FIDE President address on anti-cheating policies

[Wayne Komer]

FIDE President address on anti-cheating policies

August 19, 2020

Dear chess friends,

I believe there is a consensus that computer-assisted cheating is a real plague of contemporary chess.

We have already taken strong steps to enhance our efficiency in fighting it, including strengthening analytical tools, using detectors and scanners in all official FIDE events, training arbiters, finding a right legal basis, and having a dedicated team working on these matters.

The online chess boom brought new challenges, and although the number of suspicious cases is fairly low, FIDE must act vigorously, sending a clear message to potential violators in order to create a secure environment in our competitions.

We work together with the leading online chess platforms. We have adjusted the algorithms used for online play. Having a lot of data, we sharpened our statistical methods – and in these regards, I’d like to thank Professor Ken Reagan, who keeps improving his algorithm – and those who think his method does not work against the so-called smart cheaters, they will be surprised.

We must act, and I want to emphasize that FIDE will be ready for the ensuing legal challenges.

However, I feel that we need a broad consensus on the measures applied. Below are the main questions we would like to have your opinion on:

1. Our methods of detection, although very advanced and ever-improving, can't provide a 100% confirmation. In many cases, the probability estimated is higher than the one for DNA tests. Do you believe a statistical algorithm (or a combination of those) giving close to 100% probability of cheating could stand as sufficient grounds for banning a player? If yes - what odds would you find sufficient?

2. Shall FIDE apply sanctions for alleged online violations to over-the-board-play (and vice versa)?

3. Shall we apply sanctions for alleged violations at platforms’ own events, and other unofficial online events, to official FIDE online events (and vice versa)?

4. Shall we publish the names of alleged violators after the very first conviction?

5. Shall the violators be punished retroactively, with their prize money, rating and titles been revoked for some period preceding the verdict? And, if yes, how far back should these actions go?

6. What would you consider a reasonable banning period for first-time violators, and for repeat offenders? How strict should be the measures in youth competitions?

There are many questions and some of them are related to the moral and legal aspects of the subject. Having a fair and transparent system will require a trusted framework. The worst thing to do would be to ban an innocent player.

Likewise, the reputation of chess and our global chess family could suffer tremendous damage if a tsunami of scandals and court procedures starts to overshadow the exciting environment of international chess competitions.

We must be strict, but responsible. Firm, but accountable. And before approving a general policy, we would like to hear your opinions. You may answer the questions raised in this communication or simply submit your proposals to the following email: anticheating@fide.com

It is going to be a long battle, but I am sure we will succeed.

https://www.fide.com/news/703

[Brad Thomson]

How is it possible to detect a cheater who uses the computer program just once a game at most, and only in key positions where it finds a tactical shot?

[Aris Marghetis]

How is it possible to detect a cheater who uses the computer program just once a game at most, and only in key positions where it finds a tactical shot?

That's part of the challenge man.

[Matt Nicholson]

About seven years ago I was recruited by schemingmind.com to help work on a cheat detection system. We didn't even get past the first stage and I'm convinced sites that ban regularly (like chess.com) are only catching the blatant cheaters.

The first stage of the trial was to run a fairly basic engine detection algorithm (checking for top 3 matches on 3 different engines and generating an 'accuracy ratio') on a set of 100 games. Half the games were correspondence matches from before engines, and the other half were modern correspondence games where engine use was allowed.

The experiment couldn't find a single 'cheater'.

[Kerry Liles]

How is it possible to detect a cheater who uses the computer program just once a game at most, and only in key positions where it finds a tactical shot?

When I was a young kid I often wondered how the police would investigate and find someone who committed exactly *one* robbery (say) of a store where they carefully disguised themselves and picked a random store in a neighbourhood where they had no connection... It seemed to me that the majority of "solved" crimes were aided by the accumulation of many bits of information and data which would be missing from this 'restrained criminal'. The list of suspects would be very large and there would be basically no connection between the perp and the crime.

I don't see any clear way to reliably catch cheaters - hackers and people who are motivated enough will find gaps in almost any system - for reference, see the current internet.
Some of Ken Reagan's work to provide a high probability that someone is cheating (by making "too many" moves that are just unlikely to be found by humans*) is simply an attempt to mitigate the problem.

(* - a highly simplified description)

Cheaters will find a way and that will be discovered and blocked and everyone moves on. At this point cheating detection must be very heuristic.

[Erik Malmsten]

A player could have a board beside the computer and play through variations. Or an opened opening or endgame book. Or somebody with a computer just outside the room who could signal when there is a sudden exclaim move. But somebody who cheats doesn't really want to play chess.

Have to have an Arbiter in every room.

[Kerry Liles]

A player could have a board beside the computer and play through variations. Or an opened opening or endgame book. Or somebody with a computer just outside the room who could signal when there is a sudden exclaim move. But somebody who cheats doesn't really want to play chess.

Have to have an Arbiter in every room.

As long as the Arbiter can be trusted etc. What could go wrong? :)

[Fred McKim]

Cheating is all about risk vs reward. If the reward (ie prize fund) is kept low, then the risk becomes not worth it. When the reward becomes higher (ie an increased prize fund) then the risk has to be increased by having cameras and screen sharing, etc.

[Pargat Perrer]

Cheating is all about risk vs reward. If the reward (ie prize fund) is kept low, then the risk becomes not worth it. When the reward becomes higher (ie an increased prize fund) then the risk has to be increased by having cameras and screen sharing, etc.

There is another thing to think about in the issue of cheating: the age and talent of the person who is supposed to be cheating.

A teenage junior who is improving rabidly might just by chance get good enough to start making GM-level moves in a particular game and trigger anti-cheating software to identify him or her as a cheater in that game. If the junior objects and says s/he was just playing good moves, are you going to deny that it is possible?

Whereas if Bob Armstrong suddenly starts playing GM-level moves throughout a whole game …. lol sorry Bob, I saw your post about your ratings so just had to use you as an example! Hope you don't mind. :)

The one thing you don't want to do with anti-cheating is overapply it, because it is like saying to everyone in a tournament: Okay, here is where we EXPECT you to finish in the standings, and if you don't do that and have a surprising good tournament, we are going to take you to a back room and find out the truth....

What this means to me is that cheating is not really a disease, it is a symptom. The real disease is that chess has come to the point where cheating is even possible and as Fred says can be rewarding in some circumstances. This means chess has a disease, but the disease is not the cheating itself.

Remember the "good old days" when games were adjourned and players would be up all night analyzing? That has gone away and it is a clue about the disease. The disease is that chess is precisely calculable now by any of thousands of engines. This means chess as a human contest is no longer the same. To keep it going, you have to have now all these anti-cheating measures so that no one is playing above their real abilities. It makes it all.... meaningless.

Nakamura and Carlsen just played a best of 7 series of rapid / blitz matches and it went down to the 7th match and Armageddon game decided it. What does it all mean? Nothing really.