If he indeed found a way how to use grader's mistake then disqualification is a very unfair action of jury. It is 100% judges fault that grader allowed exploiting it in such way and punishing the contestant in this situation doesn't make any sense.

If I found a way to exploit grader during the contest, it's highly probable that (if I had enough time of course) I would also spend few minutes on sending this into testing system and writing judges a message describing exploit in case of success.

The emotional text above is legit only in case if everything was how pllk described. Can anybody provide some extra information about this case?

Yeah, my teammate peto159 talked with them (I think...) and he told me the same.

We came up with another idea that might "hack" those interactive tasks (none of us tried it, of course — but maybe some ISC guy reads this comment and tries/fixes (if necessary) it for next year):

IOI 2015 is the first IOI where multiple threads are allowed. So it is possible to fork and solve the problem in the child process with a suboptimal amount of queries — then return the result to the parent process, which submits the answer with 0 queries. I think that should work, because the child process has seperate variables (and basically seperate anything), so the parent process has no chance to know about those queries, right?

Here's a quick proof of concept (I even added a lot of comments ^^): here. The codes printes the following on my mac:

child used 10 calls
parent used 0 calls
result = 45

Would something like that have worked?

UPD: Well, the more I think about it: Forking isn't the same as a thread (haven't really realized that at the time of writing the comment.. probably because I don't know too much about this kind of stuff ^^) — so maybe it's not even possible to call fork. Does someone know more?. I'm kind of curious :)

I agree that the actions taken in this case (to the extent that I understand it) by the IC and ISC were reasonable and right.

There are however a number of other practices for which I think rules should be made clearer.

At present (or at least back when I was a contestant) there are several ways in which one may use the submissions to gain information about the test data. One possibility for example is to use "assert" statements. My understanding is that this practice has even revealed bad test data during contests in the past yet it is in my mind not exactly in the spirit of the contest.

Another way instead of using asserts would be if a contestant tried to use "TLE" as a way of gaining information about the input (by making the program terminate or go into an infinite loop depending on some property of the input). It is unclear to me whether the rules at present would give grounds for disqualifying the contestant who tried to use such submissions to gain information about test data "in good faith" (in the sense of each submission really trying to extract the information instead of just trying to slow down the grader) but still ended up making 80 submissions of such type.

(To be specific — suppose there is a task with only three ints as input. I can use 96 submissions of the above variety to in effect learn all the inputs. I then use an inferior algorithm to actually compute the outputs and finally submit a program with all these answers hardcoded for 100 points.This entire process can be fully automated. Cheating or not?)

Thank you for the answer. From your words it becomes much clearer why one can tell that he deserves such punishment. Let me clarify my words you quoted above.

Suppose that I found a way how to hack a contest environment, or grader, or compilation process, or whatever, to get score in an unfair way. First of all, I think that in this situation I must tell judges about that. Otherwise it can simply be that someone else finds such an exploit and uses it for his own benefit and jury can possibly miss such case (after all, jury are also people with about the same skills and abilities as the most skilled contestants). As a contestant I want such situation not to happen because this can affect my place in a final scoreboard.

That's why I would decide to tell judges about this exploit. After that I have two options. I may want to be sure that my exploit works by sending it in testing system and then write judges a message like "Here, look on my 100 points! They are unfair, please fix the hole in your grader and discard my score on this problem, so I can write a fair solution". On the other hand, I may just write "There is a potential exploit in this place, ...".

What I want to say is that there are reasons to choose first way rather then second. First, it may appear that the wrong person reads my message and he is not able to understand what I'm talking about. It may be that he tries an exploit by himself in wrong way and it doesn't work for him. It may even be that he decides to check it after the end of the contest because there is a lot of more important work to do now (and after that he can forget about it or find out that 10 contestants successfully used an exploit). That's why preparing a certificate of a working exploit is a reasonable action in such situation.

I understand that this violates the rules you quoted, but I'm curious what would happen if the situation was like I describe above. I'm interested mostly because there are lots of stories available on the Internet about people who find exploits and security issues on large popular sites like github.com, facebook.com, vk.com (our largest social network), some banks sites and many others. There also usually raises a question if such person should use an exploit as a certificate of that he tells the true. The interesting fact is that the behavior of different companies may be completely different. If you do not provide a certificate, you may even not be heard because you'll stuck in a support team that doesn't even understand any single word you are telling. If you decide to provide such certificate (by posting something on the web-site or creating $5 on your personal account) you may be rewarded for your help, as well as you may face threats of suing against you because you hacked their site.

Also as a programmer who is interested in how such exploits work I feel a bit upset that somebody was punished on a programming competition for showing extraordinary skills in just another area of programming. That's... well, that's not a reason for those guy to be justified, but he can do cool things, it's a pity that everything ended up in this way (even though it was intentional as you said several times).

Such an action by the contestant goes both against the letter of the rules (where it is clearly stated that the type of actions is forbidden) and against the spirit of the rules (as it is clear that the contestant is not attempting to solve the given algorithmic problem).

I'm not talking about this particular case, but IMO submitting solutions which don't attempt to solve algorithmic problem generally doesn't violate the spirit of the rules. E.g., if someone manages to fit into TL an overly optimized brute-force C++ code instead of implementing complex algorithms, is it really bad?

IMO situations when a participant uses knowledge about low-level stuff instead coming up with high-level algorithmic solution only add fun to programming competitions.