This page sets each round-one bar beside what apps of the same age in the archive actually did over a nominal 42-night window: cohort members from the call night (±1) to the decisive night, the app’s own row from its dated signed baseline to the decisive run. It states facts only: for each call, the group of tracked apps that were the same age on the call date, the spread of their rating-count gains, the bar a 90th-percentile rule would have set beside the pre-registered bar, and the call’s own observed count. Rating counts are public App Store figures, not downloads, revenue or users. Compiled 2026-09-30 (JST) from the nightly archive.
Method (one paragraph). For each call, the cohort is every tracked app whose age on the call date falls in the same age bucket the ledger scores by (0–30, 31–90, 91–180, 181–365, 366+ days). Each cohort member’s start count is its snapshot on the call night or the adjacent night (whichever is closest; the archive’s first full night is 2026-08-10), a night’s count is the highest observation that night established (the rule the published verdict counts follow); its end count is the call’s decisive night — the night of the run that judged the call (09-22 for the 08-11 calls, 09-24 for the 08-13 calls); the gain is the difference. The app’s own row starts from its signed baseline — the snapshot the bar was computed from — while cohort members start from the call night ±1. Relative gains are computed only for members with a start count ≥ 5 (a 1→2 gain is 100% and would swamp the distribution), and separately for the “same size band” — members whose start count lies within 0.5×–2× of the call’s baseline, because rating growth is strongly base-dependent. The “90th-percentile rule” bar is baseline × (1 + the 90th-percentile relative gain); it is shown for the whole cohort and for the size band (flagged when the band holds fewer than 30 apps). The v0 bars are the pre-registered ones (D+21: max(2×, +50); D+42: 3×). Each window is nominally 42 nights: cohort members run from the call night (±1) to the decisive night, and the app’s own row runs from its dated signed baseline to the decisive run; the exact member spans are stated under each section. Rating counts are public App Store figures; nothing here is downloads, revenue, or users.
Heartio: Heart Rate Awareness (call 2026-08-11, baseline 5, age at call 8 d → cohort 0–30 days)
| measure | n | 50th | 75th | 90th |
|---|
| absolute gain, whole cohort | 154 | 0 | 1 | 8 |
| relative gain, whole cohort (baseline ≥ 5) | 17 | 67% | 154% | 392% |
| relative gain, same size band (baseline 3–10) | 10 | 115% | 210% | 394% |
| bar | value |
|---|
| v0 D+21 early bar (max(2×, +50)) | 55 |
| v0 D+42 bar (3×) | 15 |
| 90th-percentile rule, whole cohort (baseline × (1 + p90)) | 25 |
| 90th-percentile rule, same size band (n < 30 — thin, shown for the record) | 25 |
| this app, observed | 5 → 6 (+1, 20%) from the signed baseline (snapshot 2026-08-10T18:00:04Z) to the decisive run 2026-09-22T18:00Z |
Cohort window: start night 2026-08-11 ±1 → end night 2026-09-22; member spans 41–43 nights (median 42). The v0 bar is judged on the 2026-09-22T18:00Z run.
Normal: Screen Time Control (call 2026-08-11, baseline 24, age at call 87 d → cohort 31–90 days)
| measure | n | 50th | 75th | 90th |
|---|
| absolute gain, whole cohort | 458 | 0 | 1 | 8 |
| relative gain, whole cohort (baseline ≥ 5) | 84 | 33% | 89% | 180% |
| relative gain, same size band (baseline 12–48) | 35 | 46% | 116% | 189% |
| bar | value |
|---|
| v0 D+21 early bar (max(2×, +50)) | 74 |
| v0 D+42 bar (3×) | 72 |
| 90th-percentile rule, whole cohort (baseline × (1 + p90)) | 67 |
| 90th-percentile rule, same size band | 69 |
| this app, observed | 24 → 36 (+12, 50%) from the signed baseline (snapshot 2026-08-10T18:00:04Z) to the decisive run 2026-09-22T18:00Z |
Cohort window: start night 2026-08-11 ±1 → end night 2026-09-22; member spans 41–43 nights (median 42). The v0 bar is judged on the 2026-09-22T18:00Z run.
Sentence: Screen Time Control (call 2026-08-11, baseline 8, age at call 165 d → cohort 91–180 days)
| measure | n | 50th | 75th | 90th |
|---|
| absolute gain, whole cohort | 678 | 0 | 1 | 11.3 |
| relative gain, whole cohort (baseline ≥ 5) | 220 | 11% | 36% | 69% |
| relative gain, same size band (baseline 4–16) | 102 | 7% | 33% | 70% |
| bar | value |
|---|
| v0 D+21 early bar (max(2×, +50)) | 58 |
| v0 D+42 bar (3×) | 24 |
| 90th-percentile rule, whole cohort (baseline × (1 + p90)) | 14 |
| 90th-percentile rule, same size band | 14 |
| this app, observed | 8 → 8 (+0, 0%) from the signed baseline (snapshot 2026-08-10T18:00:04Z) to the decisive run 2026-09-22T18:00Z |
Cohort window: start night 2026-08-11 ±1 → end night 2026-09-22; member spans 41–43 nights (median 42). The v0 bar is judged on the 2026-09-22T18:00Z run.
Ziggy: Learn ASL Sign Language (call 2026-08-13, baseline 832, age at call 23 d → cohort 0–30 days)
| measure | n | 50th | 75th | 90th |
|---|
| absolute gain, whole cohort | 154 | 0 | 1 | 6 |
| relative gain, whole cohort (baseline ≥ 5) | 15 | 49% | 130% | 300% |
| relative gain, same size band (baseline 416–1,664) | 2 | 77% | 109% | 128% |
| bar | value |
|---|
| v0 D+21 early bar (max(2×, +50)) | 1,664 |
| v0 D+42 bar (3×) | 2,496 |
| 90th-percentile rule, whole cohort (baseline × (1 + p90)) | 3,328 |
| 90th-percentile rule, same size band (n < 30 — thin, shown for the record) | 1,899 |
| this app, observed | 832 → 2,005 (+1,173, 141%) from the signed baseline (snapshot 2026-08-12T18:00:01Z) to the decisive run 2026-09-24T18:00Z |
Cohort window: start night 2026-08-13 ±1 → end night 2026-09-24; member spans 41–42 nights (median 42). The v0 bar is judged on the 2026-09-24T18:00Z run.
| measure | n | 50th | 75th | 90th |
|---|
| absolute gain, whole cohort | 913 | 0 | 2 | 24 |
| relative gain, whole cohort (baseline ≥ 5) | 456 | 6% | 19% | 42% |
| relative gain, same size band (baseline 857–3,428) | 31 | 10% | 29% | 39% |
| bar | value |
|---|
| v0 D+21 early bar (max(2×, +50)) | 3,428 |
| v0 D+42 bar (3×) | 5,142 |
| 90th-percentile rule, whole cohort (baseline × (1 + p90)) | 2,430 |
| 90th-percentile rule, same size band | 2,386 |
| this app, observed | 1,714 → 3,106 (+1,392, 81%) from the signed baseline (snapshot 2026-08-12T18:00:01Z) to the decisive run 2026-09-24T18:00Z |
Cohort window: start night 2026-08-13 ±1 → end night 2026-09-24; member spans 41–42 nights (median 42). The v0 bar is judged on the 2026-09-24T18:00Z run.
CurveSpotter