A while ago, I was wondering if there was any way to avoid losing my rating when suddenly it struck me: What if I used public code on ideone to mark a submission as a "cheat submission" to be flagged for cheating and have my rating change canceled? I decided to put it to the test, participated in Round 833 (Div. 2), answered a few questions, and then submitted a cheat solution, copied directly from a public ideone, for problem A. Loh and behold, my -200 rating was canceled, and I remained blue.
So what's the moral of the story? I don't know. Cheaters should have their rating changes removed, but should they have it removed even if the change is negative? Until CodeForces takes action and allows cheaters to have negative rating changes, we can effectively negate any negative delta and only climb up and up. (Of course, this relies on the Carrot extension correctly predicting delta changes during the contest)
Yeah, this is big brain time
I might be wrong, but I think this only applies for first time offenders
Assuming cheat detection works the way you say it does (i.e. submissions are skipped, rating changes are cancelled), one obvious solution is to assign all users flagged for cheating the rating change that they would have received had they placed in the bottom of the standings. I'm not sure whether it makes more sense to rank cheaters last and then calculate rating changes normally or to calculate rating changes with cheaters removed, then update cheaters' ratings based on the performance they would have had if they placed last, but in either case this seems like a reasonable way to ensure that there is never an incentive to cheat.
Post like these have emerged in the past, sometimes even with people admitting cheating for this same reason. Losing rank as well other measures, like, for example, being banned to participate in the next X contests if you get caught cheating should have been implemented long time ago. As of right now cheating has no real consequences.
The thing is that it might be a unlucky low probability coincidence. Hence it might be that everyone believe you cheated, but you actually haven't. But such a coincidence cannot happen twice with non-negl probability, so first offence is excused. That's the reasoning I have for this.
If I am not mistaken, on codechef you lose a lot of raiting if you cheated, even more than not solving any problem at all. I feel like codeforces should implement something similar, or some other punishment for cheaters so that people don't start cheating when they see negative delta prediction.
(However, they are only fooling themselves by having raiting far above their skill, so I don't think that it's a huge problem, unless people who participate fairly are affected)
it's not just that simple, the chance of a false positive is quite high. The way cf works in my knowledge is it accounts for multiple attempts of cheating and then finally bans the user after a few true incidents.
Indeed, these things are very common in the competition, and I hope this kind of behavior can be strictly prohibited.
i wonder if someone accidently(not cheated but coincidently matches with someones else solution) got a submission skipped not only he didn't got any positive delta plus will also get negative delta
This happens quite a bit, and that's why users can petition through a comment.