| 모델 | Founder Success Prior Mean | Founder Success Prior Precision (value) | World Success Posterior | Value given Success (value) | 특징 | | ---------------------------------------------------------------------- | -------------------------- | --------------------------------------- | --------------------------- | ---------------------------- | --------------------------------------------------------- | | Gans 2023, 2025<br><br>\cite{chen2024programs} <br>(bandit literature) | △ | x (0) | x | x (V) | passive founder learning ground truth success probability | | \cite{archibald2002should } | . | . | △(objective) | X (1) | startup maximize success probability (variable) | | **M1** | o | x (0) | △ (prior becomes posterior) | X (1) | • 성공확률 = 상수<br>• 학습 없음<br>• 정태적 모델 | | **M2** | o | x (1) | △ (Posterior) | X (1) | • 성공확률 = 확률변수<br>• 베이지안 학습<br>• φ는 변하지만 가치 고정 | | **M2'** | o | o | △ (Posterior) | X (V) | • 가치변수 모두 내생화<br>• 완전 동적 모델<br>• τ 최적화 가능 | - o: Decision Variable - △: variable but not decision (state) - x: fixed, static, exogenous [[tom_griffith]]