Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The Full Chess Cheating Report of Hans Neiman

chess.com

Post reply on HN
Print view View original

261–270 of 486 posts

Re: The Full Chess Cheating Report of Hans Neiman

#261

Post by computerdork »

berkut wrote: post #37
erdevs wrote: post #27

This report provides a detailed background of Hans' potential cheating, and detailed breakdowns of certain aspects of chess.com's cheat-detection methodology, including previously unknown (or little known) methods such as window focus change event monitoring and post-focus-change move analysis.[1] The report also reveals Niemann's engine move correlations alongside over two dozen chess Grandmasters who have admitted…

Does engine correlation actually prove anything though? Some of the 'statistical analysis' that has been posted on twitter regarding it in the last week has been against hundreds of engines, so 'engine correlation' seems to mean "the move made matched against at least one engine that would have made that move" I think?

Am not a statistician, but at least in an online analysis I saw, seems like correlation can effectively identify players who are playing too much like a computer. Because they don't just run correlations on Niemann, but on all the top players, and do comparisons (and for certain long stretches of tournaments, Niemann's is playing way, way above how anyone else has ever played).

This is video explains it pretty well, and seems like a very compelling argument (at least to me): https://www.youtube.com/watch?v=qjtbXxA8Fcc - and just know that the woman talking is a bit hard to understand because of her accent.

... oh, and to address your point about 100's of engines, my first thought was that are only a handful that everyone uses (Stockfish?) (and also, just guessing, but I get the impression that most top engines recommend similar moves, but again, just a guess!).

Re: The Full Chess Cheating Report of Hans Neiman

#262

Post by rippercushions »

rowanG077 wrote: post #254
kadoban wrote: post #180

Earlier quoted context omitted.

For shooters, bots or assists that only use vision and normal input (special mouse/keyboard) aren't hard to imagine and I'm positive already exist. I know specifically I read a while back about mice that do recoil correction to improve aim, and things like "click when something that looks like a head is in the crosshairs" would be a weekend project at most for a bored college student. Strategy games may be harder, bu…

I doubt that since a weekend project. You need to analyze a video stream of hundreds fps in real time with very low latency requirments. You'd probably need an FPGA to do that. Custom training an AI to only trigger on heads.

No, it's trivial these days, and this very task is one of the introductory tutorials to Google's Coral AI Tensorflow boards:

https://coral.ai/models/object-detection/

Which you can buy as a USB stick for $59:

https://coral.ai/products/accelerator/

Sure, it's a little more complex if you want to train on specific heads as opposed to a generic model, but not hugely so. The biggest challenge would likely be downsampling the video stream into something the model can process quickly enough.

Re: The Full Chess Cheating Report of Hans Neiman

#264

Post by xxs »

rowanG077 wrote: post #254
kadoban wrote: post #180

Earlier quoted context omitted.

For shooters, bots or assists that only use vision and normal input (special mouse/keyboard) aren't hard to imagine and I'm positive already exist. I know specifically I read a while back about mice that do recoil correction to improve aim, and things like "click when something that looks like a head is in the crosshairs" would be a weekend project at most for a bored college student. Strategy games may be harder, bu…

I doubt that since a weekend project. You need to analyze a video stream of hundreds fps in real time with very low latency requirments. You'd probably need an FPGA to do that. Custom training an AI to only trigger on heads.

You don't need 'hundreds' fps - you can do it w/ 10sec and get better reaction time than humans (reaction under 100ms in track events is consider foul). It should not be so difficult with a separate GPU and a capture card. Serial mouse/keyboard inputs are trivial as well.

Re: The Full Chess Cheating Report of Hans Neiman

#265

Post by rowanG077 »

rippercushions wrote: post #262
rowanG077 wrote: post #254

Earlier quoted context omitted.

I doubt that since a weekend project. You need to analyze a video stream of hundreds fps in real time with very low latency requirments. You'd probably need an FPGA to do that. Custom training an AI to only trigger on heads.

No, it's trivial these days, and this very task is one of the introductory tutorials to Google's Coral AI Tensorflow boards: https://coral.ai/models/object-detection/ Which you can buy as a USB stick for $59: https://coral.ai/products/accelerator/ Sure, it's a little more complex if you want to train on specific heads as opposed to a generic model, but not hugely so. The biggest challenge would likely be downsampling…

That's a generic AI. I would be very surprised you could get it to work accurately for games and hyper specific the heads of (possibly non-human)models. The box you see in the marketing material is completely useless for this use case. Besides how do you get the data to your standard second PC, scale the input the USB thing expects. Send it over USB and back, and then you have to send it to a custom mouse that receives the click command over bluetooth? I don't think 7ms just for the inference step is going to cut it.

A weekend project were you have to design an ultra low-latency software + hardware stack, a custom mouse and train a custom AI. I would be incredible if anyone manages this in a weekend.

Re: The Full Chess Cheating Report of Hans Neiman

#266

Post by lelanthran »

fsckboy wrote: post #84
Dalewyn wrote: post #75

Earlier quoted context omitted.

>The fact that online cheating is so widespread even among top chess players is certainly news to many, including me. Not to me. When the stakes of winning are so damn valuable, sometimes literally so, it would be patently stupid to not cheat and increase your odds of winning. Remember: Cheating is only a problem if you're caught. If you're never caught, cheating is completely legitimate. These aren't games of pleasu…

If your attitude about chess competition was applied toward you in all of your routine irl activities, you would be extremely unhappy. The cop does better in his job by meeting his quota of speeding tickets, so it objectively makes sense that he would issue you a speeding ticket every day with no evidence and over your strenuous objections... except you wouldn't object because the cop is working the system rationally…

He didn't say it's a good thing, nor that he approves, just that it is unsurprising that people (attempt to, and sometimes) cheat because cheating is a rational response.

It's just a different way of saying that when the measure of something becomes the target, then that measurement is gamed.

Would you also assume that Goodhart (https://en.wikipedia.org/wiki/Goodhart's_law) prefers cops who issue illegitimate tickets, or lawyers who pad their hours?

Re: The Full Chess Cheating Report of Hans Neiman

#268

Post by rowanG077 »

xxs wrote: post #264
rowanG077 wrote: post #254

Earlier quoted context omitted.

I doubt that since a weekend project. You need to analyze a video stream of hundreds fps in real time with very low latency requirments. You'd probably need an FPGA to do that. Custom training an AI to only trigger on heads.

You don't need 'hundreds' fps - you can do it w/ 10sec and get better reaction time than humans (reaction under 100ms in track events is consider foul). It should not be so difficult with a separate GPU and a capture card. Serial mouse/keyboard inputs are trivial as well.

How is Serial mouse/keyboard trivial? You'd need to make a custom mouse that receives data from your second PC. You'd probably want it to be wireless else you will have 2 cables coming out of it. And no just having a second fake mouse is not fair. That's easily detectable by software and would definitely raise a lot of red flags.

Re: The Full Chess Cheating Report of Hans Neiman

#269

Post by xxs »

maoeurk wrote: post #140
addicted wrote: post #127

Earlier quoted context omitted.

One of the stupidest phrases vert. Almost by definition the exception disproves the rule. There’s literally no logical sense in which the exception can prove the rule unless the rule is not really a rule but more of a suggestion.

Yeah it's pretty much exclusively used incorrectly. The original meaning was that an exception to a rule proves that such a rule exists. For example, "no parking on Sunday" is an exception that implies that parking on any day except Sunday is okay. https://en.wikipedia.org/wiki/Exception_that_proves_the_rule It honestly doesn't bother me as much as when people incorrectly say "begs the question" when they mean "raise…

One note about parking - normally all actions (incl. parking) are allowed, unless they are explicitly forbidden. Lack of any other rules/guidance implies the parking is allowed.

Re: The Full Chess Cheating Report of Hans Neiman

#270

Post by barrkel »

rowanG077 wrote: post #265
rippercushions wrote: post #262

Earlier quoted context omitted.

No, it's trivial these days, and this very task is one of the introductory tutorials to Google's Coral AI Tensorflow boards: https://coral.ai/models/object-detection/ Which you can buy as a USB stick for $59: https://coral.ai/products/accelerator/ Sure, it's a little more complex if you want to train on specific heads as opposed to a generic model, but not hugely so. The biggest challenge would likely be downsampling…

That's a generic AI. I would be very surprised you could get it to work accurately for games and hyper specific the heads of (possibly non-human)models. The box you see in the marketing material is completely useless for this use case. Besides how do you get the data to your standard second PC, scale the input the USB thing expects. Send it over USB and back, and then you have to send it to a custom mouse that receiv…

7ms isn't going to be the bar. Human reaction times are in the 150+ms range, you have 20x your budget.

I agree it's still tough though, most I/O paths on generic hardware will have buffers at every step that you'll have to fight to eliminate.

Post reply on HN