In the last Div. 1 + 2 round, there were quite a lot of hacks on problem B during the contest, as you can see it here. However, it seems that many of the hack tests were not added to the system tests. For example, all of my 4 hacks are distinct: 1 2 3 4, each containing 1000 test cases, but you can see only the first one was actually added to the tests as test 18. There were 10 pretests and 23 main tests (before uphack), while there were seemingly ~30 distinct hackers, so it's also likely that many of the hackers didn't even get any of their hacks added to the main tests at all.

Although most of the solutions that failed system test got "Wrong answer on test 11", there were still ~300 solutions that managed to survive on it and failed on my hack test. I can't still be sure if the ones that passed the whole system test, would've survived even if all of the hacks had been added to the main tests.

There was also this blog that claimed that their hack wasn't added to the system test on a former round, but no action has been done to it yet. I thought it was a simple mistake by the coordinators, but now I start to think that it's not like that — whether a hack test will be added to the main tests is just arbitraily chosen and they consider it as a normal decision.

I can't find a rule that specifies that all hack tests will be added to the tests, but it has always felt like that we were granted that we can add our test to the main tests by successfully hacking, so it feels weird to see cases like this. It feels also unfair and random that some hacks can luckily make other solutions fail while others unluckily can't just because it's not chosen by the contest managers.

The only reason not to add all hacks to the main tests is that it may slow down system testing, but there was a case that applied a workaround to handle this: https://codeforces.net/contest/1985/hacks?verdictName=CHALLENGE_SUCCESSFUL&chosenProblemIndex=F where we had exceedingly many distinct tests to add all of them separately, so instead they bundled many hacks into a single test with individual test case. Couldn't the same thing have been done in this case too? Literally all wa hacks could be integrated into one test.