29 November 2009

Learn to Castle Correctly (Mechanics)

It's the sixth and last round of a SchemingMind.com chess960 dropout tournament (see Pyramids and Dropouts for more about these events) held a few years ago. The starting field of 85 players has been whittled down to 10. These are the survivors, the crème de la crème at Scheming Mind, battling for the annual chess960 championship at one of their favorite online play sites.

After a few days play, a game starting NQRBBKNR (SP217) and continuing 1.Nf3 c5 2.c4 Nb6 3. b3 Nf6 4. e4 O-O reaches the following position


(After 4...O-O)

Now White timidly asks Black, 'Can you tell me, please, how can I castle? It's enough to move the king to g1 or there's something else I should do?'. Black's first reaction must have been, 'What the...? How did this guy survive the first five rounds, two games per round, without knowing the rules of castling?' The second reaction might have been, 'Now I get it! He doesn't know how to execute the castling move on Scheming Mind', followed by the same refrain, 'How did this guy survive the first five rounds without knowing the mechanics of castling?'

I looked at White's previous games in the tournament and verified that he had never castled. Game after game, with the King stuck in the center and a Rook out of play, White must have been an exceptionally good defensive player to have faced increasingly skilled competition without ever castling.

When castling in traditional chess, most software works the same way. You click on the King, then click the square located two squares toward the castling Rook. The software understands your intention exactly and moves both the King and the Rook to their castled squares.

In chess960, the castling intention is not so straightforward. Sometimes both pieces move, sometimes only the King moves, and sometimes, most illogically for traditional chess players, only the Rook moves. How to tell the software that, for example, clicking on the Rook this time means castling, but next time means a normal Rook move? There is no standard way to do this and the mechanism for castling depends on the software being used.

Before you play chess960 using new software or on a new site, make sure you know how to execute a castling move. Take the time (for once?) to read the instructions. Later in your games, it will save you much defensive maneuvering and perhaps a little embarrassment.

28 November 2009

Learn to Castle Correctly (Rules)

Here's a war story from the fourth game of a match between two super grandmasters at Mainz 2004 -- Gerling Match : Chess960 World Championship : Svidler vs Aronian. A problem occurred in the following position.

Chess960 World Championship, Mainz 2004
Aronian, L.

Svidler, P.
(After 32.e5-e6)

The story as told on the Chesstigers.de page involved an error in castling O-O-O.

This position caused quite a stir in the fourth game of the Gerling Match between Levon Aronian and Peter Svidler. Levon Aronian took his rook on a8, jumped over the King and put it on d8. However, he forgot to take his King which had to be placed on c8 to fulfill all the requirements for a-side castling in Chess960. If he would have done that, the game would be over: Qa8 check and mate!

Peter Svidler stopped the clock and arbiter Sven Noppes came to see what was going on. Castling was possible, the mistake Aronian made was that he took his rook first, so he had to castle a-side.

Peter Svidler showed that he is a gentleman on the board and allowed Aronian to correct his move. The Armenian had to play Ra7 and kept the better position. However, after this faux-pas Aronian completely lost control and finally lost the game.

Was this just a mistake by the young Armenian, or did he try to win with a dirty trick? Aronian joked during the press conference : "Off course I wanted to cheat, just like in his game against The Baron in the computer match." Peter Svidler said: "I do not have a problem with Aronians mistake, and when you have only 40 seconds on the clock, like in this case, it is possible that you mix up the castling rules".

The complete PGN game score (SP725, RKBBNQNR), again courtesy Chesstigers.de, is:

[Event "CCM4 - Gerling Match (Chess960 WCh)"]
[Site "Mainz"]
[Date "2004/8/6"]
[Round "4"]
[White "Svidler"]
[Black "Aronian"]
[Result "1-0"]
[FEN "rkbbnqnr/pppppppp/8/8/8/8/PPPPPPPP/RKBBNQNR w HAha - 0 1"]
[Input "DGT1805"]

1.e4 e5 2.f4 exf4 3.Qxf4 Ne7 4.Ngf3 d6 5.d4 f5 6.e5 Nd5 7.Qd2 Nb6 8.Nd3 dxe5 9.dxe5 g6 10.a4 a5 11.b3 Ng7 12.Be2 Ne6 13.Bb2 Nd5 14.O-O h5 15.Nf2 Nb4 16.Bc4 Qh6 17.Qc3 Be7 18.Rad1 Nc6 19.Nd3 h4 20.Bb5 Ncd8 21.Nf2 g5 22. Bc1 Qg7 23.Nd4 Bb4 24.Nxe6 Nxe6 25.Qf3 Bc5 26.Kh1 h3 27.gxh3 Nd4 28.Qd5 b6 29.Rxd4 Bb7 30.Bc6 Bxc6 31.Qxc6 Bxd4 32.e6 Ra7 33.Ba3 Qe5 34.e7 Bc5 35.Nd3 Qxe7 36.Nxc5 bxc5 37.Bxc5 Qe5 38.Qb5+ Ka8 39.Bxa7 Qxb5 40.axb5 Kxa7 41.Rxf5 Rxh3 42.Rxg5 Kb6 43.Rg3 Rh4 44.Rc3 Rh5 45.Rc6+ Kb7 46.Rc4 Rxb5 47.h4 Re5 48.Kg2 c5 49.Kf3 Kc6 50.Kg4 Re2 51.h5 a4 52.bxa4 Kd5 53. Rc3 Kd4 54.Rh3 Rxc2 55.h6 Rg2+ 56.Kf5 Rg8 57.h7 1-0

Not all opponents would be as forgiving as GM Svidler. Learn how to castle correctly in chess960. It's not that difficult!

22 November 2009

Grischuk - Mamedyarov, Mainz 2009

In More from Mainz 2009, I gave a partial transcript from an interview -- Grischuk post-FiNet win -- on Chessclub.com, and promised to feature a game from winner Grischuk in a future post. During the interview Grischuk mentioned his game against Mamedyarov, so it looks like a good choice.

The start position (SP535, RNBKQNRB) is not too far from a traditional chess game. The royalty is in the center and the Queenside pieces ('a-side' for the purists) are in their traditional places ('RNB'). Only the Kingside pieces ('h-side') are scrambled. The first four moves were 1.g3 g6 2.c4 e5 3.Nc3 Ne6 4.d3 c6, bringing the game to the following position.

8th FiNet Open, Mainz 2009
Mamedyarov, S.

Grischuk, A.
(After 4...c7-c6)

Here is an explanation, as I see it, of those first moves.

  • 1.g3: Opens the diagonal for the Bh1. The move 1.g4 is an alternative, but as Grischuk explained in the interview, he's wary of the b4/g4 ideas.
  • 1...g6: The same thinking as for 1.g3, from Black's point of view.
  • 2.c4: Prepares the development Nb1-c3 and takes control of d5, an important central square. The move is reminiscent of the English opening in traditional chess.
  • 2...e5: Prepares the development ...Nf8-e6 and takes control of d4, also an important central square. The move is reminiscent of an important counter to the English opening in traditional chess (1.c4 e5).
  • 3.Nc3: Follows through the previous move and develops the Knight to its best square.
  • 3...Ne6: Ditto.
  • 4.d3: Prepares the development of the Bc1 on its long diagonal.
  • 4...c6: Counters the influence of the White Bishop on the long diagonal and prevents a White piece from moving to d5.

So far, both players have developed their pieces in classic style by developing a Knight to a good square, opening a diagonal for a Bishop, and paying attention to the center. Now White probably considered what to do about further development. The move 5.Ne3 forces the Bishop on c1 to go eventually to d2, where it blocks the Queen from developing. The move 5.Be3 forces the Knight on f1 to go to d2, where it also blocks the Queen.

This thinking leads to the idea 5.Bh6, where White has to calculate the tactical consequences of the Bishop trap 5...g5. After these fifth moves for both players, the game continued 6.h4 gxh4 7.Be4 Bf6 8.Ne3 hxg3 9.O-O-O. White is temporarily two Pawns down, but Black is lagging badly in development on the Queenside.

Since this is a chess960 blog, and the only real difference between traditional chess and chess960 is the opening phase, I'll stop here. White has castled O-O-O, while the Black King will find shelter on c7. Here is the complete game score...

[Event "CCM9 - 8.FiNet Open"]
[Site "Mainz"]
[Date "2009.07.31"]
[Round "8.1"]
[White "Grischuk, Alexander"]
[Black "Mamedyarov, Shakhriyar"]
[Result "1-0"]
[Variant "chess 960"]
[SetUp "1"]
[FEN "rnbkqnrb/pppppppp/8/8/8/8/PPPPPPPP/RNBKQNRB w GAga - 0 1"]
[PlyCount "97"]
[EventDate "2009.??.??"]
[EventCountry "GER"]
[Source "Chess Tigers"]
[SourceDate "2009.01.01"]

1.g3 g6 2.c4 e5 3.Nc3 Ne6 4.d3 c6 5.Bh6 g5 6.h4 gxh4 7.Be4 Bf6 8.Ne3 hxg3 9.O-O-O d6 10.fxg3 Bg5 11.Rh1 Bxh6 12.Rxh6 Qf8 13.Nf5 Na6 14.Qf2 Bd7 15.Rdh1 Nac5 16.b4 Na6 17.a3 Kc7 18.Kb2 Rd8 19.Qxa7 Rb8 20.Rxh7 Ng5 21. R7h5 Bxf5 22.Bxf5 Ra8 23.Qe3 Qe7 24.Rc1 Kb8 25.g4 Nc7 26.d4 f6 27.d5 c5 28.b5 Ne8 29.Rh2 Qc7 30.Ne4 Nxe4 31.Qxe4 Qa5 32.Qe3 Ka7 33.Qc3 Qd8 34. Rch1 Ng7 35.Bc2 Ne8 36.Qf3 Qa5 37.Rh7 Kb6 38.R1h4 f5 39.Rf7 fxg4 40.Qg3 Qd2 41.Rhh7 Ra7 42.Qh4 Qd4+ 43.Kb1 Qg1+ 44.Ka2 Nc7 45.Qe7 Nxb5 46.Qxb7+ Rxb7 47.Rxb7+ Ka5 48.Rxb5+ Ka6 49.Rhb7 1-0

...courtesy of Chesstigers.de. For an explanation of the PGN specific to chess960, see my previous post Chess960 PGN.

21 November 2009

Who Is the 'Father of Chess960'?

In The First Recorded Fischerandom Game?, I noted that the 2009 edition of 'The Sorcerer's Apprentice' by Bronstein & Fürstenberg mentions that 'Fischer Random is a derivative of an idea that Bronstein himself has invented twenty years earlier'. That quote is from a Google translation of a Dutch review of the book and not a quote from the book itself. Whatever the book says will remain a mystery to me until I get my hands on a copy. In the meantime, I can still explore the idea of who is responsible for the invention of Fischer Random, better known on this blog as chess960. The title 'Father of Chess960' hangs on the outcome of the research.

The ChessVariants.com page on Fischer Random Chess says,

Fischer Random Chess is somewhat similar to the older Shuffle Chess, or Prechess (or other related variants), yet has a unique style of its own. An extensive introduction and history of the game was written by Eric van Reem.

The van Reem article, The birth of Fischer Random Chess, is available on multiple web sites. It dates the 'birth' to Buenos Aires 1996, the event that I covered in Fischer Announces Fischerandom. Gligoric, in his ground-breaking 'Shall We Play Fischerandom Chess?', gave a more detailed account. After mentioning Capablanca's erstwhile efforts to improve the game that he felt was under threat from draws, Gligoric wrote,

Fischer's concentration on this problem went much deeper. During his stay in Saint Stefan [better known outside Serbia as Sveti Stefan] in 1992, he recommended shuffling all the pieces at random on the back row before the beginning of each game. [...]

It turned out that Fischer's first plan would make 2400 different commencing positions. Immediately after his sensational return to the chess scene in 1992, he began experimenting privately with this kind of chess against colleagues and chance visitors. Although Fischer was pleased that the mathematical sum of starting positions was very large, he soon discovered that eventually having two Bishops of the same color made an unpleasant impression, producing one-sided and limited opportunities on the chessboard. It was also clear that obstructing the right to castle would mean a step backward toward the primitive distant chess past and if the intermingling of pieces on the back row made castling impossible for both sides, this would inflict irreparable damage on playing strategy as well.

The result of Fischer's constant meditation on how to give alternative life to the game of chess, threatened by the exhaustion of its creative resources, was the formulation, in September 1993, of the rules of "Fischerandom Chess" which are quoted in a separate chapter of this book. [p.36]

I covered those rules in Fischer's Rules of Fischerandom.

Prechess, mentioned above in the ChessVariants.com quote, is a variant of shuffle chess where the players take turns placing single pieces on the back rank. Since the rules don't appear to have been formalized anywhere, restrictions on the Bishops and possibilities for castling depend on who is playing and what was agreed beforehand.

Who is the 'Father of Chess960'? There are three key requirements: (1) shuffle chess with the White and Black pieces starting opposite each other; (2) the two Bishops for each side on different colored squares; and (3) castling into the familiar O-O and O-O-O positions. Other would-be inventors might have proposed (1) and (2), but Fischer proposed (3). That last contribution, which steers the game from a shuffled variant into positions fully reminiscent of traditional chess, gives Fischer the title.

15 November 2009

A Chess960 Catastrophe

Talk about opening disasters! The diagram shows the position after three moves each in a game between two 2700+ players from Mainz 2005. White played 4.Nc5, threatening the royal fork on d7 (plus the small matter of a smothered mate). However Black responds, White has 5.Qb5, reinforcing the fork and adding mate on b7 to Black's woes.

4th FiNet Chess960 Open (2005 Mainz)
Bacrot, Etienne

Aronian, Levon
(After 3...Bd8-f6)

The start position was RKRBNQBN (SP941) and the initial moves were 1.e4 e5 2.Nd3 Ng6 3.f4 Bf6. Black's last move led directly to disaster, but the preceding moves 1...e5 and 2...Ng6 were probably not the best either. After 1.e4, the CCRL engines (see the 'Resources' link in the sidebar) prefer 1...f5 and 1...Ng6. Is the position after 1...e5 definitely lost? Perhaps the losing move was 2...Ng6; more analysis required.

14 November 2009

Setting up a Chess960 Position

In my previous post, The First Recorded Fischerandom Game?, I wrote, 'For reasons that I'll give some other time, the order by value is my preferred method of setting up a chess960 start position.' By that somewhat cryptic statement, I meant that I use a little procedure -- an algorithm -- to set up the pieces at the start of a chess960 game. That procedure involves placing the pieces in descending order of value.

Setting up the pieces to start a traditional chess game (RNBQKBNR) doesn't take any special effort. Experienced players do this so often that the pieces seem to find their correct start positions without any conscious effort by the players. The fingers grab a few pieces and scatter them on the correct squares the way a coin sorter sorts a bag of random coins; they just drop into the right slots. Sometimes a Bishop and Knight, or a King and Queen, might be switched accidentally, but this is usually spotted before the first move is made.

Setting up a new chess960 position is not so automatic. The pieces are almost always placed according to instructions that differ at the start of each game. It takes a conscious effort to place each piece on the correct square, and even then, there is always a chance that the pieces might not have been placed correctly.

A long time ago I discovered a method of placing pieces on a chess board when setting up a position other than the traditional start position. This is required, for example, when working from a diagram in a book or magazine. It turns out that the same basic method works for chess 960. I described the general procedure for traditional chess in an article I wrote for About.com: Improve Your Middle Game (Part 1 - Patterns).

Here's another trick I use frequently. It's a procedure for setting up a position on a board. First, clear the board. Don't try to set up a position by adjusting the pieces already in place unless the old position is almost identical to the new. Second, place the two Kings on the board. Third, set up the Pawns. Then add the Queens (both White and Black) if they are present, then all the Rooks (White and Black), and finally all of the minor pieces (ditto).

'What's the big deal?', I hear you asking. 'What difference does that make?' Perhaps no difference whatsoever, except that it works for me.

Setting up the Kings first tells me immediately where the most important pieces on the board are located. Are they on their original squares, on the same side, on opposite sides, or in an unusual place?

Setting up the Pawns without the other pieces gives me a quick picture of the Pawn structure. Does one side have a numerical advantage? Are there any classic weaknesses like doubled or isolated Pawns? How many islands are there? Since the Pawn structure changes very slowly, it's often the key to devising a long term plan. This is one of the things Philidor meant when he said, 'The Pawn is the soul of chess.'

Setting up the Queens, then the Rooks, then the minor pieces gives me another quick count on the material. Is there an advantage? An imbalance? How do the minor pieces match? Does one side have two Bishops and a Knight where the other side has two Knights and a Bishop?

By the time I've set up the position, I've already registered a lot of information about what's happening on the board. This makes up for the lack of information from not having played the game from the starting moves.

That same procedure, with a couple of tweaks, also works well for a chess960 start position. The first tweak is that the board doesn't have to be cleared completely. Moving the pieces to the center of the board (or off the board) is even more efficient. The second tweak is that the Pawns don't have to be cleared; they can be placed immediately on the second and seventh ranks, just like when starting a traditional chess game.

After that, however, the pieces are placed in order of their value: first Kings, then Queens, then Rooks, then minor pieces. The advantages to this procedure are:

  • Placing the Kings first helps to see that the most important piece is not placed in the corner, where it is never allowed to start the game.
  • Placing the Queens early allows to see where the most powerful piece on the board is located.
  • Placing the Rooks early provides a quick visual check that the King has been placed between the two Rooks. As a bonus, the castling choices are more obvious when there are no minor pieces on the board.
  • Placing the Bishops at the same time provides another visual check that they are starting on opposite colored squares.

And that's it. On top of these visual checks for the individual pieces, placing the pieces of the same value at the same time for both colors helps to ensure that the right pieces start opposite each other.

I've also discovered that this procedure makes it easier to set up the pieces when I'm sitting on the Black side of the board. I just take my instructions, QRKNBBNR for example, turn them upside down (RNBBNKRQ), and place the pieces using the same method I just described. It's not flawless, but it works much better than placing the pieces randomly.

08 November 2009

The First Recorded Fischerandom Game?

Between Fischer's announcement of his variant -- TWIC 88 (24 June 1996), Fischer Announces Fischerandom -- and the collapse of his kickoff match -- TWIC 91 (14 July 1996), Fischerandom's Inauspicious Start -- TWIC carried another item on Fischerandom.
13) Fischerandom chess game between Bronstein and Douven. Tom Furstenberg reports: IGM David Bronstein - IM Rudi Douven. This game was played on the 1st board during a friendly match on 30 boards between Chess Club Anderlecht of Belgium vs. Eindhovense Schaakclub of The Netherlands on 30.06.96. The time limit used was 1 hour 45 minutes for the first 40 moves, thereafter 15 minutes for the rest of the game.

The initial position for this game of Fischerandom Chess was White: Kd1, Qa1, Rb1, Rg1, Be1, Bh1, Nc1, Nf1. Black: Kd8, Qa8, Rb8, Rg8, Be8, Bh8, Nc8, Nf8

1.d4 d5 2.Nd3 Nd6 3.g3 e6 4.Bb4 a5 5.Bxd6 cxd6 6.a4 Qa6 7.Nd2 Nd7 8.b3 Nf6 9.c4 Bd7 10.e3 0-0 (The black King goes to g8 and the black Rook to f8) 11.Ke2 g6 12.Rgc1 Rfe8 13.Qa3 e5 14.dxe5 dxe5 15.b4 e4 16.Nc5 Bg4+ 17.Ke1 Qd6 18.cxd5 Qxd5 19.bxa5 Qh5 20.h4 Bf5 21.Nxb7 Qg4 22.Bg2 Nh5 23.Nf1 Be6 24.a6 Be5 25.Nh2 Qf5 26.Nd6 Bxd6 27.Qxd6 Qa5+ 28.Kf1 Rxb1 29.Rxb1 Rd8 30.Rb8 Bc8 31.Rxc8 Rxc8 32.Bxe4 Qxa4 33.Bb7 Rc2 34.Qb8+ Kg7 35.a7 Nxg3+ 36.Kg2 Nf5 37.a8Q Nxe3+ 38.Kf3 Rxf2+ 39.Kxf2 Nd1+ 40.Kg3 Qb3+ 41.Bf3 Black Resigns. [TWIC 90, 7 July 1996]

This was followed by the game in PGN format, where Crowther commented, 'Of course I've had to split the game in two because of the castling.' Of course! Even today, 13 years later, there is much chess software that doesn't understand chess960.

When I read Crowther's report, my first reaction to the initial position was 'That can't be right!', because the King was not placed between the two Rooks. Then I realized that Crowther had listed the pieces in order of value (KQRRBBNN), rather than in order of board position (QRNKBNRB). For reasons that I'll give some other time, the order by value is my preferred method of setting up a chess960 start position.

The start position QRNKBNRB is no.491 (SP491). After White's 10th move, the players reached the following position.


(After 10.e2-e3)

Now Black castled 10...O-O (Crowther: 'The black King goes to g8 and the black Rook to f8') and White played 11.Ke2. Here the game starts to look very much like a traditional chess game, except for the Black Bishop buried on h8. After 11...g6, even that difference fades into the game score.

Given the general lack of material about chess960, it's not surprising that there is very little on the web about the Bronstein - Douven game. Google locates David Bronstein herzien?, a Dutch review of a new (2009) edition of 'The Sorcerer's Apprentice' by Bronstein & Fürstenberg. That's the same Furstenberg who reported the game to TWIC in 1996.

Google also does a reasonable job of translating the book review, where we learn the reviewer's main objective: 'The question was whether the need is so new and expanded version. And to answer that question, go look for the differences.' That's a good question, because I have the 1995 edition of 'The Sorcerer's Apprentice', and I would also like to know what's new. It turns out that the chess960 portion of the review is:

And gives a lot Fürstenberg Bronstein - Douven, from a friendly game Anderlecht - Eindhoven, with random chess is played. The reader is warned of passing and the fact that Fischer Random is a derivative of an idea that Bronstein himself has invented twenty years earlier.

Perhaps not coincidentally, near the end of the 1995 edition is a brief Q&A passage explaining why Bronstein, a World Championship challenger in 1951, might have been attracted to chess960 in 1996:

Q: Classical chess appears to be in a state of crisis. • A: Because chess has already been studied so thoroughly. The aura of mystery has vanished! Today chess relies on the splendid memory of the young. They complain that they have to work hard, that our generation does not understand them, does not realize how difficult it is to digest thousands of games, commit a mass of information to memory, and not confuse the proper order of moves in reproducing a 20-move long variation on the board. I appreciate that this involves enormous physical and mental strain. But this has nothing in common with the chess that was played by my generation. (1995, p.289)

What about the statement that 'Fischer Random is a derivative of an idea that Bronstein himself has invented twenty years earlier'? That would also make a good subject for another post.

07 November 2009

Fischerandom's Inauspicious Start

The press release issued for Fischer's press conference in June 1996 -- see Fischer Announces Fischerandom for the full text of the announcement -- contained two mentions of a forthcoming match.
The object of the conference was to publicize the launch of Fischer's new game, Fischerandom Chess, and to announce the match of Fischerandom Chess between Philippine Grandmaster Eugene Torre and two-time Argentine champion International Master Pablo Ricardi which begins July 12.


Now the world awaits the first Fischerandom Chess match which will commence at Pasaje Dardo Rocha in the Argentine city of La Plata on July 12. The match between Torre and Ricardi will be played until one player accumulates four victories, draws not counting.

What was the result of that match? The only mention I could find was in TWIC 91 (14 July 1996).

8) Fischerrandom Match between Torre and Ricardi. Eric van der Schilden is there for the match and will cover it on the TASC www page. It has been postponed for a few days due to the continuing illness of Ricardi see: http://www.tasc.nl/frandom

The tasc.nl link is long defunct, but the page lives on in Archive.org: web.archive.org/web/*/www.tasc.nl/frandom/. Here's a screen shot of that page, titled 'FischeRandom Match : Torre-Ricardi'.

After informing us that

On July 11 the final press conference saw the presentation of the clock and shuffler, two electronic devices essential for the new chess: [Fischer's clock and shuffler]. From July 12 onward, Eugenio Torre and Pablo Ricardi should have played the first ever FischeRandom-event: [Rules for the Torre-Ricardi match].

the page ends abruptly with (sic)

Torre-Ricardi match cancelled. Finally the true reason of the delays in starting with the Torre-Ricardi match has surfaced. It appears that the Argentinean organisers couln't show their part of the prize fund upon arrival of the players as was stipulated in their contracts. Thereupon Fischer, who was to donate the other part of the prize money, refused to let the match start. Last-minute attempts to solve this impasse remained invain. The money could not be provided. Fischer set an ultimatum to the morning of July 17. Nothing happened and he and Torre packed their bags and left La Plata that same day.

Over three weeks passed between Fischer's press conference and the cancellation of the kickoff match.

01 November 2009

Fischer's Rules of Fischerandom

When I wrote the post Fischer Announces Fischerandom, I omitted the rules published by Fischer in June 1996. They can be found here...
Bobby Fischer's Rules of Fischerandom Chess

...For a much simpler version of the rules, see Fischer Explains the Rules of Fischer Random. The rules were also copied into Gligoric's Shall We Play Fischerandom Chess?, in the chapter titled 'Fischerandom Rules' (p.86-92). For other copies of the rules, which likely lead to other resources about chess960, see Google: "pieces on their respective back rows receive an identical random shuffle".

Fischer's rules included the following explanation...

White and Black have identical positions. From behind their respective Pawns the opponents' pieces are facing each other directly, symmetrically. Thus for example, if the shuffler places White's back row pieces in the following position: Ra1, Bb1, Kc1, Nd1, Be1, Nf1, Rg1, Qh1, it will place Black's back row Pieces in the following position, Ra8, Bb8, Kc8, Nd8, Be8, Nf8, Rg8, Qh8, etc.

...where the position used in the example is shown in the following diagram.

Start Position 760

Did the position have some special meaning for Fischer or was it also chosen at random?


Later: Gligoric mentions earlier in his book (p.36) that the rules he quotes were formulated in September 1993, almost three years before the press conference announcing them.