> You're banking on there being a significant shift in technology/knowledge that occurs right between the time you are diagnosed and the time you have left to treat it. Just as a Bayesian probability, it's extremely unlikely to be the case.

The probability that I think you're talking about here is not actually relevant, given that the trial is already underway. Basically, the "potential advance" event has already happened. (Similarly, the probability of getting shot by an arrow is not relevant once the arrow's already poking out of your body.)

Given the trial is already underway, the probability that matters is the fraction of treatment candidates at the phase you are considering that make it all the way through; for phase 3 trials, this is 58-65% [0], which is actually pretty good. Of course, rates for earlier phases are much worse because there's attrition at each phase.

0: https://patentpc.com/blog/clinical-trial-success-rates-how-m...