But what should someone do when the problem statement does not match with sample I/O & finally one has to get AC by sending a solution for which sample I/O doesn't match (Problem C for the yesterday's contest, surely it did not have a multiple corrector judge program). How can a programmer assume which Sample I/O is correct & which must be corrected by himself?
Yes, problem C was really annoying. I finished my program after 7 minutes, and then tried to figure out what I did wrong. After convincing myself, that my output was correct I submitted, ignoring the fact that I couldn't reproduce the sample output and got Accepted.
But also in problem B it was not good explained what to do in case of multiple solutions; I think this is the reason why nobody got Accepted.
And problem E didn't define what kind of alimentary chains are allowed - if one animal can appear more than once or not.
From someone of my university I heard that he found out with asserts that limits for problems D and F were wrong in the judge input file.
So only problem without mistake seems to be problem G.