Track record
Other validators publish user counts. We publish outcomes. Everything on this page is computed from the pool, kills included. Nothing is curated.
Survival scores across the judged pool, 0 to 100. A flattering scorer piles everything on the right. This pool bunches in the middle and dies on the left.
Zones follow the verdict bands: under 40 is kill territory, 40 to 69 wounded, 70 and up survives.
Two judges scoring the same idea far apart is signal, not noise: it usually marks an assumption one model bought and the other refused.
The panel now records the score gap between judges on every run. The current pool predates that instrumentation, so there are no recorded gaps to show.
The standard test for a scorer that flatters everything is a control run: feed it deliberately absurd ideas and historically successful ones, then check it can tell them apart. A validator that gives everything 75 out of 100 fails that test in one pass. Skeptral publishes its controls on this page.
No control runs have been scored yet, so there is nothing to show in this section.
Put your idea through the same panel
Validate your idea