 4 hours of wall time. Touches: