I’ve never played this game myself, but many of my friends and students have taken part in the traditional Fischer-random tournaments in Mainz. Most of them liked the new game. They were very happy not to have to waste time preparing for the game, and it was interesting to test themselves and compete with their opponents in solving original tasks. That being the case, one can only welcome the continued hosting of such events, and hope there will eventually be more of them.
This and the following excerpts are from Part 2 of 'Polemic Thinking' (PDF). [Polemic - 'an aggressive attack on or refutation of the opinions or principles of another' (Merriam-Webster.com)] Note the curious phrase 'to waste time preparing for the game'.
But this can hardly mean that chess960 should be promoted as the designated successor to everyday chess. [...] The problems involved with such an enormous change in the rules should be examined from all sides and tested, with all aspects considered in order to find out whether there are drawbacks that might prove dangerous to the future of chess.
One of the main criteria of beauty (along with subtlety and originality) is the soundness, the correctness of the moves, of the individual ideas, or of entire games. And here is where I have some doubts about the future of chess960.
Doubts? What doubts?
In Fischer chess, where the majority of the pieces – if not all of them – are standing in unusual positions, we must deal with many new and unknown elements. As a result, a chessplayer has almost nothing to refer to in looking for a move; he’s playing “without line or compass.” I can assure you that even leading grandmasters play a weak game of chess960, full of both strategic and tactical errors. [...] So these games almost never show us any aesthetic value. If we remember how hard it can be to discover the secrets of a position even in traditional chess, where we can refer to many generations’ worth of experience, what I’m saying becomes logically obvious.
As proof that 'even leading grandmasters play a weak game of chess960', Dvoretsky gives two examples. The first is the same game I used in A Chess960 Catastrophe.
The level of play demonstrated here by grandmasters isn’t much different from (to take an example from traditional chess) the efforts, successful or unsuccessful, to exploit the weakness at f7 from the starting position, and deliver the "scholars mate". Of course we need to take into account the fact that in Mainz, the games were played in rapid chess; however, I suspect that, even under a classical time-control, the quality of play would not have risen very much. In the early days of chess, many such naïve games were played. As experience grew, so did the understanding of the principles of opening play; new schemes of battle appeared and were worked upon, and those that didn’t work out were tossed aside.
In chess960, there will be practically no accumulation of experience: there are too many opening positions, and too many differences between them. And thus, the concept of the opening phase will find itself frozen, for a long time, at a childhood level.
Let me summarize, briefly: Playing Fischer-random is undoubtedly interesting (and probably even useful: overcoming routine, and developing an unfettered approach to the position). But studying played games is of no interest, because it’s almost impossible for anything creatively important to come from them (when measured against the level that both amateurs and experts in classical chess have grown accustomed to). So switching to this new game involves a serious risk that we may lose the aesthetic element of chess – and consequently, a great number of its adherents.
This argument is similar to the one I addressed in More Arguments Against Chess960, where I quoted Tim Krabbé writing, 'Any form of shuffle chess puts chess back 200 years.' This highbrow dismissal of chess960 because errors occur early ignores the reality of modern grandmaster chess. The start of a game is two players following a known path for 'X' number of moves, after which they follow computer based preparation for 'Y' number of moves, after which they are on their own. At this point there are three possible outcomes: either they agree to a draw, or one of them blunders, or they continue playing as best they can.
The example I used in that previous post was the first game from this year's Anand - Topalov match where Anand blundered on move 23. As we later learned, the blunder occurred because he forgot his preparation ('Y') and mixed up his ideas. As for agreeing to a draw as soon as move X+Y is reached, I could give lots of examples, starting with the 2004 Kramnik - Leko match, which saw a humiliating loss by Kramnik because his computer preparation was faulty.
The reason we see errors earlier in chess960 is not because the games are played at a 'childhood level' (Dvoretsky's phrase). It's because the players are on their own earlier. That they play chess better than they play chess960 is an illusion, a fiction, a fabrication due to conveniently overlooking the X+Y unoriginal moves that preceded real play. The truth is that GMs play chess960 very well.
As for Dvoretsky's remark that 'studying played [chess960] games is of no interest, because it’s almost impossible for anything creatively important to come from them', this implies that the only creative phase of a game is the opening. Is there really no creativity in the middlegame or endgame? If there isn't, I can throw away Dvoretsky's own books plus all the other books I've mistakenly acquired on those subjects.
As I wrote in the response to Krabbé, 'Playing over chess960 requires playing slowly from the very first move, just like playing a chess960 game requires real thinking from the very first move.' That last thought is worth repeating: chess960 requires real thinking from the very first move. Real thinking, creative thinking, has little to do with memorization. There is no 'X'; there is no 'Y'; there is only chess.
Later:See also 'The Chess Instructor 2009' (New In Chess 2008),Ch.2 'Mark Dvoretsky: Controversial Thoughts', section 'Should we all play chess960?' (p.30). On reading through this post a second time, I realized that Dvoretsky has selected a difficult chess960 position (RKR in the corner, where castling is particularly problematic) played at rapid time control and used it to condemn the entire idea of chess960. Anyone for sophistry?