It's interesting, because it means that on average you'll do substantially better than your expected hit rate. For poisson processes a lot of the expectation of

It's interesting, because it means that on average you'll do substantially better than your expected hit rate. For poisson processes a lot of the expectation of the time taken "comes from" long tails. If you're rolling a fair dice, if you cap at 12 (i.e. the 12th roll will land a 6 no matter what) then your expected time to roll a 6 goes down from 6 to 5.4. So by applying a "correction" on strings of misses without applying a similar correction to a string of hits it skews the mean down which actually changes what your average hit-rate is. It would be interesting however if the tested hit-rate was adjusted lower to account for this effect, that would make it "fair" again by the hit-rate stat reporting the average hit-rate, but would make the randomness generate "appear more random" to humans.