28-05-2013, 11:32 AM
Thanks for this Graham.
On out-performing Carlsen…Carlsen would surely not be a good choice for such a comparison - going by ratings, he is the most likely player to either play stronger than top computer programs, or play as well but by playing in a different way from them. So while playing much less strongly than Carlsen, another player could ‘out-perform’ him at the (different) game of matching program moves. As Alan Jelfs said, the right comparison would be with other players of the strength at which Ivanov is accused of achieving by cheating.
“Anyone who still doesn't believe he was cheating is just naive.”
I personally think the chances are he was cheating, but you need objectivity to take precipitate action, and anyone claiming 100% certainty despite there being uncertainty is not being objective. The accusation is of electronic cheating, but the ‘evidence’ only relates to the weaker hypothesis that it is unlikely a 2200-rated player could have achieved such a strong match with computer moves if playing normally. Most of the debate then centres round this part of the story as if it were the smoking gun itself. True, the analysis of the computer matches is compelling. Even this part is not objectively overwhelming though, because the fit between theory and evidence isn’t completely uniform and needs explanations for where the evidence doesn’t fit the theory.
In your analysis Graham you don’t say why it is justified to remove the second game? More generally, it’s quite difficult to produce objective analysis, including ‘explanations’, if you have a prior belief in guilt – such analysis usually reads like the case for the prosecution. For example, in his video presentation <!-- m --><a class="postlink" href="http://www.youtube.com/watch?v=qhfCUdy2Tzk&feature=player_embedded">http://www.youtube.com/watch?v=qhfCUdy2 ... r_embedded</a><!-- m -->, Valeri Lilov says things like ‘this is a computer move’, ‘this is not a human move’, which suggest a prior disposition. Of the people Ivanov lost to he says something like ‘not only were they 1900s but they were kids’, temporarily forgetting the concept of junior additions! It’s not that he doesn’t make a ‘good’ case but that’s the point, it’s a case AGAINST someone - it should not be treated as if it were an objective evaluation. Also the more subjective explanations that are required (even if they are compelling), the less the theory can be claimed to be objectively robust.
Evidence of cheating method is lacking. Mike Scott made an excellent point regarding the Sally Clark case in which a physical crime was inferred from medical/statistical evidence – evidence that was found to be statistically wrong. But even if the statistical calculation had been correct, the medical ‘evidence’ that a child had been murdered was non-existent. It’s sobering that a courtroom of educated people had still accepted obviously flawed logic – i.e. that if the probability of something happening ‘by chance’ was small enough, then something else in particular can be relied upon to have happened. Ironically, the legal world has an expression for this – it’s called a non sequitur, or ‘it doesn’t follow’. The ‘something else’ was not the only possible explanation – the technical experts were guilty of failing to recognize the limitations of their own knowledge, and were actually just speculating outside it.
Going back to Ivanov, what other explanations are possible? Well maybe for example, Ivanov is bipolar and can only play half the time. Just saying…also we haven’t heard much from him. There’s a rather uninformative interview here <!-- m --><a class="postlink" href="http://en.chessbase.com/home/TabId/211/PostId/4008781">http://en.chessbase.com/home/TabId/211/PostId/4008781</a><!-- m --> but with not very searching questions.
I think it would have been best to let the clock run a bit more on this ‘difficult position’. Rather than refusals to play and ostracism that are damaging to the game, isn’t the best action simply to not transmit his games while he is under suspicion, or to transmit them with a delay, and see what develops?
This suggestion is also the action that closely matches the evidence for the accusation – which relies heavily on the explanation that when the transmission stopped, his strength stopped too. This means that if he was cheating, he was relying on the transmission of the moves from the tournament. It is also by far the easiest way to cheat – in fact it may require no technical activity at all from the cheater, just an accomplice and a means to signal moves.
Trouble is, with the ban we’ll not get any more evidence now!
Cheers
On out-performing Carlsen…Carlsen would surely not be a good choice for such a comparison - going by ratings, he is the most likely player to either play stronger than top computer programs, or play as well but by playing in a different way from them. So while playing much less strongly than Carlsen, another player could ‘out-perform’ him at the (different) game of matching program moves. As Alan Jelfs said, the right comparison would be with other players of the strength at which Ivanov is accused of achieving by cheating.
“Anyone who still doesn't believe he was cheating is just naive.”
I personally think the chances are he was cheating, but you need objectivity to take precipitate action, and anyone claiming 100% certainty despite there being uncertainty is not being objective. The accusation is of electronic cheating, but the ‘evidence’ only relates to the weaker hypothesis that it is unlikely a 2200-rated player could have achieved such a strong match with computer moves if playing normally. Most of the debate then centres round this part of the story as if it were the smoking gun itself. True, the analysis of the computer matches is compelling. Even this part is not objectively overwhelming though, because the fit between theory and evidence isn’t completely uniform and needs explanations for where the evidence doesn’t fit the theory.
In your analysis Graham you don’t say why it is justified to remove the second game? More generally, it’s quite difficult to produce objective analysis, including ‘explanations’, if you have a prior belief in guilt – such analysis usually reads like the case for the prosecution. For example, in his video presentation <!-- m --><a class="postlink" href="http://www.youtube.com/watch?v=qhfCUdy2Tzk&feature=player_embedded">http://www.youtube.com/watch?v=qhfCUdy2 ... r_embedded</a><!-- m -->, Valeri Lilov says things like ‘this is a computer move’, ‘this is not a human move’, which suggest a prior disposition. Of the people Ivanov lost to he says something like ‘not only were they 1900s but they were kids’, temporarily forgetting the concept of junior additions! It’s not that he doesn’t make a ‘good’ case but that’s the point, it’s a case AGAINST someone - it should not be treated as if it were an objective evaluation. Also the more subjective explanations that are required (even if they are compelling), the less the theory can be claimed to be objectively robust.
Evidence of cheating method is lacking. Mike Scott made an excellent point regarding the Sally Clark case in which a physical crime was inferred from medical/statistical evidence – evidence that was found to be statistically wrong. But even if the statistical calculation had been correct, the medical ‘evidence’ that a child had been murdered was non-existent. It’s sobering that a courtroom of educated people had still accepted obviously flawed logic – i.e. that if the probability of something happening ‘by chance’ was small enough, then something else in particular can be relied upon to have happened. Ironically, the legal world has an expression for this – it’s called a non sequitur, or ‘it doesn’t follow’. The ‘something else’ was not the only possible explanation – the technical experts were guilty of failing to recognize the limitations of their own knowledge, and were actually just speculating outside it.
Going back to Ivanov, what other explanations are possible? Well maybe for example, Ivanov is bipolar and can only play half the time. Just saying…also we haven’t heard much from him. There’s a rather uninformative interview here <!-- m --><a class="postlink" href="http://en.chessbase.com/home/TabId/211/PostId/4008781">http://en.chessbase.com/home/TabId/211/PostId/4008781</a><!-- m --> but with not very searching questions.
I think it would have been best to let the clock run a bit more on this ‘difficult position’. Rather than refusals to play and ostracism that are damaging to the game, isn’t the best action simply to not transmit his games while he is under suspicion, or to transmit them with a delay, and see what develops?
This suggestion is also the action that closely matches the evidence for the accusation – which relies heavily on the explanation that when the transmission stopped, his strength stopped too. This means that if he was cheating, he was relying on the transmission of the moves from the tournament. It is also by far the easiest way to cheat – in fact it may require no technical activity at all from the cheater, just an accomplice and a means to signal moves.
Trouble is, with the ban we’ll not get any more evidence now!
Cheers