Recent and related:
Lichess: The free and open source chess server - https://news.ycombinator.com/item?id=32045763 - July 2022 (65 comments)
I’ve seen a video of Magnus Carlsen playing on Lichess before. Does Irwin ever accidentally flag him or people like him? Do these sorts of folks have to be verified in some fashion?
Unless they've made significant improvements, this thing is borderline useless.
About all it does is act as a nice PR/gimmick. Magic thinking.
When last I checked about 5 years ago, they were including client-based stats in the analysis (time per move, focus, and timing between clicks to pick a piece up and drop a piece).
The issue with doing that is the client is fungible, you can set any state you want locally; and so by manipulating these stats you could skew the input going into these models.
To give people a layman's short overview of the model:
It uses a several CNN Pooling Layers for feature detection and LSTM (attention embeddings) in a siamese network architecture.
CNN models notoriously fail to detect features that that are larger than its kernel size.
The LSTM embeddings can only be as useful as the features that it is trying to detect.
There were a number of problems with the model at the time, a high false positive rate, overfitting, and too much weight was being given to client-controllable input.
When the client-controllable input was held stable (constant). Certain book openings would be skewed with additional weight towards a false positive, and earlier book moves activated more often than lesser seen positions.
I brought the issues up to ornicar years ago in an issue, but they closed the issue without comment. The posts are mostly gone now.
There were a number of disagreements (mostly about certain people with admin privileges abusing their authority, not sure if it was ornicar or one of their flunkies; either way not important just completely unprofessional).
I guess it was good enough for them despite the high false positive rate (driving account churn).
Related to this, there's a really good talk by the founder of lichess that includes an overview of the cheating problem, and the techniques they use to detect and manage it.
The relevant section of the video on YouTube is: https://www.youtube.com/watch?v=LZgyVadkgmI&t=1080s
If the system suspects a cheater, maybe it can just match them up against a GM (or beyond GM) bot to confirm?
This is really cool. Rampant cheating is the reason I completely abandoned playing chess, so nice to see something happening.
Now we have a computer effectively accusing people of fraud (if I understand correctly). How does Irwin address the obvious risks and dangers?
I love chess. At my absolute best, I had a over-the-board USCF rating of 2156. I never made it to master no matter how hard I tried. So, ticked off, I gave up serious play. Years go by and while I'm no where near as good these days, I started enjoying playing again. And, then I start running into people online who were clearly using engines instead of their own skills and wits. Now, I only pay with friends and over-the-board. If this can help wipe out the cheats, I may finally go back to playing online again.
I don’t understand what someone gets out of winning chess games by cheating. It would actually make me feel worse about myself.
how can you cheat on chess?
just hire Kamsky
undefined
undefined
I see a lot of comments about the general topic, but none on the actual tool, which appears to have no (public-facing) updates in about three years. Presumably it could be a tool that incrementally improves over time with the explosion of data lichess has, but surely there would be a single commit since 2019. Do we know that this is what's actually running on lichess's backend today?