Columns | October 15, 2011 21:44

Non-random Fischer Random

Non-random Fischer Random

Watching the 7th game of the Kasparov-Short blitz match last week made me realize once again how radical Fischer’s proposal to shuffle all the pieces on the first rank was and still is. If you want to avoid boring theoretical chess duels, all you have to do is force the players to play an unexplored variation or opening – problem solved.

The 2009 poster of the Mainz Chess Classic, which doesn't exist anymore. There Chess960/Fischer Random was played every year.

In that blitz game, Nigel Short played the St. George Defense (which usually arises after 1.e4 a6 but now appeared on the board after 1.Nf3 b5 2.e4 a6) and even managed to beat the 13th World Champion (who, admittedly, blundered an exchange in the middle game). He thus created an appropriate echo of another St. George game, in which another British Grandmaster, Tony Miles, managed to beat  former World Champion, Anatoly Karpov (at the Skara European Team Championships in 1980).

In fact, Short also beat Kasparov with the now-rare King’s Gambit – and with a rare line within the King’s Gambit at that - providing another argument for those people (and I think I consider myself to be among them) who claim that all it takes to solve boring computer preparation is some creativity in the opening. Is that too much to ask of professional chess players?

Of course, some will say that it’s easy to experiment in blitz, but that one can’t expect the chess elite to actually start studying the King’s Gambit for important tournament games. Playing this dubious gambit in serious competition will cost them serious money! I only partly agree with this argument, because there’s no reason why playing blitz should, in principle, not be as profitable as playing classical chess. (It seems that even Kirsan Ilyumzhinov agrees with me on this point!)

But let’s for a moment assume that it’s impossible to force the King’s Gambit (or the St. George, or any other opening that’s not considered to be ‘main stream’) down professional players’ throat:  what if we simply adjusted the starting position a little to help the pros make up their minds? Suppose from now on everybody would needs to start their game with the following position:

PGN string

So much for all Najdorf and Ruy Lopez theory! And that’s just the beginning, of course. All openings would have to be studied anew, because the slight modification will create all sorts of subtle and not so subtle differences. The game would still resemble chess sufficiently not to lose the interest of the general public, but the nuances would be different enough for the insiders to immediately appreciate the complete make-over of “boring” chess opening theory. 

Perhaps some will argue that this new beginning position is actually to Black’s advantage, even though it’s still White to move. Well, that might turn out to be true, but how ‘fair’ is the current starting position? Isn’t that considered to be better for White? Even so, to make it a bit fairer maybe we shouldn’t put a black pawn on a6 (which might also makes queenside castling slightly less attractive), but a black knight?

PGN string

Heck, we could even have this position and let White choose whether he wants to play with White or Black. It still would be a much more modest change and thus be much more likely to be accepted by both professionals and laymen. Doesn’t this modest change of the initial position makes the ‘real’ Fischer Random chess look absurdly radical?

We can get rid of all the special regulations of Fischer Random chess, or Chess960 as it seems to be called these days: no more need to create the various starting positions with a special computerized algorithm; no more need for confusing different castling rules. We can simply play the above position for the next, say, 600 years - until theory has evolved so much again that we’re ready for the next step – put a white knight on h3 as well. (And in the mean time, we can enjoy the evolution of completely new chess openings as more and more games are coming in.)

Perhaps even more importantly, the quality of the games will be much higher than those played under the current Fischer Random rules. Why? Simply because Fischer Random opening positions are too unfamiliar and weird even to super-GMs, causing them to blunder in a much higher percentage of the games than in regular chess. Normal pattern recognition is mostly useless in Fischer Random. As Tim Krabbé once said, “Fischer Random puts us back 200 years.”

I suspect Krabbé was even being polite in his estimation. Maybe it puts us back not 200 but 400 years. Remember those games from the 17th century in which even the strongest players in the world used to fall for what we now think of as ‘cheap opening traps’? Well, I recently went to watch the Dutch Fischer Random Championship in my hometown, Amsterdam. I was just in time to witness what everybody felt was the ‘dream final’ - the decisive game between Dutch GM Dennis de Vreugt and Yasser Seirawan, who in regular chess beat many a World Champion in his best days, including Karpov and Kasparov.

I was in for a disappointment: Seirawan blundered a full piece as early as move eight, thus robbing the audience of a thrilling finale (and handing De Vreugt his well-deserved title on a silver platter). I felt bad, not only for Yasser but also for the tournament organizers. Honestly, I think no audience in the world likes to see such ‘drama’.

And that’s hardly the only example from GM-practice in Fischer Random chess. I remember Gata Kamsky blundering on move six a few years ago. I think it was against Aronian – I’d love to show you the game, but unfortunately, because to my knowledge there isn’t any good commercial database storing Fischer Random games, let alone allowing for any kind of simple search functionality (no handy ECO opening codes, sorry!), I can’t. Whereas some people are still debating Lasker-Lasker, New York 1924, any game of Fischer Random or Chess960 seems utterly forgettable.

Why not get rid of this artificial stuff and just move a pawn or piece to a6 in the starting position instead? Even if you think I sound like a Luddite, it’s hard to deny that it’s a lot easier for all.


Share |
Arne Moll's picture
Author: Arne Moll


arkan's picture

Finally possible to comment :)

I don't think this will solve anything at all, a pawn at a6. Randomness should be a factor, maybe something like let the tournament software randomly choose beteen the a2-h2 or a7-h7 pawns before each game?

Also i don't see why chess960 is so bad? It's just not very common yet

Arne Moll's picture

Yes, the comments option was switched off by accident.

Anyway, why should randomness be a factor? Isn't the main goal of this form of chess to avoid any kind of heavily analyzed opening theory? Well, that can also be achieved by simply putting a pawn on a6 in the starting position.

(If the goal is to play a different kind of mind game, then I think the game of Go is a wonderful alternative!)


Macauley's picture

Nyah. Chess 960 is much more interesting to watch than just putting a pawn on a6. In St. Louis last month they didn't even use a computer to select the starting position. It's not hard to follow the rules for setting up a position. Also, doesn't DGT make a clock that will give you the position a the touch of a button then times the game?

Arne Moll's picture

I'm not denying Chess960 is interesting to watch, or fun to play - and watching or playing one game with a pawn on a6 would be equally interesting (or uninteresting, depending on your personal preference).

What would, in my opinion, become more and more interesting after just one single game, is to see this position occur in more games so that opening theory, one of the key 'scientific' properties of chess, actually gets a chance to develop.

This is an aspect that's completely lost in the Chess960 proposal, where each round a new position is chosen (or was this different in St. Louis?). It's this 'reset' principle that, in my view, completely destroys the beauty of opening theory evolution. This, and the 'homework' aspect, are not lost in my proposal, nor would it be hard to collect and distribute games played with this genre in a normal database file and explain its nuances to the general public without having to go through the basics of the starting position each and every time.  

thechamp's picture

Sorry, but boring proposals. Capablanca, Fischer and everyone advocating randomness of sorts is trying to kill classical chess. Possibilities are endless in chess - no need for any adjustments to the starting position. Why not leave this old and still vibrant game alone.

Rudy's picture

Moving a black pawn one or two square from it's starting square (at random or not) doesn't give rise to a new kind of chess. E.g., the starting position with a black pawn on a6 is the same position with colours reversed after 1.a3 in a normal game of chess. So, this instead limits one's options. A similar argument holds if black places his knight on a6.
Thus this doesn't look like the way forward for our game of chess.

st32's picture

I would have been the first to mention this if the comments were not switched off :)

Arne Moll's picture

I did realize the situation was the same as after 1.a3 with Black to move: the reason why I kept the pawn on a6 was to preserve the comparison with the St. George Defense.

Not sure how this would limit one's options. That's the same as saying that in Chess960, having a bishop on b1 instead of c1 'limits one's options' because you can't put your bishop the a1-h8 diagonal anymore!

With a black pawn on a6, 1.e4 c5 would surely look like a Sicilian but it would not be the same, thus killing all existing Sicilian theory and providing a new basis for fresh ideas. It might turn out that 2.b3 is now the best move, or that 1.b3 is a better way of dealing with the situation. It would take decades to establish this. It sounds pretty exciting to me.


Rudy's picture

The suggestion of putting a black pawn on a6 limits the number of reachable positions compared to the normal configuration of pieces. Thus, in a sense, the complexity of the game decreases and there is less room for creativity.
But this problably isn't the reason why you suggested the starting position should be adjusted. As I understand it, there are two reasons. In the first place, it renders established opening theory useless (well, not totally, since there still are ways to transpose to known theory). Second, it enhances our understanding of the game. It would be beneficial to know exactly how a different starting position influences winning chances (or drawing chances for some of you out there!) and then use this knowledge in a normal game. Come to think of it, would Adorjan still say that black is OK if the pawn is on a6? In a way, black has lost some of his reactive possibilities. I'm guessing that a hedgehog with a tempo up for black is disastrous, since he can't put the tempo to use (and this may hold for other sicilians as well).
But certainly, i can't disagree it isn't exciting, but it does look a bit artificial.
As for chess960, should we not play chess960, because a GM blundered a piece in the first couple of moves?

adam's picture

interesting proposal; however, imho chess960 is a more promising alternative, it would just need _much_ more support from players, sponsors as well as fans... for instance, although world champions have been crowned multiple times by now, having no database in an age with live rating sites updated several times a day clearly describes the situation
i want to ask something that may be trivial: what is lasker-lasker (1924)?

Levon's picture
adam's picture

thx, very nice game. must have heard about it a long back, but couldn't recall

Kenneth W. Regan's picture

At Hans Bodlaender's, I'm on record as favoring this form of "Fischer Non-random":

With 960-squared possibilities, probably a few thousand really meaningful ones, it would really set back "theory" 200 years!

thechamp's picture

It's interesting that Arne Moll wants to "kill all excisting sicilian theory". The openings - their names and the long tradition of developed theory - is a living cultural treasure. Think again Arne!

Arne Moll's picture

Thechamp, I think you misunderstood me, or maybe I didn't make it clear enough: I love our game as it is and I think there's plenty of room to avoid boring theoretical duels without doing anything to the starting position!

But some think more radical measures are needed, such as Fischer Random. This article is mainly written for them: my position is that you don't need Fischer Random to avoid existing opening theory: if you want to get rid of current opening theory, then all it takes is making minimal changes to the starting position.

Dan's picture

Arne, I've read chess websites for a decade and never commented because I've never felt strongly enough. But this idea of yours is beautifully brilliant and I think it should be the next step in the evolution of chess. Like many such ideas, it seems destined for a long time to be misunderstood and unappreciated by many, possibly even the majority. Also like many profound ideas, it is deceptively simple, but this is a large part of its brilliance. As you rightly point out, the "minor" alteration becomes increasingly significant as one moves up the rating ladder - to GMs, it's a monumental difference. And yet, unlike F-Random, it preserves the essence of the game. Fischer Random's flaw is that it's too wildly different, as you point out. As its name and creator remind us, it's random (incoherent, meaningless), and therefore disrupts in too violent and shocking a way the inner coherence and logic of chess that is its essence. A game perhaps appropriate only for Bobby Fischer himself, or someone of his inner chaos and insanity. If only we could stop idolizing far-and-away the single most insane and dangerous of chess genius, we may be more receptive to good ideas.

Many amateurs won't appreciate the idea because they won't think it's a major difference. They like Fischer Random for that reason. But the truth is a pawn on a3 or a6 is a monumental difference. Some of those who bemoan the dying of chess by opening theory, in my view, are plain dishonest with themselves. They laud themselves as ultra-creative as a defense mechanism to defend bruised egos. Their problem isn't really with opening theory, it's that they lack comprehension, may be a bit lazy (or frustrated with past attempts) and, yes, may lack creativity compared to better players. Wanting to "invent" from move 1 is not a sign of brilliance or creativity, people! Like some spoiled child who slaps paint on paper and wants to be praised a brilliant artist, they want to be appreciated as creative geniuses without doing any work or respecting the history of the game. In what other field - math?, science? - do we praise people who want to invent everything anew, without absorbing the body of material collected by humanity first? Most theoretical chess opening lines leave us in early mid-game positions that are unclear, with many possibilities reflecting different styles and values. That's where the limitless creativity kicks in, and if you listen to any GM review his or her games you won't help but be filled with an appreciation for his/her creativity. Do some opening lines lead directly to equal endgames? Sure. The exception proves the rule.

I am co-owner and Director of the Chess Club of Fairfield County, With bias but also good reason, I believe we are the best chess club in the world. We are a 4,000 square foot building, newly renovated, dedicated 7 days/week to chess. We do not rely on philanthrophy - our funding is from the chess community. We'd be happy to have a "St. George's" tournament or other event with your idea as the centerpiece. Feel free to contact me at anytime to discuss! Thanks again for the great idea, Arne. Don't be deterred!!

Alfonso's picture

A "pedantic" note: Skara 1980 was not a chess Olympiad, was an European Team Championship. The 1980 Olympiad was at Malta.

Arne Moll's picture

You're right of course, thanks. Corrected.

patyolat's picture

Suppose that Fischer Random was originally invented first. How do you think people would react if someone suggested that 959 possible starting positions should be discarded and only one used from now. They correctly would pont out that whole games could be calculated by a computer than memorized by a player, and "played" it in a tournament. Fischer considered this cheating and he definitely had a point. Unfortunately people as we know like the status quo, and insist to play only that one old boring starting position.

By the way I prefer to call it Fischer Random, since I believe the inventor deserves that his name is attached to his invention. We don't call Rubik's Cube Cube54 or Cube4325200327448985600 either.

Arsen Babayan's picture

"We can simply play the above position for the next, say, 600 years"...

And I think this is the error in author's theory which triggers it totally void. 600 years? Seriously? I agree that it took 600 years for the modern chess theory to emerge and get to where it is now. But do not forget, that all the strategic and tactical theory is still out there and even if we completely disregarded the role of computers in modern-day chess theory, it would take no more than 100 years for the opening theory for a new starting position to grow and overgrow. In this computerized era it would take 5-6 years, let it be 10, to get another starting position analyzed and published in all details, while you would never do that for 960 possible positions of Fischer's chess. Fischer's idea was "show me you can play CHESS". By switching between starting positions you do not fulfill Bobby's desire, and if a super-GM blunders a piece on 10th move playing Chess960, sorry, then there are only two explanations for that: either the guy is a theory-freak or that's just an accident, which happens sometimes and the author's suggestion doesn't solve any of those in any way.

Another major drawback for this is that it's very difficult to find another position in line with author's idea which would not be of big... if not decisive advantage to either side. Black pawn to a6 - and Black is much better, as the author correctly mentioned. But the alternative (knight to a6) is a disaster for black - not only they are deprived of very important defensive piece on c6 or d7, they basically will have to play a piece down most of the game, as 1.d4 will become an automatic first move for white taking c5 square from the knight, let alone all possible captures Bf1xa6 by destroying black's queenside pawn structure.

So briefly, I think this is just another "nice try". In my honest opinion nothing has yet been suggested to even compete with Fischer's idea of revolution.

Mattovsky's picture

The idea is not really new. I can't find the source right now, but Dvoretsky made a very similar suggestion years ago.

chandler's picture

Yes, he made it in his chesscafe column a year or two ago; Arne maybe you should have a look at it. His motivation is also that positions should be understandable, but he explains it better and in more detail. And I think his suggestion is different... moving just a pawn or something like that. Please search for it and give a link :)

GeneM's picture

Dvoretsky discussed FRC-chess960, and his own alternative, on 2008/Jan, in "Polemic Thinking, Part Two", at...

Yes, we need to discard the 'Random' from Fischer Random Chess.
But No, not the way this ChessVibes column says to.

I believe that one of the 959 non-traditional setups should be annointed for the next couple decades, so that grandmasters and amateurs, both armed with Fritz and a creative spirit, could develop a whole broad & deep opening theory for the new setup, to rival the depth and breadth already achieved for the traditional chess1 setup.
Watching that new theory grow from nothing would be facinating. It would also be educational in ways we cannot foresee.

But what should the new setup be? I recommend the following, which after 1. e3 e6 in the traditional setup, can be reached in nine more legal move-pairs:


Importantly, S#549 has no corner bishops, and the two white knights start on the same shade of square. Positions which lack either of these two characteristics should not be considered at this time.

I was redirected here from... , its entry dated 2011/Nov/05.

GeneM (2011/Nov/05)
CastleLong .com

Arne Moll's picture

Sounds like a sensible suggestion. For me, the most important condition is that this position is fixed for at least a couple of decades, allowing theory to develop and games to be analysed properly by a large amount of people, rather than just a few specialists.

GeneM's picture

Fritz_13 has a new cloud-derived collaborative analysis feature named "Let's Check". This feature could have a tremendous effect on the rate of growth for opening "theory" for one new stable reused chess960 position such as S#549. The whole chess playing planet could contribute to a central repository of analyzed variations of all the not-yet-discovered opening systems for S#549.

Seems like the "Let's Check" feature is a significant new reason to consider adding reuse of S#549 to the long existing reuse of the traditional setup S#518.

harryo's picture

I think that the 1.a3 idea is too limiting for the future of chess. What would be really wonderful is to change chess competition so that players play three types of games in equal proportion in the really big global tournaments:

1) A traditional chess portion of competition
2) A fixed inter-generational FIDE assigned Chess960 start position portion of competition
3) A genuinely randomized Chess960 portion of competition

If we give these names it would be:
1) "Classical" portion of competition
2) "Fixed-position" portion of competition
3) "Fischer-random" portion of competition

Each of these three divisions has the following academic advantages:
1) Classical-start -> test of memorization transitioning into the mid-game
2) Fixed-start -> test of good research techniques transitioning into the mid-game
3) Fischer-start -> test of creativity transitioning into the mid-game

The big tournaments of the world would feature the elite players facing up in all these three forms of chess competition. The chess rating system would legitimize and recognize these three forms of play by giving them each an official ratings system.

Then in other parts of the world specialist tournaments would start to show up focusing on one of these disciplines depending on what the regional interest is. For example there are regions of the world right now that are highly focused on traditional chess. There are regions of the world willing to try fixed-start positions and there are regions in the world such as Germany that already have an exciting Chess960 culture.

I appeal to the chess world. Can Chess players of the world finally accept that traditional Chess and Chess960 are not mutually exclusive, but actually complement each other? Please I ask that the chess world lessons it's black and white mentality just for a short time while we rethink the future of chess.

Thanks for listening

woolyworm's picture

Yasser Sierawan's chess variant, where a hawk ( a piece combining the moves of bishop and a knight) and an elephant (a piece combining the powers of a rook and a knight) are introduced on a vacant square on the back rank, maintains the patterns of the classic game while cranking up the voltage. Yasser's game gets exceedingly complex once the new men are placed, but there is no reason why the players can't agree to use the same mechanism to introduce another bishop, say, or any other (or any number of other) pieces during a game. Using this technique preserves the contours of the traditional game and the reliance on opening theory, though valuable, is considerably less so. Try it. You'll like it.

woolyworm's picture

Yasser Sierawan's chess variant, where a hawk ( a piece combining the moves of bishop and a knight) and an elephant (a piece combining the powers of a rook and a knight) are introduced on a vacant square on the back rank, maintains the patterns of the classic game while cranking up the voltage. Yasser's game gets exceedingly complex once the new men are placed, but there is no reason why the players can't agree to use the same mechanism to introduce another bishop, say, or any other (or any number of other) pieces during a game. Using this technique preserves the contours of the traditional game and the reliance on opening theory, though valuable, is considerably less so. Try it. You'll like it.

MatsW's picture

You should have a look at this, an alternative to Fischer Random where the pieces are manually (non-randomly) relocated:

MatsW's picture

By the way, I have already suggested a training variant, Chess256, close to Moll's suggestion:

Edwin's picture

From what I can tell watching the recent Tata Steel tournament wherein we saw the best players in the world square off, this talk of the death of chess is as premature now as it was in Capablanca's day. In many of the Tata games the top GMs got in trouble in the openings.

T's picture

Pick 1 960 position and use that for the next 10 years, then pick another and so forth ;)

Honestly I think normal chess has a lot of life still though.

Anthony's picture

I've stumbled across this interesting discussion, and like to add something. First, I should say that I'm not much more than a casual player - I did play for the school team, and I played a bit of club chess later, but if I tell you that I own only two chess books - a 1974 "Ideas Behind the Openings", and Fischer's 60 Memorable Games - you'll know that I'm pretty much a tyro in these computer-driven times. But I do love the game, and I think I agree with those who bemoan the fact that there's so much theory now that you can't even say who just won a game - was it you, or the sound variant you managed to remember move by move from a book? But I also understand why there's such a resistance to Fischer Random and similar variants. It's not "real" chess - it's not "classic" and so on. So here's a suggestion - daft or not, only you really serious players can say. But this suggestion has certain advantages; it's real chess - its' classic, so the purists should be satisfied. But it pretty much wipes out most - not all, and that's important - opening theory. You'd have to start again, which gives you another fifty years to build up theoretical lines! And, finally, I promise two things should any of you try this- and you should try this by playing half-a-dozen games with someone whose game you know well: one, that you'd be as frustrated as hell for the first two or three games - but then, your opponent would be in the same boat. And, two,that it would really test your ability to create - and - as Fischer put it - just play chess. This suggestion isn't even new- in fact, it's several hundred years old!! Let's call it Indian Chess. (Which is where the King's Indian, and Nimzo-Indian get their names of course - which is why I said that not all opening theory gets jettisoned), So what you do is revert to chess - classic chess - as it was played in India. That is, reverse the rule that allows pawns to move two squares from their original position. It was done to "speed up" the game of course, but maybe it's time now to slow it down again! By doing this, nothing about the game as we know it now gets chucked out. But just try this, and think of how little use opening theory is - and yet you're playing real, classic chess. For instance, suppose I happen to hate facing the Ruy Lopez: well, when I see e3 played by white, I can avoid any possibility of that opening by playing a6 straight away. Then my opponent has to think again - and really think again. It would need a mathematician to calculate how many combinations are possible in Indian chess after each player has made, say, two moves each - but whether it's a smaller or larger number than at present hardly matters, because without the book or the computer to do it for you - over the board, that is - you'd have to fall back on your native ability to play the game.
Classic Indian Chess: any takers?

Your comment

By posting a comment you are agreeing to abide our Terms & Conditions