Hack Protection LLC

Revision en2, by V--o_o--V, 2018-11-12 21:20:34

Remember http://mirror.codeforces.com/contest/1039/problem/B or http://mirror.codeforces.com/contest/843/problem/B and the hack fiesta with reproducing the pseudorandom sequence?

It might be the case that some of you don't want to be seed hacked. There is a post by neal about the issue. The post is well written but I strongly disagree with the solution described there.

"This should be different every run." I'm fairly sure that it indeed is (well there is still is a small probability of collision but this is kinda obvious). Because of that you are no longer able to conveniently debug you program.

You can hide that under an IFDEF and this fixes the aforementioned problem. However hiding the code under the IFDEF takes 3 lines of code and you may accidentally change something in the IFDEFed part and never notice that locally (the latter is somewhat made up). And I consider IFDEFs ugly but it is a personal thing.

Another way is to obfuscate the generation of the seed. It is definitely against the rules though. There are some grey area ones like hashing a string of tabs and spaces.

There is another way.

template <size_t S>
constexpr bool ls(const char a[S], const char b[S]) {
    for (int i = 0; i < S; ++i) {
        if (a[i] != b[i]) {
            return a[i] < b[i];
        }
    }

    return false;
}

static_assert(ls<8>(__TIME__, "18:50:00") || ls<8>("20:15:00, __TIME__)", "((");

I don't think this requires any further explanation. It does not compile between 18:50 (should be shortly after the submission) and 20:15 (should be the time you expect the contest to end). This was tried on the Educational Codeforces Round 54.

Initially the hacking attempt received something along the lines of "Invalid Verdict". Later the solution and the hacking attempt were ignored (we were deemed trolls or smth like that). However, this is somewhere in the grey area. It might be considered an attempt to destabilize the testing system but I assume that the testing system should be fine.

It is not known how will similar submissions be treated in a standard contest. While I do think that hacks are cancer and would never hack myself I understand that in almost every problem the hack is beneficial to the hacked person. Therefore I am not willing to try this code in a Div1 contest unless there is a problem with a randomized solution. However, someone might and I would be curious to see the results.

Irrelevant Fact

The actual reason for my preference of this approach to any other is because it seems like a middle finger one to a person who tries to hack the seed but without resorting to swearing. Feels good man.

History

 
 
 
 
Revisions
 
 
  Rev. Lang. By When Δ Comment
en2 English V--o_o--V 2018-11-12 21:20:34 26
en1 English V--o_o--V 2018-11-12 21:14:13 2745 Initial revision for English translation
ru1 Russian V--o_o--V 2018-11-12 21:13:29 2745 Первая редакция (опубликовано)