Colonel Stok: Do you play chess?

Harry Palmer: Yes, but I prefer a game with a better chance of cheating.

Michael Caine played Harry Palmer in the movie Funeral in Berlin. This was released in 1966—long before computers could play terrific chess. Perhaps he would have a different answer today?

Today I thought we might look at chess cheating in a way that complements what Ken does.

Of course Ken refers to Ken Regan. He is one of the world experts at detecting chess cheating. Detection is based solely on the statistics of move choice. The only data given to Ken about the games are the moves that were played and the overall time allowance. But this ignores the players and equipment on the scene. People can cheat in ways that are closer to issues in computer security and protocols.

Caine knows a lot about the latter. He is a doyen of heist movies and those increasingly involve security. In the 1966 comedy Gambit it is as simple as working around an alarm, but the 1969 version of The Italian Job has a switch of computer data reels and jamming traffic cameras. His latest movie Tenet centers on an algorithm for inverting time and entropy on Earth. He is also a fan of chess. He tangles with his co-star Laurence Olivier amid chess sets in the movie Sleuth. Most of all, his character in 2009’s Harry Brown is a chess player, who discourses on the 17th (not 7th as said) game of the 1972 championship between Bobby Fischer and Boris Spassky.

Cheating at Chess—the Easy Way

Derren Brown is an illusionist—a magician. He claims that he is a weak chess player. But he had Britain’s Channel 4 broadcast him playing nine strong players, including two grandmasters. Yet he won the match 5-4.

This is how he did it. He used an ancient trick. Say he plays two games: one against Alice and one against Bob. He plays black against Alice and white against Bob. After he gets Alice’s first move he plays that exact move against Bob. Then after he gets Bob’s move he plays that one against Alice. And so on.

Suppose he loses both games. Thus Alice wins and Bob wins. But that means that Bob’s answer to Alices’ move was a winner and so on. This is impossible and so he must win at least one of the games. Thus he wins one game unless both end in a tie.

The illusion is, how could he win five games against nine players given two were grandmasters? Brown played one of the nine games for real—he won that one. The “table trick” only works for an even number of games. It is not, of course, a new idea:

Alekhine and his twice world championship challenger Bogoljubov were once challenged separately by a relative patzer to games of correspondence chess at money odds. In effect, of course, the anonymous opportunist was playing in neither game. The story goes that the two players, who were friends away from the board, met up one day and latched on to what was happening.

The imitation trick is a real potential issue in Basque chess, where two players play two games simultaneously, one as white and one as black. By copying each other’s moves, they would always tie. An early description noted the issue but Ken has not been able to find how the rules forbid it.

A similar situation happened recently in a real tournament—a world championship qualifier, no less. Two games at adjacent tables played almost twenty of the same opening moves. The chief arbiter—someone Ken corresponds with several times a week—moved one of the games to a different area. International Master Danny Rensch, who heads the major online playing site Chess.com, made a video “Are You Copying Me?” of the incident. None of the four players involved was cheating, but this illustrates the kind of people dynamics one needs to watch for.

Cheating at Chess—the Too Easy Way

This is to cheat by consulting a computer program that is stronger than all human players, such as the free program Stockfish, without anything impeding one’s ability to access the program’s recommendations during the game. This is often the case in online chess without sophisticated measures to detect the access.

Ken’s statistical model can still judge the moves, but this is after the fact. We would like to prevent cheating. This hasn’t happened. Already last year, Ken was quoted in the UK Guardian newspaper saying, “The pandemic has brought me as much work in a single day as I have had in a year previously.” A Wall Street Journal story that featured Ken also noted:

The data showed something curious. More people were playing chess. Yet the fair play violations were surging even faster than the number of overall games. “Which makes us think that there has been an uptick in the rate of cheating,” said Gerard Le-Marechal, head of cheat detection for Chess.com.

This year has brought no letup—see Ken’s statement prefacing a non-chess post that June and July were the worst. Chess.com and Lichess and other playing sites have the final say but Ken is often used for both early warning (his “screening” step is agile and gives officials an informative snapshot of an entire tournament) and for explaining verdicts afterwards, since his model is transparent and not compromised by divulging explanations.

But again, this is after the fact. I recall a post we wrote about ways computer security is like “closing the barn door after the horse has already left.” The open question that I find interesting is not how cheaters can be detected, but is there some way to make it hard for them to cheat at all.

ARPANET history source

Of course, this is not just about chess, or other games forced online by the pandemic. It extends to administering courses and tests, at a time when the prospect of a “normal” in-person Fall semester is being roiled by the surge we’ve previewed and tracked on this blog.

Cheating at Chess—the Harder Way

The number of ways people have cheated at in-person chess is legion. Wikipedia has a long list. Ken put the ways in cases he’d encountered to a Dr. Seuss rhyme midway through his 2014 TEDx Buffalo talk.

On March 15, 2020, the New York Times published an article on tech in chess cheating. It drew analogy to the Houston Astros scandal, including the same example we just covered of whether José Altuve was wired for his series-winning home run in 2019. It touched on online chess and quotes Le-Marechal but showed no inkling of the impending pandemic and its effect on chess. Its first sentence about chess alludes to the 1978 incident in which Viktor Korchnoi alleged that Anatoly Karpov could receive coded information about their match games via the flavor of yogurt delivered to the table.

I, Ken writing this part, have been part of discussions of how a yogurt spoon dropped audibly could be one of myriad possible signals from the audience. The pandemic caused this year’s Tata Steel tournament to be played without audience, while some other elite events are played in an “aquarium” with one-way glass. But that does not work for larger-scale Open tournaments. Jamming RF signals is generally illegal. I agree with those recommending that an illusionist like Brown—someone with an eagle eye for watching people—be employed to help the arbiters at large events.

Yet for all the ways and means out there, it is still hard to cheat at in-person chess. Its state is one that organizers of online chess would gladly reach if they could. FIDE has promoted a hybrid form in which players travel to regional rooms watched by arbiters, but this is hard to manage on large scale. Dick and I have debated all year what to do for online chess, and we’ve converged on two poles of answers.

Way #1: Standardized Playing Tabletops

The paradox, noted this week by International Master Nisha Mohota in her recent video on cheating, is that the popularity of chess online has burgeoned during the pandemic. But this also enhances the following dilemma:

• Having a second camera—side view supplementing screen view—has been an effective measure.
• But requiring even one camera has been an acknowledged obstacle to expanding the reach of chess tournaments.
• And it takes extra human resources to monitor two video feeds per player.

Online education has also recognized the importance of a second camera, requiring one in some cases. Yet allowing the user to control how the side camera is positioned may allow circumventions, and mandating one connotes distrust and negativity in a bare sense.

My suggestion is to try to turn around the negative aspect into a positive by marketing a standardized and hopefully-inexpensive “Online Tabletop Arena.” It would have three walls to feel like an alcove. The walls, one with a side camera built in, would limit hand movements as well as sight lines. Standardization would lessen stigma and help monitoring. It could also be used for online test taking.

Way #2: Give In

The real import of our mentioned security post is to stop trying to stick thumbs in all the dam holes. Mohota in her video laments kids being exposed to chess programs and advocates training without them, but that strikes us as trying to close ten thousand barn doors while a million free horses are out there.

So let’s give in: Allow the players to use computers freely. The more, the merrier. But as in the 1967 Caine-as-Harry-Palmer movie Billion Dollar Brain, we reward the humans for how they disobey the computer calling the shots.

One way to implement this would be to have the chess playing site appoint one unknown (say, randomly selected) strong chess program as the official scorer of all games. A player’s score for a won or drawn game would be proportional to the total difference from in Ken’s metrics. Perhaps credit could be given also for a valiantly lost game.

This scheme would directly reward players for the amount of non-cheating they do. Or put more positively, human ingenuity apart from computers would bring the reward. The ability to sleuth strategy beyond computer moves was already demonstrated in so-called freestyle tournaments held in 2007-08 and 2014. A particularly nice example of playing a sacrifice that the computer does not like was executed at turn 18 by Magnus Carlsen in his World Cup win today over Vladimir Fedoseev.

Open Problems

We admit the tabletop suggestion is far from electrifying, but has anyone come up with better? As always we welcome suggestions from our readers, or pointers to forum discussions that you agree with.