Methodology
How calls, signals, and scoring work
Obvious Later exists to let people make timestamped calls before they see the community consensus. Signals are the underlying evidence layer. The methodology below applies to both.
Principles
- Clear resolution criteria
- Every call defines in advance what evidence resolves it YES or NO. Ambiguous questions are not published.
- Answer before consensus
- Users make their call before the community consensus is shown. This is enforced in the interface, not requested by norm.
- Timestamped predictions
- Calls are recorded with the exact time they were made. Records are not editable after submission.
- Honest conviction
- Probabilities should reflect genuine belief. A 60% answer means the caller would take fair odds around 6-to-4.
- Resolved with defined evidence
- Outcomes are resolved against the criteria set at publication, using pre-agreed sources.
- Accuracy needs time
- A meaningful record requires many resolved calls across topics. Early scores are directional at best.
- Contrarianism is not virtue
- Disagreeing with consensus is only valuable when it turns out to be right. Being contrarian is not itself a signal of skill.
- Calibration matters
- A good forecaster is both accurate and well calibrated: 70% claims should come true roughly 70% of the time.
What this product is not
Informational only
Obvious Later is informational. It is not investment, legal, medical, or financial advice.
No wagering
Predictions are personal calls of conviction. No money is at stake. The product is not a prediction market.
No financial return
No monetary reward is attached to any prediction, regardless of outcome.
Forecasts are uncertain
All forecasts may be wrong. Track records are personal and provisional until resolved.
Forecasts are personal and may be wrong. A user's Foresight Score is a preview until enough of their calls have resolved to be statistically meaningful.