'Reform Chess'

Pleased that I actually learned something from old rec.games.chess (rgc) posts -- see my previous blog post, Foraging the 'News Groups' -- I returned to the resource looking for more material on the early days of chess960. Here's a snippet by IM Mark Ginsburg from May 1993, titled Fischer (again).

Fischer met with the Polgars recently on the Yugoslavia / Hungary border. He attempted to negotiate a match when he would play Judit a variant of chess, known in some circles as "Reform Chess" - where the piece arrangement for both sides on the first rank is determined at random before the game starts (except that the Kings, and the Rooks, stay on their 'normal squares' to allow for eventual castling).

No doubt Fischer feels that Judit is a product of intense training (and by extension, memorization) and wants to 'prove' his superior talent. Whatever the case, the Polgars have rejected his initial overtures to a Reform Chess Match and they went back to Budapest - and I suppose he likewise retreated to a hotel room in Belgrade. One anecdote that came out of this meeting is that Fischer is now being asked to pay for his hotel room daily, in cash.

A follow-up post, also titled Fischer (again), but from another rgc-user, gave more details about Fischer's new idea.

While some may interpret this as Fischer being unwilling to play Judit straight-up, I'd counter that Fischer has been promoting the idea of randomized chess for some time. Also, during a [Fischer - Spassky] II press conference (perhaps it was the one before the match started?) Fischer was discussing randomized chess.

Randomized chess was experimented with early on by Aaron Alexandre (1766-1850) and later by Erich Brunner (1885-1938). Of course, Bird (1830-1908) and Capablanca (1888-1942) were also very big supporters of unorthodox chess to get away from mechanistic memorization of openings. Fischer is obviously of the same belief.

This squares with other facts that I've featured in earlier posts. See, for example, Pictures of a Fischer Random Precursor. As for the youngest (and strongest chess player) of the three famous Polgar sisters, Judit is on record as having said, 'To be honest, I never was a big fan of Fischer Random'; see KC-Conference with Judit Polgar.

Foraging the 'News Groups'

The heyday of rec.games.chess (rgc; a piece of a technology once known as 'news groups') was long ago, but the forum's posts through the mid-2000s are a valuable archive for the early days of the Internet. What can rgc tell us about the early days of chess960? For example, what can it tell us about the evolution of the 'chess960' name? I searched rgc for the first occurrence of the most popular chess960 synonyms.

'fischer random'

The two word spelling was the earliest reference I could find.

• 1996-04-14: Bobby Fischer Announces New Rules of Chess

April 1996 was after the first (tentative) formal mention by Fischer in November 1995, as I recorded in An Aborted Announcement, and the formal announcement in June 1996, Fischer Announces Fischerandom. There is not enough information in the rgc thread to identify what happened in April.

'fischerandom'

• 1996-06-02: The Week in Chess No. 85 • 'One person who always attracts attention when he appears is the long lost Bobby Fischer. He is about to appear in Argentina to promote his newchess game FISCHERANDOM with a press conference later this month.'

The spelling with one 'r' was Fischer's preferred spelling, but it didn't take long for the two 'r' version to appear.

'fischerrandom'

• 1996-07-25: Fischerandom rules?

That thread uses both 'r' and 'rr'spellings in posts by two different people. The chess960 name, which has since been adopted as the preferred name (except perhaps in the United States), emerged more than six years later.

'chess960'

• 2002-12-29: Wishes for the coming year 2003 • 'My computer chess related hope for 2003 is for more chess programmers to integrate the Fischer Random Chess (FullChess / Chess960) variant into their engines.'

The post was signed 'Reinhard Scharnagl', another of the early promoters of Fischer's greatest invention.

Queen Safety in the Center

'Do not bring your Queen out too early', as I once quoted in a post on Fine's 'General Principles' of Opening Theory. There are at least two reasons for this: (1) the best square for the Queen is usually not apparent until some other pieces and a few Pawns have moved, and (2) the opponent can develop more pieces harrassing the Queen at the same time.

While this is a tried-and-true principle of traditional chess, some chess960 positions leave room for more specific considerations. One such position that I encountered, SP679 QRBKNNRB, is shown in the top diagram. I played it as White in a pyramid game on Schemingmind.com (see Pyramids and Dropouts for an explanation of the event format). I had been playing chess960 for less than a year, my opponent, a top-10 player on Schemingmind.com, was the highest rated player in the pyramid, and I was curious to see if I could hold my own against him.

The initial moves were cautious, 1.b3 b6 2.Nf3 Ne6 3.Bb2 Nd6 4.d3 Bb7, as is often the case when the Queen starts in the corner. After the first contact between the forces, both players decided to castle O-O-O: 5.e4 f5 6.exf5 Nxf5 7.N1d2 O-O-O 8.O-O-O. This led to the position in the bottom diagram.

The game continued 8...g6 9.Bxh8 Rxh8, exchanging the dark squared Bishops. Now I wondered what to do with my Queen. I finally hit on the idea of 10.Qe5, placing it in the center. This might seem paradoxical, given that all of the pieces, except the Bishops that were just exchanged, are still on the board. Black, however, doesn't have many possibilities to harrass the Queen. The move ...d6 leaves the Knight en prise. If the Knight first moves away, or if Black moves the other Knight followed by a Rook lift to f5, White can retreat the Queen down the e-file.

The game continued 10...Rhf8 11.g3 a5 12.a4 Rf7 13.e4 h6 14.h4 Rdf8 15.Rde1 Nfg7 16.Nh2 d6, finally attacking the Queen, which retreated 17.Qc3. I eventually lost in the endgame, after declining my opponent's draw offer, but was happy with the course of the opening. I learned that there are chess960 positions where the Queen is quite safe in the center.

More on Displacement and Distance

Responding to my post on 'Bizarre Castling Rules'?, HarryO suggested examining the R***K**R positions (see Non-Random Chess960 Trial Game 9: SP864 after 4)...e6), e.g.

1) Do these 18 starts allow [full] Queenside and Kingside castling strategies or are they restricted in some way? [...]

with the further restriction that

The main culprits we should investigate are the six SPs with Knights on the same color probably starting with the four N*N SPs.

I had some difficulties to compare the different positions, so I created a chart in the same style I used in The KID Family Goes the Distance.

With the left Rook (zLR), King, and right Rook (zRR) fixed to their original squares, the list includes SP518, while the 'distance' (Dist) from that position to the other positions is typically low. It turns out that HarryO's suggested positions can be identified easily from a calculation based on the displacements of the pieces (zLR, zLN, etc.).

For example, the first position in the list, SP414, has the left Knight (zLN) displaced by one square and the right Knight (zRN) by three squares. The sum of these -- four squares -- is an even number; the Knights start on different color squares, because they start on different color squares in SP518. For the second position in the list, SP430, zLN = 0 and zRN = -3. The sum of these displacements (using their absolute values) is an odd number, so the Knights are on the same color square (N*N in SP430).

[By similar reasoning, the combined displacements of the Bishops must always be an even number. The Bishops are on different color squares in SP518, a requirement that holds for all chess960 positions.]

Of the 18 positions in the R***K**R family, HarryO is right in pointing out that only six have the Knights on the same color. That means twelve positions have Knights on different colors. Why the imbalance in the two cases? Because Knights on the same color mean there are fewer possibilities for placing the Bishops.

Another curiosity of the same type is the number of positions in the family where a Knight starts on the same square as SP518. There are eight positions where zLN = 0, but only six where zRN = 0. Why the difference? Because in R***K**R, the left Rook and the King are on the same color as the right Knight. This again limits the choice for placing one of the Bishops.

That's all nice to know, but it doesn't tell us anything about the dynamics of the different SPs. Some practical tests will be necessary to do that.

A Two Pawn Gambit for Black

Along with the game featured in my previous post, Dynamic Imbalances Are Instructive, I played another unusual game in the same LSS event. The start position, SP305 BNQBRKRN, is still visible in the top diagram, where my opponent opened with 1.b3.

Here I had the idea to sacrifice a Pawn by playing 1...e5, which my opponent accepted with 2.Qa3+ Be7 3.Qxa7. The game continued 3...Na6 4.Qe3 b5 5.Qh3, when I sacrificed a second Pawn with 5...O-O 6.Bxe5, reaching the bottom diagram.

What's the compensation for the two Pawns? Black is better developed and will gain further time by threatening the Bishop on e5 and the Queen. The Knight on a6 will come into play via c5 (less likely is b4), after which the move ...Qa6 will initiate an attack on the Queenside. The move ...f5 will activate both Rooks and threaten to advance into White's Kingside, supported by the light-squared Bishop on the long diagonal.

All of these objectives were met in the following moves: 6...Ng6 7.Bb2 f5 8.e3 Nc5 9.Ng3 Qa6. Black sacrificed two more Pawns with 10.Nc3 f4 11.Nh5 Ne6 12.Be2 fxe3 13.dxe3 Qa5 14.Bxb5 Bb4 15.Bxd7, when White was no longer able to cope with the pressure. Short of time, he blundered on the 19th move and promptly resigned.

My note to my first move said, 'Not sure it's sound, but it's definitely interesting!' That's one of the attractions of chess960. There are new ideas to be tried as early as the first move.

Dynamic Imbalances Are Instructive

Studying traditional chess often yields benefits in chess960. One obvious area is the endgame, where the old chess and the new chess are virtually indistinguishable. Another crossover occurred while I was working on the series Practical Evaluation on my main blog, in particular the topic 'material imbalances'.

I was playing White in a game on LSS that started as SP009 QNNBBRKR. One of the problems both players have is how to castle. I took a leaf from some ideas documented in a previous post -- Extreme Barbecues -- and decided I would lift the f-Rook up the f-file, thereby getting it out of the way in order to castle O-O. That explains my first moves, 1.f4 d5 2.e3 e6, reaching the position shown in the top diagram. Black has played a couple of normal developing moves.

Now I became attached to an idea that I couldn't get out of my head. It started with 3.f5. In my notes, I gave it '!?' and commented, 'Looks interesting, but is it sound? Willing to take the risk to find out...' My opponent played the obvious moves and the game continued 3...exf5 4.Rxf5 Nd6 5.Rxd5 Bc6 6.Rc5 b6, reaching the bottom diagram where the Rook is effectively trapped. Now I sacrificed the exchange with 7.Rxc6 Nxc6 8.O-O.

It looks like White is simply down the exchange with a single Pawn as compensation. According to IM (now a GM) Larry Kaufman's database research, this is more accurately calculated to be a 0.75 advantage for Black. As compensation, White has the Bishop pair (a 0.50 advantage for White) and two unopposed center Pawns (a 0.20 advantage for White). Since -0.75 + 0.50 + 0.20 gives Black a trivial advantage of -0.05, the material is almost equal. On top of this, both sides are approximately equal in development, with Black perhaps having a slight edge.

The game lasted until move 49, during which time the material imbalance transformed to a Bishop for White versus three Pawns for Black (plus other matched pieces for both sides). Although the result was a draw, the shifting dynamic imbalances were both fun to play and instructive.

'Bizarre Castling Rules'?

From 'Counterplay / Readers Respond', Chess Life (CL), July 2013:-

CASTLING IN CHESS960: ANOTHER APPEAL FOR SIMPLICITY

Chess960 is a noble, but flawed attempt to force players to start thinking from the very first move of the game. The biggest problem with Chess960 are the bizarre castling rules. For anyone not familiar with Chess960, consider the following, which is just one bizarre aspect of “castling.” Depending on the opening setup, when castling, the king can move anywhere from five squares to zero squares to minus one squares (yes, the king can actually move in the opposite direction than it normally would). It would be difficult to teach this maneuver to anyone not familiar with standard chess. A variant called Chess480 seeks to simplify these castling rules, but in doing so creates some of its own issues.

I propose a variant which achieves the goal of eliminating memorization of openings while avoiding the failings of both Chess960 and Chess480. This variant, which I have dubbed Chess18, has a randomized opening setup just like its “predecessors.” The difference is that the rooks and the king start on the same squares that they do now so that castling remains exactly the same as it is now— problem solved!

An additional benefit of Chess18 is that it avoids the situation in Chess960 where with some opening setups White can attack an undefended black pawn with her first move.

When Bobby Fischer met with former FIDE President Kirsan Ilyumzhinov to propose the switch to Chess960, Ilyumzhinov advocated “step-by-step” changes mindful of the heritage of chess. Well, here is such a step.

David Couture, via e-mail

CL also provided a response:-

Chess Life asked Damian Nash, a two-time Utah state champion who ran small Chess960 tournaments at the U.S. Opens in 2010 and 2011, and also conducted small break-out sessions on the topic, to reply:

David Couture hits the nail on the head. Bizarre castling rules are a serious problem with Chess960 (Fischer Random). His solution is novel: Leave the rooks and king on the same squares as classical chess, thereby keeping familiar rules intact. Chess18 is a logical first step toward the evolution of the world’s greatest game, expanding opening books by a factor of 18. Another interesting alternative is “Moab Random,” a form of pre-chess that replaces castling (already a bizarre move in classical chess) with the much simpler ‘evacuation’ of the king to any empty back-rank square.

Kudos to Mr. Couture and other game theorists who attempt to wrestle chess out of the grip of the brilliantly obsessive memorizers at the top, who hold Ph.D. equivalents in opening theory. Consistent with Bobby Fischer’s hope for the future of the game, Chess18 could help return chess to the vastly larger audience of brilliant tacticians and strategists worldwide; at least for a little while, until opening jargon catches up. In classical chess, opening experience usually trumps raw talent. But in ChessX, as X increases, natural ability and sound strategy will yield progressively better results.

What do you think? Are 'bizarre castling rules' the 'biggest problem with chess960'?