Trust ledger

The engine should be judged too.

Reasoning, confidence, predictions, challenges, and outcomes make the engine inspectable instead of just persuasive.

Trust rule

A report is only useful if it can change a decision. The trust surfaces below make IdeaForge more falsifiable, less flattering, and harder to hide behind generic AI prose.

Public predictions

Opt-in analyses can publish falsifiable predictions so engine confidence can be compared with crowd judgment.

Idea graveyard

Outcome follow-ups at 30, 60, and 90 days create a record of what actually happened after the report.

Anti-portfolio

A page for misses, false negatives, and lessons from ideas the engine judged incorrectly.

Leaderboard

Opt-in calibration rankings let founders compare judgment quality, not just idea quality.

Reasoning

Every meaningful score needs an explanation the founder can inspect.

Calibration

Predictions and outcomes matter more than confident-sounding language.

Challenge

The product should invite better context and a sharper second pass.

See Calibration Metrics Read Methodology