Shall We Play Chess960?

My post on The Rampant Expansion of Theory quoted GM Gligoric on the inspiration for his chess960 book. It reminded me that ever since using Gligoric's title on my first post on the subject -- Shall We Play Fischerandom Chess? (*) -- I've wanted to devote one full post to the book. Having read it cover to cover twice and flipped through its pages many times, it has been my primary source of written inspiration on chess960.

A list of the book's contents on the Schachversand Niggemann page, Shall we play Fischerandom Chess?, tells us that the book is divided into three sections: a brief history of chess, the development of Fischer's idea, and sample GM games. Despite having been written and published in 2002, when there was little practical experience with the game that would later be called chess960 (when was it so named?), Gligoric's work still offers the best material on the subject.

Why did Gligoric write the book? He explains in the 'Author's Note'.

It was my pleasure to witness part of the process of creation of the randomized chess game, invented and formulated by the world chess champion [Bobby Fischer], who, having probably liked my interpretation of his match versus Spassky in my earlier book on Reykjavik 1972, at the time suggested the idea that I try to write a book myself about the unknown subject of this new version of the game of chess.

Like chess960 itself, the book has been largely ignored by the mass of the chess playing public. There are, however, a few reviews available on the web. Amazon.com currently has three Customer Reviews. One of these, by Gene Milener, author of one of the few other books on chess960, is another list of the contents. A second is more of an opinion on chess960 ('Fischerandom chess is unlikely to replace "classical" chess for much the same reason that the aluminum bats have not entered professional baseball: it would take a beautifully crafted game with a long tradition ... and throw it all out the window.') than on Gligoric's book. The third is unintelligible ('very useful for those who think clasical chess in diing into uncatchible theory+computer combo').

Of the reviews found on chess sites, the first chronologically is on Chessville.com (2002): Reviewed By David Surratt; 'If you are interested in FRC, or even just in chess history - buy this book.' The next is on Jeremysilman.com (2004; 'Shall We Play Fisher Random Chess?'): Reviewed by John Donaldson; 'If you have any interest in random chess you will want to get Gligoric's book.' The most recent is a discussion on a Chess.com forum (2010): Interesting Chess960 FRC books; Milener surfaces again and presents his real opinion, 'I ask you honestly whether anything in Gligoric's Fischerandom book says anything at all about Fischerandom?', followed by a few concepts from his own book.

My favorite section of the book is a postscript to the eight games played between GMs Leko and Adams at Mainz 2001. There are comments on Fischer's version by Leko, Adams, Yusupov, Bronstein, Kasparov, and Kramnik. Here's Kramnik:

I tried many different starting positions and all these were somehow very unharmonious. And this is not surprising as in many of these positions there is immediate forced play: the pieces are placed so badly at the start that there is a need to improve their positions in one way only, which decreases the number of choices.

It's a good point and would provide an interesting kickoff for another post.

The Rampant Expansion of Theory

In Who Is the 'Father of Chess960'?, I related an account of the creation of 'Fischerandom' as given by GM Svetozar Gligoric in his book on the subject. Gligoric had more to say about the subject in a recent post on ChessInTranslation.com: Analysing by the riverside with Bobby Fischer.

[In preparation for his 1992 match with Spassky], Bobby asked if I could play a training match with him. At first I didn’t want to, but I had to give in to his wishes. He was panicking about how theory had developed during his twenty-year absence from chess. That was why he came up with his own version of chess, where the starting position would be determined by the drawing of lots. And he began to torment me with persistent requests to write a book about it.

I told Bobby that I had very little information, but he wouldn’t let it go: "Write that book! You have to do it!" In the end I started to gather a few crumbs of material, and a few years ago a book on "Fischer Random Chess" was published in London.

There I wrote that "Fischer Random Chess" would never replace classical chess, but could exist in parallel with it. And I turned out to be right: there are now tournaments in Fischer Random Chess, and moreover great success has been achieved in it by the same players who play well in classical chess.

I don’t think that classical chess will ever die out. Capablanca feared the spectre of the "draw death" of chess, while Fischer feared the rampant expansion of theory. Perhaps a time will come when grandmasters can’t think up anything new in the opening, but then the struggle’s centre of gravity will shift to the middlegame, and the endgame. To a degree we can already observe a situation like that now.

Before seeing this, I hadn't known that Fischer had any connection with Gligoric's book. I should have guessed.

When Castling Undevelops a Rook

In Castling Patterns Visualized, I used a simple technique to show the different ways that R, K, & R can be distributed across chess960 start positions. The technique can be used to count how many positions exist where castling O-O-O or O-O is possible on the first move. Many newcomers to chess960 are impressed by this contrast to traditional chess (SP518 RNBQKBNR), where castling O-O is only possible after at least four moves, and castling O-O-O takes even longer.

Another curious pattern is where the rightmost Rook starts on the e-file. There are 102 such positions, shown in the following table (the second column is the count of chess960 positions that share that pattern).

R**KR*** 18
*R*KR*** 18
**RKR*** 18
R*K*R*** 12
*RK*R*** 18
RK**R*** 18

These patterns are special because castling O-O displaces the Rook from the central e-file to the off-center f-file, at the risk of undeveloping it. From the f-file, it might have to be moved back to its start square by Rf1-e1 (or Rf8-e8). When this happens, castling O-O actually loses a tempo. In cases where the Rook is already performing an important duty, the move O-O might even be too dangerous to consider.

This quirk doesn't happen with castling O-O-O. When the Rook starts on the d-file, as in the following patterns, the O-O-O move leaves the Rook in place, where it is often already active.

***RKR** 18
***RK*R* 18
***RK**R 18

Considerations like these make the castling move so important to chess960. That's why I always consider the castling options when evaluating a new start position.

Castling Patterns Visualized

Some time ago, in a post titled Introduction to Chess960 Geometry, I worked out the number of unique castling patterns in chess960. By 'unique', I mean a type of position where the King and two Rooks start on different files. It happens that there are 56 such patterns. Listing them in logical sequence gives the following picture.

RKR***** 18
RK*R**** 18
RK**R*** 18
RK***R** 18
RK****R* 18
RK*****R 18
R*KR**** 18
R*K*R*** 12
R*K**R** 18
R*K***R* 12
[...]
***R**KR 18
****RKR* 18
****RK*R 18
****R*KR 18
*****RKR 18

The number in the second column is the count of different chess960 positions that share that unique castling pattern. For example, there are 18 positions that have the 'RKR*****' pattern and 12 that have 'R*K*R***'. In the previous post I worked out why some castling patterns encompass 18 positions (there are 48 such patterns) and others only 12 positions (8 patterns). It has to do with positions where the R, K, & R all start on squares of the same color.

The same type of visual table can be used to show the number of positions where castling is possible on the first move. There are 72 positions where castling O-O-O is immediately possible:-

**RKR*** 18
**RK*R** 18
**RK**R* 18
**RK***R 18

And there are 90 positions where O-O is immediately possible:-

R****KR* 18
*R***KR* 18
**R**KR* 18
***R*KR* 18
****RKR* 18

Some day it might be interesting to examine patterns from real games and determine how many times the players castled O-O-O vs. O-O. For now, though, there are too few recorded games to make this worthwhile.

Top GMs and Traditional Development Patterns

After writing the post on Traditional Development Patterns, I searched for other examples on the same theme: start positions with 'RNBQK***' or '***QKBNR'. My collection of games from the Chess Classic Mainz series (CCM) yielded only a few examples, most of them from machine vs. machine games, but searching on the twins ('RNBKQ***' or '***KQBNR') yielded many examples.

The first batch of examples consisted of ten games played in CCM9 during the first round of the 8th FiNet Open. Because the event used the Swiss system for pairings, the highest rated players, all GMs, were paired against the top players from the bottom half of the rankings. This was similar to the situation I discussed in GMs vs. 2100-2200 (I) and GMs vs. 2100-2200 (II).

The second batch of games came from the eighth round of the same event, where the GMs played against each other. This reminded me of the post How Top Players Treat the Same Chess960 Position. The games played in the second batch used the start position shown in the following diagram. I featured one of these games in the post Grischuk - Mamedyarov, Mainz 2009.

SP535 RNBKQNRB

The initial moves show considerable variety, but are drawn from the most popular moves used in the traditional start position.

( 1.g3 g6 2.e4
    ( 2.c4 e5 3.Nc3 Ne6 4.d3 c6 5.Bh6 g5 {Grischuk - Mamedyarov} )
    ( 2.d3 e5 3.c4 Ne6 4.Nc3 c6 5.b3 Qe7 {Navara - Moiseenko} )
2...e5
    ( 2...d6 3.Ne3 Nc6 4.f4 Bd4 5.c3 Bb6 {Bologan - Buhmann} )
3.d3 d6 4.Nc3 Ne6 5.Bd2 Nd4 {Kamsky - Grigoriants} )

( 1.Nc3 g6
    ( 1...g5 2.g3 d6 3.b3 Nc6 4.Bb2 f5 {5.O-O-O Bd7; Azarov - Akopian} )
2.g4 e5 3.b3 Nc6 4.Bb2 d6 {5.O-O-O Be6; Movsesian - Sargissian} )

( 1.e4 g6 2.Ne3 b6 3.Nc3 Bb7 4.g3 Nc6 5.b3 e6 {Landa - Malakhov} )

( 1.f4 g6 2.g4 c5 3.Nc3 Nc6 4.d3 b6 5.Bd2 Bb7 {Naiditsch - Stevic} )

( 1.d4 g6 2.Be3 e5 3.dxe5 Qxe5 4.Nc3 Ne6 {5.O-O-O O-O; Nielsen - Zvjaginsev} )

( 1.c4 e5 2.g3 d6 3.d3 Nc6 4.Nc3 Nd7 5.Bg5+ f6 {Gashimov - Sebag} )

Other than releasing the Bishop on h1, I don't see any reason why 1.g3 should have been such an overwhelming favorite. It was chosen by four of the world's best players.

Traditional Development Patterns

An interesting class of chess960 start positions (SPs) has the first five pieces in the same order found in the traditional SP: 'RNBQK***'. Logic says that there should be four such positions. Since the White Bishop starts on a dark square, the other Bishop must be on one of the remaining light colored squares, f1 and h1. For each of these possibilities, the Rook and Knight can be switched on the remaining two squares. Searching the list of all 960 SPs identifies the following SPs as belonging to this class. (The twin of each SP is given in parentheses).

518 RNBQKBNR (534 RNBKQBNR)
519 RNBQKNRB (530 BRNKQBNR)
614 RNBQKBRN (246 NRBKQBNR)
615 RNBQKRNB (242 BNRKQBNR)

Using similar logic, there are four positions that fit the pattern '***QKBNR'. They are shown in the following table.

226 BNRQKBNR (631 RNBKQRNB)
230 NRBQKBNR (630 RNBKQBRN)
514 BRNQKBNR (535 RNBKQNRB)
518 RNBQKBNR (534 RNBKQBNR)

By definition, the well known SP518 RNBQKBNR appears in both tables. What makes these positions interesting is that a majority of the pieces can follow the development patterns known from SP518. Last year I had the opportunity to play one of these positions and the game evolved as shown in the following diagram.

SP514 BRNQKBNR
After 4...Bf8-e7

My opponent outrated me by several hundred points and I was surprised to see how he conducted the opening. While I followed classical opening principles, his moves fit into some other pattern of logic. It was one of the games that gave me the inspiration for two posts on this blog: Extravagant Openings in Chess960 and Extravagant Openings (cont.).

White's neglect of the center in the early moves -- 1.b4 d5 2.f4 Nf6 3.e3 e6 4.Nf3 Be7 (diagram) -- is striking. The game continued 5.Bb5+ c6 6.Bd3. I had already decided that the move ...c6 would be necessary before ...b6, in order to prevent the cramping b4-b5. When White forced this move by 5.Bb5+, I was convinced that I had won a tempo and was happy with the position. The game was eventually drawn after 66 moves.