Each optimization hypothesis models rank improvement as a Bernoulli trial initialized with prior strength \(N_0\) where wins increment \(\alpha\) and non-improving tests increment \(\beta\), shifting the expected win probability \(\mu = \frac{\alpha}{\alpha + \beta}\) as empirical evidence accumulates.