The cheating wars

The cheating wars

IM Cryptochess

Well, it has been a looong time since I've posted something on here. Unfortunately the topic drawing me back is less than ideal, but with the all the recent press surrounding cheating and the numerous very long discussions I've had with people as a result, I wanted to get through my thoughts here and hopefully start a more abstract discussion on the topic, as it seems to me that current discussion surrounding cheating largely misses the point.

Cheating in chess is hardly a new problem, dating back to more or less the invention of the game itself. While the ubiquity of computers and relatively recent advances in information dissemination have brought the issue into the limelight time and time again, computer assistance is certainly not necessary for cheating. In particular, sandbagging (intentionally losing to lower ones rating) is an incredibly prevalent cheating method, and one that is receiving renewed attention after the disaster that was last weekend's USCF Junior High Nationals. Thus it is important to separate the rather nebulous concept of "cheating" into a few distinct categories, which I'll get back to in a moment.

First, however, I want to discuss the various incentives surrounding cheating and its detection. There are more or less five actors here: the cheater, the players, the organizer/directors, the federation, and the general chess public. For the cheater, the incentives to cheat are usually pretty straightforward: some combination of prize money, rating points/titles, and publicity. Players, obviously, have similarly straightforward incentives to detect cheating, as playing against cheaters is more depressing than trying to find an advantage against the Berlin. But from here, things get much less clear. Organizers, from what I can tell, actually have little to no incentive to detect cheating; in particular, investigating cheating is a huge administrative hassle with either no (if the accused is found innocent) or negative (if the accused is found guilty, which can spark legal action) benefit. Thus, from the organizer's perspective, as long as a cheater is undetected by the players, it is better to ignore them. This is not a view held by all organizers -- mostly only the large business-minded ones -- but it's held by enough of the market share to be a serious problem. One need only recall the case of Borislav Ivanov, which took a tremendous effort involving a 20-something player boycott and well over a year to finally force action.

This perverse incentive is obviously very bad, but more concerning is that it exists more strongly at the federation level as well. Taking punitive actions against a cheater, particularly at high levels, is even more extreme than an organizer -- while an organizer can legally defend themselves via a standard "private business" defense, federations have no such recourse as they are ostensibly preventing one's profession. Beyond the threat of legal action, the publicity generated from punishing a cheater is rarely positive -- it is sure to spark debate on how appropriately harsh the penalty is, raise questions as to how long the cheating went on, and various other topics that a federation would understandably rather not deal with.

The takeaway here is that if a cheater is not caught by a player, organizers and federations are disincentivized from independently investigating the case. FIDE's toothless ACC, as I will discuss later, drives this point home. In particular, except in the most egregious cases (where a player is caught red-handed), the only way to generate an investigation and/or meaningful punishment is by publicly calling out the suspected cheater.

Obviously, this is a terrible state of affairs, because public accusations are disastrous to everyone if wrong, and even when right their effectiveness is often unclear. They also cast organizers and federations in a poor light, because it is indirectly calling out their inability and/or unwillingness to handle the problem themselves. As a result, it is not surprising that organizers and federations have a vested interest in discouraging public accusations, even when accompanied by convincing or even undeniable evidence. Recently, FIDE took this a step further by punishing the accuser, notably not for the accusation, but for "[...] conduct likely to injure or discredit the reputation of FIDE [...]". It's a transparent and targeted message: keep quiet about cheating, because it makes us look bad. One may also note that in this decision, FIDE did not bother to evaluate the actual accusations themselves even though such evaluation already existed, only that they were made.

Though FIDE has taken this message a bit far, they are hardly the only ones to take a very laissez faire approach to cheating in general. USCF, for instance, responded to the blatant case of sandbagging mentioned above with a strange statement that essentially boils down to "not our problem". This fits a long pattern of inaction; there are a number of well-known cases of players straight up being caught (even at Nationals, which is USCF-run -- a case where the organizer and the federation are one and the same) that received little to no punishment from USCF, perhaps most notably Will Fisher whose case was never publicly explained and who continued to play events as little as a week after being caught (even today, the organization makes a point of making it clear that Fisher was never banned). His efforts even earned him a spot in the U.S. Juniors, an opportunity that many promising U.S. talents would kill for.

So what options does this leave the chess public? We've seen compelling evidence that:

  • Organizers and federations will not/will rarely act on complaints unless accompanied with undeniable physical proof.
  • Public accusations, regardless of merit, are silenced by the relevant federation and the accuser is punished in either case.
  • Public accusations, even if wrong, are extremely harmful to the accused.

The answer is: not many. I, for instance, have experienced this firsthand: there is one player who has undeniably cheated against me more than once, as well as against a teammate. Appealing to directors/organizers had little to no effect, though at least seeing me talking to a director spooked the cheater a couple times into backing off. Though I never seriously considered making a public accusation, it would likely have been futile anyway; in a USCF article containing one of his games, a comment suggesting that something was amiss in his play was promptly removed by the federation.

Indeed, the one recourse available to players is to join the ever-growing rumor mill, culminating in a sad state of affairs where a necessary part of my pregame preparation is messaging friends and asking if my opponent is "sketchy". This is not just an academic exercise; there are a number of strong players -- even some top US players -- who have been caught in the past but the details kept hush-hush (forcing us to learn about them from the slow spread of information rooted at players in the same event), so that the answer to the question is often "yes". But for obvious reasons, this is easily exploitable -- there are just as many strong/top players who have had unsubstantiated accusations against them (e.g. a well-known GM even once semi-publicly accused me after giving away a piece, which was laughed off but spawned a rumor that apparently reached another even more well-known GM halfway around the world). As a result, this situation is even worse than public accusations -- at least public accusations can be evaluated, defended, etc. especially when games are public, while rumors are impossible to contend with.

So we've now established that of all possible methods of handling cheating, federations have forced -- and actively continue to force -- the chess public into the worst possible option: whichever of "rumor mill" and "crowdsourcing" sounds more appealing to you. And while this works for blatant cases of computer assistance, which is what most people think of when cheating in chess is addressed, there are a number of other types of cheating that are simply much harder or impossible to "crowdsource".

One of these, and probably the most common (even more common than computer cheating), is sandbagging and more generally match fixing. We've already seen the most blatant sandbagging case a couple times now, but there are a number of more subtle (well, not even necessarily that subtle) cases that are very hard for players to accurately detect. Match fixing is also extremely common, most notably prearranged draws at the highest level, but what is lesser known is that this is often combined with sandbagging in a team effort to have at least one person claim the top prizes; there is one locally famous such ring in the NY area, and I'm sure this is not the only instance. This really brings the whole incentives thing I've been talking about to an ugly head: organizers and federations are the only ones who can reasonably be expected to monitor and detect this sort of thing, but they have a vested interest in not doing so, so these sorts of rings run rampant.

By the way, as a quick aside, this sort of "team cheating" is, in my opinion, by far the most dangerous threat to chess. We already know of many many cases involving a single accomplice, but distributing the work negates many of the obvious countermeasures (e.g. delaying broadcasts) against that. We've already seen this on a small but elaborate scale at the highest level, in a scheme that would make even the dramaticized version of the MIT blackjack team jealous, so it is not hard to imagine this on a somewhat larger and somewhat more intelligent scale, possibly even already happening.

And in fact, in many ways low-tech cheating is far more concerning than computer assistance, which we can at least pretend to have statistical measures against (though for many reasons I am not very convinced of their general applicability, which is a subject for another time -- e.g. a single bit of information is often enough). For instance, when I was a kid playing a for a large prize in an U1600 section (which, in particular, does not save or broadcast games), I played against someone whose method of cheating was decidely undetectable with such a small directing staff: her husband, an IM, simply told her what to play in a foreign language. An even simpler cheating method is to simply write some notes, say on opening theory, and stick them in a wallet -- this is extremely unlikely to be caught even if an organizer undertakes a search (which is rare, and generally only looks for electronics), and is completely undetectable by statistical methods as theory is generally (and certainly in Regan's approach) pruned out of the data. It certainly is not something that a player can reasonably detect.

So we are back to the main points:

  • Organizers and/or federations are the only ones in a reasonable position to detect most kinds of cheating, and certainly the only ones in a position to apply meaningful sanctions.
  • At the same time, organizers and federations are disincentivized to investigate and/or sanction cheating.
  • As a result, even when catching or finding compelling evidence against cheaters, players have no meaningful recourse. To me, it is clear that this is by either by design at the federation level or a very unfortunate, and very anticipatable side effect of federation policy.

Now, to me at least, it is obvious that something here needs to change. Of course, it would be nice if federations/organizers independently came to this conclusion and decided to implement some countermeasures, but outside of some good organizers (of which there are many, which I'm very happy about!) and some poor lip service from federations there seems to be little reason to expect this to occur. Instead, something needs to change at the incentive level in order to effect a meaningful shift.

The problem is that, for players, this is quite tricky. In general as consumers the answer is some form of boycott, but as federations are de facto monopolies in their respective regions and organizers in some areas enjoy more or less the same status, for many this would be roughly tantamount to quitting tournament chess which is simply too much to ask. 

However, the Borislav Ivanov case does offer some small encouragment: with sufficient public pressure, even federations are forced to act; the tricky part again becomes that generating such public pressure is virtually impossible without public accusations that are now being actively discouraged. It seems to me that this leaves essentially one option: decouple the investigative bodies from federations and organizers, so that it is fully independent and publicizes their findings if and only if compelling evidence is supplied and/or found. In particular, the FIDE ACC seems to be quite useless ala-TSA, partially because its incentive structure is aligned with FIDE's (as far as I can tell, the ACC has made a grand total of two rulings over 4 years, one a transparently political tool to disenfranchise Bulgaria and one to discourage even accusations that were privately supplied to organizers) (EDIT: it appears the ACC "handled" a third case, though it seems gratuitous to credit them for anything meaningful here as the player was caught red-handed) and partially because FIDE doesn't bother to meaningfully support it ("The ACC feels that it needs more financial and political support from Fide in order to resume its standard setting and policy making action"). Straight from the horse's mouth: "The current level of funding of the Commission is so low that it is heavily hindering its functioning. Commissioners have not been meeting since early 2015 [...]". I don't believe the USCF has a dedicated anti-cheating body analogous to FIDE's (as far as I can tell it's enveloped into the main body), but my view is that an independent one should be established.

My reasoning here is that such a body meaningfully affects the incentive structure in a way that forces federations to act -- a body that has already done the brunt of the investigative work for them, and publicly details their results, in theory should put enough pressure on federations to take meaningful action. This also more or less removes risk of false accusations substantially damaging the accused, and with appropriate implementation (the FIDE guidelines are surprisingly decent) such a system is relatively difficult to abuse.

Of course, the devil is in the details here; I'm not sure how to make this work logistically or what the exact incentive structure for this independent body will be, but I'm hoping to spark some discussion on this concept and whether it is feasible (or a good idea). In any case, I hoe that I've managed to explain what I think are the root causes of these issues and hope that others have better suggestions on how to move forward in the face of them; in particular, I'm hoping to make clear that computer cheating, while substantial, is not the only or perhaps even the most concerning method of cheating, and as such anti-cheating proposals should take this into account.

Finally, I want to make clear that while I've used some harsh language here, I don't intend this as a call-out of any particular organizer or any particular federation. I understand that the topic is extremely difficult to handle, and that from a federation's (and sometimes organizer's) perspective there are often no good options. But, at least to me, it is clear that the direction we are heading is simply harmful to the chess community and ultimately unsustainable; as a result, to me it is clear that we as a community -- not just as a federation, or as an organizer -- need to hold organizations accountable and make fundamental changes if we want any hope of winning the cheating wars.