Brier = (p − outcome)², 0 is perfect, 0.25 is a coin flip on everything. Every resolved entry counts; void entries are listed and excluded; a superseded original is excluded inside its correction window and shown struck through. Nothing is removed.
Where an entry publishes its reasoning, any later change in my view is a new dated record beside it: which factor changed, from what to what, why, with a source, and the new result. Nothing is edited. The original number and its date remain the graded forecast. The reasoning can be corrected while its entry is still a draft; from the day the entry is logged, it is append-only, like the entry.
A number I sealed before readers forecast is published, with its salt, when forecasting closes. If I can no longer prove one, the entry is marked void: commitment unverifiable, scored as a Brier of 1.0, and counted in the mean. Void commitments so far: 0.
| Entries | Graded | Open | True | False | Void | Brier mean |
|---|---|---|---|---|---|---|
| 22 | 3 | 19 | 2 | 1 | 0 | 0.3083 |
| Bucket | n | Mean p | Observed | Brier |
|---|---|---|---|---|
| 0.0–0.2 | 0 | — | — | — |
| 0.2–0.4 | 1 | 0.25 | 1.00 | 0.562 |
| 0.4–0.6 | 2 | 0.48 | 0.50 | 0.181 |
| 0.6–0.8 | 0 | — | — | — |
| 0.8–1.0 | 0 | — | — | — |
| Sector | Resolved | Open | Brier |
|---|---|---|---|
| Space | 2 | 12 | 0.383 |
| Defense | 0 | 2 | — |
| Energy | 1 | 1 | 0.160 |
| Cross-sector | 0 | 4 | — |