Greengard's ChessNinja.com

Topalov Shuffle

| Permalink | 29 comments

He also talks about comps in this article. Nothing much new, although I was surprised to hear that Topalov has never played shuffle chess since so many top GMs have participated in the Mainz Chess960 events.

My radar has picked up news of a very high profile shuffle event later this year, btw. I've never been a big fan of it and certainly don't want to see it replace classical chess. But it's clear that computers both databases and engines are breaking down the openings at an ever-accelerating rate. While there is still plenty of scope for originality, prepared variations reaching past move 25 and to the very end of the game are increasingly common and sheer memorization is more important every year. Balloted openings are one possibility. The problem there, of course, is that you eliminate most of the innovation and all of the balloted positions will be analyzed to dust.

Kasparov has suggested a form of balloting for shuffle chess. Each year, or each event, for example, a few positions will be used instead of starting each game from a surprise position. This allows for some work and some of the goofier positions can be discarded. Many positions give White a huge advantage or are so bizarre that they hurt the eyes, not to mention leading to terrible blunders and 20-move games. It would be nice to keep some of the patterns and geometry of the classical game.


He's right, except it won't take ten years!

Is Rybka the leading contender to win the next computer chess championship? Does anyone know how I can reach the developers?

http://www.rybkachess.com/ has his address at the top. I'm running an interview with him in a few days.

What positions exactly give white a huge advantage? With symmetrical positions, it shouldn't matter, should it?

It matters a great deal. The sharper the position the more the value of the move increases. Some starting positions are practically forced wins for White because of the precarious position of the kings. Of course mistakes can happen, but computers can be used to look at them objectively. Some positions get relatively large plus scores for white and allow for forcing lines straight away.

Hi everybody, I have just got the problem-solving attitude 8) :
A solution for this problem with the white pieces having a great advantage in some initial positions is [1] to force all players play from both sides of the board during a single tournament (against the same opponent or not : depends on the format)
Of course we should use luck to determine who is to play in the first rounds : the first to play the black will have a disadvantage since position will be less analyzed.
Another option [2] is to throw some initial positions out : sounds quite heretic, doesn't it ?
Or [3] give anyway the statistical elo disadvantage back to black regardless the result of the game ! (Sonas can help here ! ;-))
(Well, he would most likely be the man to hate by the winners who would lose elo-points because of his calculations, hehe!)
Kill preparation of classical chess by : [4] forcing all tie-breaks be played with a standard variation of shuffle chess initial positions (eg. 16 or 32, level in theory). [do it when Topalov retires please 8)]

There is a great problem with all those shuffle plans : a person who don't know how to play the 4th move in French Defence will always find it fascinating to play Frence Defence and win or lose regardless of the shuffle variation between eg. Anand-Topalov game. This will make the gap of understanding between players of the different caliber even more huge : this will not help the game itself : Shuffle games will be incredibly difficult to understand during the phase of the opening.

Are there shuffle chess websites?

Rule changes to make shuffle chess more fair: Black can veto the initial position and get a new one (once per game). Or while we're changing the rules, how about no pawns can move two squares until the third move. That should give the king time to find safety.

Mig, can you give an example of a starting position that is "practically a forced win"?

Not off-hand. But I was around when my friend Pablo Ricardi was preparing to play a "Fischerandom" match against Torre when Fischer visited Buenos Aires in 1996. Obviously specific preparation is worthless, but in practice we found positions that leave black feeling helpless very quickly. Of course humans wouldn't find every line over the board, but it doesn't seem too fair to give positions where the score goes from the usual 55-45 for white to 60-40 or worse. I've seen a few of these with computer testing, although not with the Fischer rules.

The basic point is that you have super-sharp positions immediately and white has the initiative, so black is on the precipice to start. Of course this balances out in the great scheme of things - sometimes you are on the white side - but getting a "bad" position with black in a critical game would be obnoxious.

It's tricky to compare statistics against regular play because they have opening books. And there aren't many collections of shuffle chess games because they aren't standard PGN. But I don't think white would score 60+% in computer play without books. But I haven't seen tests with the Fischer rules that allow castling, which probably increases Black's defensive chances.

What happens in modern chess deeply annoys me. Every year chess is transformed from a spiritual, creative game to a sheer memorization contest. I know players about 1700-1800 who memorize whole books trying desperately to gather some elo points, so one can easily imagine how this phenomenon is oversized in professional chess.

Random chess or some other variant maybe are the only hopes for the survival of the game. In theory of course, these variants have a lot of disadvantages in comparison with modern chess, but be sure that improvements will be generated right after their application. Anyhow some fresh air and originality are desperately needed!

ps. I am forced to play curious and unorthodox openings just to shortcut my opponents' opening preparation. And that because my job and limit leisure time do not allow me to study openings (i am about 2100 fide rating).

Hi Mig.

So who is going to be playing in this high profile shuffle match?

So Mig, who's going to play in this high-profile shuffle match?
Topalov-Kasparov? Karpov-Fischer? Bush-Putin? Kobe-Shaq?
Dying to know...

Unfortunately with some events and rumors I can't give details or the people who host them get pissed off about having their big press release and press conference surprise ruined. So I make comments like that one so, when it is finally announced, everyone thinks, "gee, Mig knew about this months ago, he must be really well connected, cool, and handsome." Tricks of the trade. Hey, is this thing on?

Sometimes I'll extort a day or two of advance announcement, that's another way to do it. As in, I won't blab this to everyone now but let me be the first to announce it so I can appear well connected, cool, and handsome. I usually find these things out from conversations with the players and sometimes just blab. But if I know the organizers I'll ask if it's cool for me blab. No blab on this one yet. But you'll be the first to know.

Thanks Mig. I agree that some positions seem to favour White more than the standard initial position, just thought what you said sounded a little exaggerated. Kasparov's idea is interesting, I think Kramnik and Anand have been talking about similar things as well.

Personally I have always favored the variant where the players alternately place their pieces on their back ranks, so that the resulting starting position is itself a product of the players' strategic choices rather than chance. This has never really gotten popular, but I honestly don't understand why not.


There is no doubt something must be done, and Shuffle Chess, even altered as mentioned by Kasparov, and Mig is a decent option.

I know a young man at a local chess club who I defeat regularly over the board, but he has a particular opening memorized, of which I avoid, that he not only can defeat a master with, but plays it within the 5 second increment on the clock!

This is a perfect example of why something needs to be done, or looked into. He is not strong enough to defeat a class A player (myself) but knows a particular opening like the back of his hand, along with all it's sidelines.

The question: Is he playing chess, or simply copying moves discovered by the greats before him?

Option 2:
Introduce Chinese chess, I forget the name, but I hear it is more in depth than the game we know, and has far fewer draws.


If some shuffle positions offer a clear advantage to white, maybe we could shuffle starting colour? In other words, if we are free to shuffle the pieces, why not shuffle the start so that black gets first move? It could be handled via coin-flip or (in the case of matches or tournaments) by stating that day one is white, day two is black, etc.


Like Mig, I do not want to see FRC/960/Shuffle chess (whatever you want to call it) replace classical chess. In fact if it did, I would probably find myself a new hobby. I would also demand someone pay me back all the money I have spent on classical chess books! ;-)

This leads to my main point. If there is a "problem" with chess, where is it and who should change? If there is a "problem" it is really only at the very highest levels. My experience is quite different from christosk's. I am at an otb level near his and I don't find the same problem. A 17 or 1800 can memorize all the theory he wants but how much does he really understand of the moves he is making? If he understood them fully, he wouldn't be a 17 or 1800! Theory hounds are often easily in trouble if you make one move they don't know. Greg Shahade recently made the comment that he would still play the Elephant Gambit even after he became a master and that in his experience openings didn't matter as much at that level. A pretty high level to be sure. I know he has posted here in the past so I hope he'll correct me if I am wrong.

Also, if we did only change chess at the highest levels (if it isn't, as I believe, "broken" at the lower levels) what effect would that have? Do we want a chess world where the top players are playing a different game and the rest of us are still playing chess?

Lastly, Dan Heisman has been asked about this on his ChessFm show and while I don't know how much thought he has given to the matter, his opinion was that changing to 960 wouldn't really make chess any less drawish of a game (at the highest levels or with perfect play). I believe he is of the opinion that to do that you need to play an entirely different game like Gothic chess (which has an extra piece and a bigger board). In fact, I have the impression that Heisman is something of a supporter of Gothic chess or at least he has mentioned it a number of times on his show. I have enormous respect for Hesiman as a chess teacher but I am quite satisfied with playing chess as is.

I'm not a theory memorizer by any means, and I've always found it relatively straightforward to throw a monkey wrench at players who just memorize openings. The only exception is in gambit openings and the like, where every tempo can potentially make a huge difference, and what I do there is pick a line that's non-forcing but solid.

I think the theoretical explosion is vastly overstated, primarily becuse top players, who are highly visible, keep bitching about it, and to be fair, the problem really *can* get acute at the top. But it doesn't seem to be a huge issue with most GM's, and I've read interviews with 2600+'s who don't study much at all.


If some shuffle positions offer a clear advantage to white, maybe we could shuffle starting colour? In other words, if we are free to shuffle the pieces, why not shuffle the start so that black gets first move? It could be handled via coin-flip or (in the case of matches or tournaments) by stating that day one is white, day two is black, etc.


This doesn't solve anything. You could achieve the same effect by flipping the positions of the pieces horizontally (ie. a1 and h1 flip, b1 and g1 flip, and so on) and painting the white pieces and black and the black pieces white. Of course it's not the color of the pieces that matter, it's who has the first move. But flipping the positions from day one to day two doesn't promise to have any effect, really. What are you saying is equivalent to saying that flipping the starting king and queen's positions in normal chess will solve white's opening advantage problem, which it doesn't (since it is still the same position).


Art, the problem is that (assuming a symmetrical position) whoever moves first has a big advantage, so by "shuffling the color" you just give the advantage to the other player.

However, this might not really be so bad over multiple games. Suppose for example one held a shuffle tournament which started everyone from a common opening position in every game, the position being selected maybe a day ahead of time? Then even if you got one of these sharp positions it would be just a matter of increasing white's normal opening advantage, and if the color assignments balanced out then everything would be ok. Mostly :-)

I still like the variant where the players put their own pieces on the board though.


I dunno about mirror reflections; I always seem to do better playing right to left.

Too lazy to look category: I infer that "shuffle" chess you can't castle? I thought this was the saving grace of Fischerandom, that no matter the initial position you could always clear the intervening pieces and go to the standard king-rook configuration...

Finally I say: If a certain chess960 position is a dead loss for black, if the board is no longer a reasonbly level playing field, then too bad for the black player! No, no, wait, hear me out! Seriously! The obvious solution is to redefine a game as two starts from the same position with the players flipping sides for round two. Now the guy with black at the start has an advantage because he can benefit from any intelligent play by white in the first game.

Ok, the other alternative is to filter the 960 to some smaller number of 'playable' positions but of course this risks contravening the premise... which is I guess what Mig is saying, that computers are rapidly reducing chess960 to a partially-solved problem. What's interesting to me is the baby-with-the-bath-water aspect: Chess is fun partly because we get to learn and then use pattern recognition which our brains are built to do, so going from 1 to 960 we throw out some of those patterns... and if 960 is still not enough we can introduce still more randomness, say like how knights move. But then there goes still more of all that built in pattern recognition and we have to play as beginners... popular with just about nobody (except those GuitarCraft dudes over in the corner).

Anyway I still like the dual-game idea. -r

Watching children play chess, no shuffle chess is necessary. It's us adults who wish to memorize it all. But even Ivanchuk, with his alleged photographic memory, couldn't get to the very top. I imagine the opening book and software publishers would not prefer a change here. The memorization secret, and how much of any opening anyone really knows, may be revealed in an an opening theory contest, which I doubt will ever be organized.

Just to clarify, I use the term shuffle chess for the general variant and that includes Fischerandom. That's just a variant of a variant and wasn't the first time castling was tried.

Playing two games with colors reversed is fair enough, but it just takes a long time. Unless of course the time controls are cut in half again, further randomizing things.

My principal argument against shuffle chess has always been a larger version of what Rob describes. It removes work and knowledge from the game and on a fundamental level reduces the skill of the game. Its champions say it will favor talent, which means it will favor tactical talent. A talent for hard work, a talent for study and preparation, a talent for strategy (which is based on application of absorbed patterns, i.e. analogy), all that is washed away with a tactical melee from move one. The endgame is still around, at least. Until they decide that's too much work too and abolish stalemate or something.

Great point, Mig.

Another good point I read at another chess forum (can't remember where or who) is that what shuffle or FRC loses is the dialogue between different games. We can see how Kasparov handles the Ruy and compare it to Karpov. That sort of thing seems much less likely with any sort of shuffle chess.

Some random thoughts in no particular order...

Regarding forced wins from certain inital positions in fischerandom, I'm the skeptical type. I'd have to see it to believe it.

Personally, I feel that shuffle chess is still a long way from being necessary. Players over the ages have moaned the death of the game. We get these ideas because the very top echelon have a super selective opening repertoire, and tend to be extremely prejudiced regarding anything that is even mildly off beat. Needless to say, for the other 99.99999% of the world, this is irrelevant.

Look at Morozevich. The guy has been at or very near the top ten for a while now, using openings that are deemed coffeehouse, bizarre, or even just plain pathetic. Obviously his monstrous talent enables him to escape some very bad positions, but his attitude infuses some of those openings with a completely different energy which, to put it mildly, must be quite unnerving to the opponent. How quickly we forget that chess is a human game.

Gary Kasparov revived to EVANS GAMBIT briefly, well over a hundred years after its heyday. Every openings manual had deemed this try for white as not a serious attempt to win---virtually a forced draw. And he won against ANAND, for god's sake.

Obviously there are many dubious openings that cannot be made viable. The really good openings are well known. But there is a lot of middle ground where tepid souls are afraid to go---territory that has psychological and practical value. I wish more players would give this place a try. It's so much more rewarding in my opinion.

A word about computers---they don't "solve" openings. They blunder check them. They can't evaluate possible endgames that might arise from them, or judge intangibles. People that claim the openings are getting all played out are just being lazy. If you just take time to sort through variations given in ECO and other tomes, you will encounter all kinds of positions with decent alternatives to what is given in the "book". Yes, it takes work. If you love chess, it's also fun.

I don't think a variant is necessary. Sure, memorization is becoming more important (more at the top level than anything), but that's not a big deal. The greatest players will be more likely to have great memory. No grandmaster no matter how powerful their memory will be likely to remember all the analysis even if it's all laid out before our eyes. The creative elements will never go away.

I basically agree with Chris Anderson's view 100%, especially about the moaning GM's with narrow repertoires!

For at least 99% of players, chess will always be about general strategy, tactics, patterns, intuition, creativity, psychology, etc. It will always be fun. It's too complex to memorize.

Two days after this blog posting about chess960 (FRC), a new chess book was published entitled "Play Stronger Chess by Examining Chess960: Usable Strategies of Fischer Random Chess Discovered" (ISBN 0-9774521-0-7). Searching Amazon.com books for 'chess960' will find it.

Fischer promoted chess960 as a way to fix broken traditional "chess1". This book argues that a bigger and more important reason to give chess960 a chance is that chess1 has been hiding a lot of proper chess middlegame themes from us. The chess world has simply not yet realized this.

Chess960 certainly produces different middlegames than does chess1. So, was it just a fantastic stroke of improbable luck that the ancients who invented chess circa 1400 happened to choose the only good initial setup of the 960 they had to choose from? Wow, what incredible luck! Or is it that hundreds of initial setups lead to interesting middlegame patterns that chess1 rarely or never lets us see?

Logical positions in chess960 arising from intelligent play are not awkward, but thinking them so reveals that our experiences from chess1 are more narrow than had thought.

Chess960 is a richer capture of fundamental, basic chess.

It is not an either-or situation. Some tournaments should be chess960.


"The basic point is that you have super-sharp positions immediately and white has the initiative, so black is on the precipice to start. Of course this balances out in the great scheme of things - sometimes you are on the white side - but getting a "bad" position with black in a critical game would be obnoxious."
- - - - -

I understand why a person would make this comment. But to me it only shows how boxed in our minds are from having played only chess1 since forever, and not having played any chess960.

The rule that White goes first is probably necessary in chess1 because the initial setup is not only static, it is also symmetrical on the two wings 'a' and 'h'. This lack of variety places a harsh limit on the range of first plies for White.

But in chess960 we can finally ditch the old awful rule, which has given White a highly unsporting lead in wins, 4 White wins for every 3 Black wins: awful (though like sheep we chess players think of it as tolerable because we were born into it and it is all anyone has ever shown us).

In chess960 the variety allows us to say:

"Player AA begins the game by moving a White piece. Now on his first ply, player BB has the option of choosing which color pieces he wants to play."

If AA makes a very strong move, BB will choose White. So the chess960 opening rule motivates AA to open with a move that keeps chances even between the two colors. Problem solved.

[To avoid the hassle of rotating the board or switching chairs, the chess960 book explains an easier "Fair First Move" method of Black "adopting" White's too-strong first ply.]

Gene Milener

Twitter Updates

    Follow me on Twitter



    About this Entry

    This page contains a single entry by Mig published on January 24, 2006 12:18 AM.

    Corus 2006 r8 was the previous entry in this blog.

    Corus 2006 r9 is the next entry in this blog.

    Find recent content on the main index or look in the archives to find all content.