Codeforces is filled with great people, but it is also has plenty of users that abuse the system, and dampen other people's fun though their actions. A few problems I put under the spotlight
A unrateds, alts, and rating-manipulating contestants that win contests below their division, stealing the victory and some rating from legitimate participants. See the last 4-5 div 2's for sufficient examples of experts and specialists losing wins to these fakes. This can be extended to cheaters in general, but there is already a system for that in place.
B Users who make alts to shield themselves from backlash when commenting, potentially to make hurtful statements. Many of us have seen the abuse that they spew.
I've labeleld things as A and B to refer to the problem I am talking about.
I've done my research:
A Make div 2 have some trusted participant system like div 3 does.
A Construct a neural network that analyzes style (easier said than done)
B Ban unrateds from commenting.
A&B The current MikeModeration, which works to some extent.
My suggestion, inspired by the stackoverflow model, is community based moderation (I'm not sure if this has been brought up before):
My suggested solution:
A We implement a sort of community based flagging. Perhaps, upon noticing a suspicious account, a user with good standing (by either rating or contribution [that would be a good way to put that stat to use] or otherwise), can be able to flag a particular participant. Then the community votes based on whether they think the participant is legitimate or not.
B We can have a similar community moderation system for users who seem to have accounts solely to make abusive comments. (We can tell by looking at their participation, submissions and contest history).
A community system confers many benefits.
-Most community members are honest coders who want to compete, learn, win, and have fun. They will be rather impartial when casting votes, and thorough in detecting cheaters. There are plenty of people who make blogs dedicated to catching cheaters!
-The collective intelligence of hundreds of votes is significantly more intelligent than any automated banning or flagging system, as the watchful eye catches what the bot cannot. Ultimately, it will make illegal actions require too much effort to evade detection.
A Most cheaters tend to be very obvious. We, as users, can rapidly notice them through their handles, or their rating graph (goes from master to newbie), or an unrated finishing a contest in 30 minutes. Bots and systems can be tricked, but users are a bit tougher to hoodwink. If we see a known GM style, we can tell fairly quickly. Yes, I am very aware that there are talented unrateds. I am also very sure that at least 99% (made the number up but we know its correct) of people who participate in their first contest WILL NOT, and STRONGLY WILL NOT solve the entire contest like it is a breeze. False positives will not be plenty (we can always give benefit of the doubt), but the hoards of obvious abusers will be cut down. It will be worth it to give honest div 2 members a true shot at actually winning the division.
B A user can tell what a true account is and what a "made to spout venom" account is. We wouldn't want a systeem to flag Harbour.Space for example, and most rational users would know that this is not an abuse account! Similarly, a user who tries cheating the system by making random submissions for problems to appear like a normal member will be caught by the vigilant eye of the community. In fact, it takes more effort than its worth to hide ill-intent from a watchful community!
So, what do you guys think?