Главная страница.

This writing has been caused by my recent disqualification – this time after the tournament held at playchess.com. As in the previous case, I mean the ICC event; I was banned due to the expert analysis. The experts reached the conclusion that I was playing with computer assistance. I see that the real situation in Internet chess is leaning towards one of the worst-case scenario I have predicted in my earlier writings on this subject. Here I am trying to systemize my view on the problem once again and to convince those who are open to sensible discussion. I will have to repeat my old arguments, none of which has met worthy objections so far. However, I will give new reasons as well.

So, to begin with, what’s the problem?
For several centuries chess has compared favourably with, for instance, card games thanks to the virtual lack of cheating. Common rules and an opportunity to see the pieces on the board at any time worked to the advantage of a chess game. Moreover, chess was almost free from referee arbitrariness, which also was an asset. A referee could cancel a goal, give a low grade to a figure skater but was unable to repeal the checkmate and annul the result of a chess game. Unfortunately, we have reached the stage when chess is losing the above advantages.

Progress is appeared to be the main trouble. Internet provided those chess fans who were short of money and free time with an opportunity of playing their favourite game. Chess software updates were of great help to professional chess players in analysis of the opening and ending position as well as in annotations. However, the combination of the above benefits did not seem to be that optimistic. A cheat has come to the scene, i.e. a person who uses computer assistance in his Internet games, which gives him an overwhelming over decent opponents. Almost all interested parties (including myself) agree that cheating is a harm threatening the chess prosperity and that it should be stopped. The opinions differ while we start referring to the specific steps.

ICC tried to solve the problem in an easy way – providing registration of every client computer with Troyan that gave an opportunity of remote control of every computer connected to ICC for chess software. I don’t know how expensive this invention was but every cheat had to buy another computer. The next step was obvious. Every chess player who has ever analysed games with software knows that human games differ a lot from machine games. The organizers of the WICC set up a committee responsible for the expert game analysis that later disqualified the player who took the second place in this tournament. ICC and playchess followed the example.

What’s wrong with it? Why do I consider the way, chosen by a lot of distinguished people supported by other distinguished people, to be a dangerous deadlock? The answer is clear: it deprives me of money and reputation while these people think that both have been acquired unfairly. Nevertheless, as my first statements on the subject were made before the first sanction I hope there are those who will think my arguments over. I will try to explain.

Question number 1 – to what degree can we trust any experts?
Mankind has accumulated a rich experience in research and there are pretty good procedures that allow to differentiate between the scientist and the charlatan. In our case we could use the following method:
A person who claims the job should be given a test. For instance, he can be provided with one hundred anonymous games that he has never seen before. Some of the games are played by humans, some – by machines and the rest – by advanced players. If he manages to score, say, 80%, let’s consider him an expert. Then when we need to check a tournament for cheating, we give the expert a heap of games (which are again anonymous, including a test lot, i.e. games from other tournaments played by humans, machines and advanced players) and if he classified the tournament player as a machine or and advancer, the conclusion is obvious. Does everything work? Yes, it works but also breaks at every stage.
First of all, I would contend there are no experts who are able to pass the above test. Secondly, are we ready to unfairly accuse every fifth player to stop cheating? What about every tenth of every hundredth? Thirdly, the games played by low-rated people differ a lot from those played by grandmasters; the same is true for different chess software. As for advanced players, every one of them has their own style that depends on the personality, the software he uses, his computer and the time control. The task is much more difficult that it seemed to be at first sight. Fourthly, a thorough control of cheating will take a lot of man and machine hours even at one tournament. As a result, the organisers have to save on something.

For instance, at ICC I was disqualified along with the woman grandmaster because an “expert” of a second grade decided that she had been playing too well. I also remember a gross mistake by a real expert – a well-known grandmaster, one of the best specialists in computer chess. The drawbacks of the subjective approach, when an expert reviews a game being fully aware of the chess players’ names and acts based on his idea of who can play well and who can’t, are clear. But we don’t have anything else but the witch-hunt and the identification process when a witness can see the only person who is accused.

Sport has faced the similar situation with drugs and a lot of people compare cheating with drugs in chess – but this is off topic.

Let’s try to take a look at another check method, now a purely machine one. Here we can rule out the possibility of prejudice and abuse. Software analyses every move based on the time record and compares it to those suggested by various chess programmes. The percentage of coincidence gives rise to accusation/acquittal. Software won’t concentrate on the beginning of the game, say, until the novelty is used, and on the ending (converting of a substantial advantage which can be done with dozens of ways), it is also possible to exclude forced variation with the only moves without a human expert. But how are we going to explain to the machine what an “obvious” move means? And how long will it take to write and adjust this software to the trustable results? I can easily imagine a situation when this software accuses, for example, Capablanca of cheating? Even after that, my second and third arguments are also important here.

So, there are no experts who deserve absolute faith. Now I will try to prove that such experts are impossible to appear. An advanced cheat can consult a software only a couple of times during his game having had no substantial influence on the percentage of moves that coincide with the machine ones and having gained a decisive advantage over his decent opponent. Let’s suppose we acknowledge the existing situation when a certain percentage of coincidence with the machine moves is considered to be criminal. Then we raise this percentage as high as possible - a clever cheat will always avoid punishment because unlike a decent player he can manage the situation deliberately making weaker moves if necessary. If we lower the percentage, we will “catch” much more decent players than it happens nowadays. If we take the borderline a decent player is on the mining here – one extra strong move separates him from disqualification and disgrace while he does not know which one. But a cheat knows and will easily avoid danger.

Many people are likely to think about identifying an indecent player based on the special diagram that reflects time a player needs to think the moves over. Having such diagrams for several hundred games of one chess player we can obtain something similar to an individual “handwriting” and to reach conclusions based on deviations from it. I think that any reader will find a lot of disadvantages of this method without effort.

In my view, here I came to the main reason why my so unpopular standpoint has not met any reasonable criticism for three years. None of the experts wishes to reveal his procedure for two reasons: 1) if a cheat knows about it, he will easily avoid it; 2) any procedure is vulnerable to criticism and can’t be faultless (see the reasons above) which is required. Nobody likes to allow legalization of advanced chess because it is almost the same as to advocate legalisation of drugs or narcotics.

In our moral middle age that dominates in Internet chess now I think it’s sensible to suggest a specific measure taken from the middle age. I mean, chivalrous courts of honour. According to the rules of such courts the accused could challenge his accuser to an encounter. It was considered that God could not allow injustice and the one who was right would win. It is not that difficult to observe for the rules in an encounter. Any playing zone and any organiser of an Internet tournament is able to provide this opportunity. If an expert is too weak to uphold his accusations, any wandering grandmaster that shares his opinion, can play for him, or he can hire a grandmaster for money. Who knows maybe it will revive the profession.