Hello everyone! This may look like yet another unnecessary blog from a low rated user. Some may even feel it is true when they reach the end of the blog. But I'm trying to address a major issue poisoning Competitive Programming and Codeforces contests, that is cheating.
Some may advice people to not be bothered with cheating and cheaters are losers and will head nowhere in life. I totally agree to most of this argument. Who am I? I'm a purple coder from my main doing competitive programming for 2 years now. Of my 20 years of life, this is the first thing I have genuinely enjoyed doing and put efforts at.
Yes, cheating does bother me. And how hard people try to ignore it, it does bother them even if it is minimal. As long as cheating affects them, it should bother them. I'm saying this from a position where I only do CP for joy and not rating. But positive deltas give me even more joy. There is no fun in playing a game when you don't win. Again, this is from my personal viewpoint which I believe others may share as well. If rating is a metric to measure somebody's skill, it needs to be protected.
I may only be affected from a small fraction of cheaters. But this number may be large for other people who are trying to find joy in this sport. I have plenty of time. So, I will try to make cheating public if I can't stop it.
Every week. I will post a blog about cheaters with proofs and evidences from this account which you guys will send me through messages. I won't be missing any week unless I died. There are lots of cheaters who simply get away without being caught. With this account, I will put all the cheaters to public. Of course, I will validate your evidences myself before posting about them. If Mike wishes, may look at the cheaters I post about and ban them.
I would love to know your opinions if this may help reduce cheating in any way.
If you make a high quality content about high rated cheaters it might increase awareness about cheating among high rated participants. With that being said, I highly doubt that it can reduce cheating. I once wrote this blog, which completely exposed a cheater harsh__h, got hundreds upvotes and nothing happened. No reaction neither from the CF team, nor from the contest coordinators.
In the last Codeforces Round 1008 (Div. 2), the authors of contest even congratulated a cheater Manasvi with top 4, despite numerous comments and even a separate blog that exposed him being a cheater.
Despite Mike's September blog about how CF is trying to be up to date with the AI cheating to keep integrity of competition, I am very disappointed with the state of things we are in right now, when they basically don't do a single thing. Even such outright dumb cheaters like Beserker69 are not banned and have no skipped submissions after the Antiplag check.
MikeMirzayanov
MikeMirzayanov
MikeMirzayanov
ok
Well we have already seen the effects the scary AIs have on actual contests: almost nothing. In the last div. 2 contest, I think that people were saying F was solvable by chatGPT, but not even $$$150$$$ people correctly got it. So that puts an upper bound on the number of people who are willing to cheat on hard problems.
My dear friend, while I do agree with you that the cheating problem might be somewhat exaggerated due to numerous blogs about cheating, I still believe that it is too serious to ignore.
Here are my points, my personal stance on the situation.
Free LLMs can more or less consistenly yield 1800-2100 performance when used without human assistance in reasoning, but with assistance with feedback (CE and which error, WA, TL, etc). I derived this conclusion from my observations of cheaters.
Most cheaters need to secure the blue color only for their successful placements. (I do deliberately put the phrase this way to avoid downvotes lol)
I also believe that most of cheaters are not dumb enough to have their rating graph taking off from green/gray to purple/yellow in couple of contests, or to be in the top 10 participants of div2/div3, while there are such cases.
The last div2F has 2248 rating at clist, which can be indeed too hard for most free LLMs with reasoning. While to get blue quick enough you need to solve ABCD with, let's say, 1800-1900ish average performance.
Average cheater performance falls in the range [1800, 1900], in my opinion. However, I may be wrong, and most of them have such performance range not to hide and be below radars, but because LLMs they use are not smart enough (yet).
TLDR. There is a huge cheater effect at blue performance range.
Perhaps you are right, there are a lot of cheaters who don't get super high ranks. But if getting blue is really that important to Indians, then there will always be a market for cheating. Like whatever Mike Mirzayanov or the competitive programming community does to combat cheating, it's gonna be destroyed by the top cheaters, because there's just more money in cheating than in preventing it. Even if we get super sophisticated plagiarism checks, there's gonna be some guy that gets paid to just quickly edit cheaters' code before they submit so they don't get caught. There's nothing we can do.
Indians should start as cheaters and code their way out.
lol
Here I agree. When I started CP last summer, there were hundreds (or even a couple of thousand) cheaters who submitted leaked solutions from Telegram groups.
However, it is clear that today there are far more tools (mostly LLMs) for cheating than there were just a year, or even several months ago.
I haven't conducted thorough research on this (sadly, though, as I'm genuinely curious about a proportion of cheaters), but based on my subjective judgement of the top 200 participants of the last div2s/div3s, there are at least 100% more cheaters than there were last summer.
Cheating can not be eliminated for sure. However, if the issue is left unaddressed, I expect we will soon have more cheaters than legit contestants in div2/div3/div4 (take a look at the userbase by country :) ). Let's all pray that CF rating will be no longer relevant for placements there.
So is it mean we need to be crazy strong to reach CM?
If you are going to post about cheaters with proof and evidence, I think you should do so with your purple account; it will be more effective.
Rating correlates but never was a metric to measure the skill.
MikeMirzayanov Do something Please!!!! The cp community will die this way if no action is taken. All the high rated people started someday where there was minimal cheating and they had motivation to move forward. But today even people who deserve to be specialist expert pupil whatever are not able to attain that due to your teams negligence of not taking down the cheaters or taking an appropriate action. As Christine- has stated in their blog about the cheater harsh, still no action has been taken. We just need this that only genuine people are been taken into the consideration of rank list after the contests.
If You cannot put the cheaters down, atleast see our blogs where people are posting about cheaters with their id's and proof. We are making your work easier but even we are ignored. I had posted a blog about a Youtuber who streams during contest's with 200+ live viewers but no action was taken even with that. I had tagged Mike thinking he will atleast see the blog someday. I deleted that blog because i felt it was all useless work of typing the whole blog, adding proof's etc showing genuine cheating had taken place.
Similarly with the user manasvi, many blogs were posted against that user but all went in vain.
Please do something it's our humble request please sir
Cheaters are ruining competitive programming by taking unfair ratings and disrespecting honest coders.
They may gain points, but they don’t improve their skills. In the end, they are only fooling themselves. Competitive programming is about learning and hard work, not shortcuts. While cheating is frustrating, exposing people publicly can be risky because mistakes can happen. A better way is to support stronger anti-cheating rules and report cheaters properly. Real success comes from effort, not cheating.
Cheaters might win for now, but when real skills are needed, they will fail. Honest coders should focus on improving and enjoying the competition.
Have you considered not smurfing first? Doesn't smurfing have the same effect as cheating — "taking the rating away" from other participants?
I mean, he has only given one Div. 3 and solved 2 questions there. I would not call that smurfing
it's like everyone and their mother has a second account