Investigating Cheating on Chess.com by Looking at Speedruns
Vložit
- čas přidán 6. 03. 2023
- UPDATE: I got some data on it, check out my followup video • 4% Cheaters?? Chat GPT...
What do you all think about the prevalence of cheating on chess.com? My curiosity has been piqued recently so I thought I'd explore it a little bit!
Definitely a different kind of video but I hope you enjoy it anyway. If you have any background in statistics, please leave a comment with your criticism of my take! Or what tests to perform to get better data and reach a conclusion!
Links I mentioned in the video:
@blazemarc8585 video about how he got away with cheating for months: • I cheated on Chess.com...
@g0atmoth video about Escape from Tarkov cheaters: • The Wiggle That Killed...
@DanielNaroditskyGM recently used speedrun account: www.chess.com/member/frankfur...
@chessbrah speedrun accounts: www.chess.com/member/kayenveebe and www.chess.com/member/sterkurs... - Hry
I'd guess that cheating is more rampant in Rapid vs the shorter time controls. I can't imagine trying to cheat during a bullet game!
Agreed! I made that same comment on another video about cheating, but it turns out, sadly, there is cheat software that just overlays the right move on the screen using arrows. So you're not relaying moves to another device, you're not switching between tabs, you just follow the arrows.
In bullet there're also cheaters: they use bots that make moves in a couple of seconds.
There are an insane amount of cheaters in bullet.
Selective engine consultation can't be caught. If a player only uses an engine for 2 or 3 critical moves how can he be detected? I have no idea how anyone could believe that only 0.2% cheat. It's b.s..
While I do agree accounts cheating is higher than .2%, I can’t verify that. There are a lot of things to consider. I doubt the number they used a lot of resources, but they have more than us.There is some variables. Things like alternate accounts exist, and they screw with numbers. In fact, I’m pretty sure the range is 0-1% of players cheating just the number of accounts is higher. Also good job on the games! Very interesting. I love watching you.
I agree there are some important factors to take into consideration and I'm glad to hear your thoughts on it. The point about alt accounts is interesting. So you're saying that cheaters are more likely to have alt accounts. Makes sense, they open new accounts whenever one is banned. Perhaps I should be more careful and talk about the percentage of accounts, not people. I'm not sure how much of an impact that has but it is a fair point!
About chesscom having more data than us, yes, but they also have a motive to underreport cheating since that hurts the value of the website. That's one thing that really solidified for me in watching the videos about cheating in Escape from Tarkov -- when you hear reactions to the knowledge about widespread cheating, so many people were saying they don't feel like playing anymore and were angry with the developer for not addressing the issue.
About the 0-1% range, I actually followed up a little bit on the data since apparently no statisticians watch my channel (or chose to comment). I used Chat GPT to analyze the data and it gave me a 95% confidence interval of 1.5% - 10% (roughly, I don't have the numbers). But even on the low end that's 10x the official rate. I think I'm going to work on another video with more data to narrow down that range.
I'm beyond sick of seeing people "Peak" in arena and tournaments and it's so obvious they're cheating. Lifetime accuracy is around 70% yet in tournaments they're all of sudden at 92+ in 50 move + games. It's insane, all you do is get a small msg and rating adjustment and some of the accounts dont' even get closed for good, they just get a warning or something.
I played at least three cheaters on my chess journey (I've been playing for 10 months already and have played about 300 games).
Used to often get points back from cheaters getting caught. But after getting above 1600 I have not had any points returned for months.
I think that probably more that 0.2% of people cheat but the majority of players are bellow 1500 and most cheaters gain rating and hence cheating is much more common at higher ratings.
Also unfortunately cheating is much more simple than using "cheating software." Most cheaters will just open the engine on a second tab / device and input the moves as the game goes on. And hence selective cheating is extremely hard to catch
Great points, yeah I've thought a little about where the bulk of cheaters are in the rating distribution and it's really hard to say. A pure cheater who maximizes ELO will get to 2500 right away and be banned pretty quickly. A selective cheater could really end up anywhere. The HurricaneAle account that played against me in this video has 239 rapid games, an overall losing record, and an ELO of around 1300. Blitz rating is only around 900.
You're right about not needing specialized software of course, but from what chesscom has said, those are the easiest cheaters to catch. They use alt-tabbing as a metric in evaluating whether a game had cheating. And if you open up the game on another device as an observer, I'm sure that's another huge red flag -- most games have no observers, and it should never be the same ip address. You could use a vpn on the 2nd device but now we're getting into special software.
Oh wow I didn't realise cheating was so sophisticated. I assumed people who couldn't play chess wouldn't be able to work out things like that. Its really unfortunate
When u make all those moves and all u win is a pawn u know they are cheating , cause at higher levels thats all the edge u need. At lower levels losing a queen is where most people resign
That is a pretty good point and I know exactly what you mean, the handful of times I have played against a confirmed cheater it's just like what you describe. It's not that they magically find a 5 move checkmate in the opening, it's that they do a bunch of confusing moves just to win a pawn, then they bludgeon you for the rest of the game with that single pawn advantage.
I have looked at some of your games and you have a tendency to resign in either equal or only slightly losing positions. Don't do that. At your level you should never resign.
Hey thanks for the feedback! I'm curious why you think resigning is bad? I've heard that advice many times, and I have experimented with not resigning losing positions and I feel like I don't really get anything out of it, but maybe I'm missing something.
Obviously its fair to resign if your down like 10+ points of material with no compensation but if you lose a piece its very possible to come back.
Also its an actual skill to be able to create winning chances in lost positions so it can actually help with your attacking chess (and in some cases help with your time management because time is also a piece)
@@jonbbbb It's not a waste of time to play till the end of our level. Firstly, blunders are not an uncommon occurrence. You can draw many lost games. By playing till the end, you can learn how to defend most resiliently and also how not to blunder a draw when you are winning, if you have been actively looking for it in lost positions.
@@jonbbbbwhen you blunder you don’t lose at the spot, you can still get back in the game. When you resign you lose and have no chance to even draw the game
@@jonbbbbI have 2300 blitz. I have saved to many games by not resigning.
less than 2 percent? oh no. i don't believe that. how many time have i played out the timer on players who are playing like geniuses but they can't think for themselves and cannot make good moves on their own. they lose on time. or they give them such a winning edge they feel like its safe to play on their own and end up losing at the end. i hate the games like this.
On my main account I have about 1.5k games and I have faced 1 confirmed cheater
I am currently researching where these engines are coming from. Not all are humans using free software. I suspect that the site developers use bots as an always ready opponent and to further the AI agenda.
It varies considerably across rating levels and time controls. Faster time controls are cleaner - not because cheating is impossible, but because many cheaters don't know about bots, and also because cheating in fast time controls is harder to hide even if you have a bot that displays the best move.
You said something about most games not getting checked - that's actually untrue. Automated systems are doing at least a basic scan of millions of games every day.
You also said something about cheaters being less frequent at higher levels because they get weeded out on the way up. In terms of sheer numbers this would be true, but in terms of percentages, the opposite is true. At lower levels, they're heavily diluted by vast hordes of clean players.
Another issue - many of those games were actually won by the speedrunner. Many cheaters play some games clean - and these are also the cheaters more likely to last longer and play more games. Simply counting games with a fair play icon will result in over-counting due to these games actually being clean. Of course, unbanned cheaters will have the opposite effect - and we have no way of knowing which effect is greater.
Another issue: you're seeing a lot of new accounts getting paired together. Is this random pairings or an intentional policy to help protect established clean players? If the latter, speedrun accounts that are all new accounts will have inflated numbers.
As for accounts closed for abuse that may have also been cheating - in general, there's no particular pattern as to one taking precedence. Whichever that happens to get picked up on first will get the closure.
it is easy possible If you want we can play, I will cheat in fast time control like 3+0 or even 1+2 and you wont we ll see how it will go