Please subscribe to the official Codeforces channel in Telegram via the link https://t.me/codeforces_official. ×

### l_WhiteEagle_l's blog

By l_WhiteEagle_l, history, 13 months ago, I have some reasons for that, if you have any doubts please comment below.

Firstly, you can't be sure whether your current solution is correct or not. I mean, how can you know if a solution with AC probability of 0.12345.... is going to pass? How to know if that's the intended solution? And even after getting AC you'd still be worried that your solution might get FST, and the probability was not the "intended" one.

Secondly, it does not measure the skill of an individual properly. For example, a middle rated participant (CM, Expert) would like to make the probability of AC as high as possible(to avoid FST), they might think that current probability is not high enough. So they spend more time looking for better one. While some low rated participant(pupil, specialist), wouldn't care about probability that much, because they don't have much rating to lose if it FSTs. They could write some solution which "somehow" gets AC, while that middle rated is still looking for better probability. Note, that I'm not talking about high rated people(Master, GM, etc...), I mentioned middle and low rated ones only.  Comments (3)
 » A problem where the probability of an expected solution passing is about $0.12345$ is definitely not a good one. But probability-based problems are not like that. The probabilities that the authors consider are much closer to $1$, most of the time the model solution has no more than $10^{-6}$ chance of failing; sometimes even better than that. For example, in yesterday's problem D, I am pretty sure that any reasonable implementation can process at least $100$ different prefixes of $s$ as the candidates for the answer. So, the probability of the model solution (and any its implementation) failing is about $2^{-100}$. Do we seriously need to consider this?