30 May 2010

Lasker on Basic Opening Principles

After his discussion of basic opening principles, outlined in my post Knights before Bishops?, Lasker gave 64 pages of specific opening variations. He summarized these as follows (p.106, abridged again):-
  1. It is of great value for either side to dominate points.

  2. He who dominates central squares is better off than the ruler over wing squares.

  3. He will have gained the advantage in the opening who first succeeds in bringing Pawns and pieces to dominate the extended center.

  4. To fix the exchange value of the Pawns and pieces and the move... (see my post on Lasker's Table of Opening Values).

With the possible exception of that fourth rule, Lasker's considerations apply to all chess960 start positions. What can be learned from other authors of chess primers when they introduced the subject of chess openings?

29 May 2010

Knights before Bishops?

In his 'Manual of Chess', Lasker introduced his chapter on openings with a number of rules in historical order (p.41, abridged here):-
  1. In the 18th century they announced their first rule: 'Sortez les pieces' -- 'Get the pieces out'.

  2. It took a hundred years before a new rule was announced [London 1851]: 'Move that one of your pieces which is in the worst plight, unless you can satisfy yourself that you can derive immediate advantage by an attack'.

  3. A few decades went by, tournaments became of frequent occurrence, and the masters, coming together oftener than before, evolved a 'public opinion'. That tended toward the rule: Avoid the moves of Pawns in the opening as far as possible.

  4. I have added to these principles the law: Get the Knights into action before both Bishops are developed.

The first two 'rules' definitely apply to chess960 and the third probably applies, but I'm not sure about the fourth. Is it a specific rule that applies mainly to the traditional start position (SP518: RNBQKBNR) and its twin (SP534: RNBKQBNR), is it a rule of thumb that applies to a family of start positions, or is it a general rule for all such positions?

23 May 2010

Tiger on the Prowl

In Chess960 Simuls @ Mainz, I noted that the last chess960 simul at Mainz was given by GM Levon Aronian in 2006. He lost one game, gave up three draws, and won the rest. That one loss was to Mike Rosa, a member of the organizing staff at Mainz whom we have already seen in a post titled Chess Tiger.

The diagram shows the position in the game (SP800 BBRKQNRN) after the first five moves: 1.c4 e5 2.Nhg3 Nhg6 3.b3 f6 4.e3 b6 5.Qe2 Ne6. Although castling O-O-O is already possible on the first move, both players have declined to use that option.

Chess960 Simul, Mainz 2006
Rosa, Mike

Aronian, Levon
(After 5...Nf8-e6)

GM Aronian has used his first five moves to open lines for the Bishops on adjacent diagonals, to develop the Knight on h1 to its natural square, and to develop the Queen to e2. He probably intends to continue Ng3-f5, Nf1-g3, and O-O. Chess Tiger Rosa has used his five moves to open the line for one Bishop, to develop both Knights, and to stake a claim to the center with ...e5. The move ...f6 was necessary to prevent the loss of the e-Pawn after Bb1xg6.

Now Aronian played 6.Qh5, attacking the undefended h-Pawn and rendering ...O-O more difficult. Here is the full game score.

[Event "CCM6 - Chess960 Simultan"]
[Site "Mainz"]
[Date "2006.08.17"]
[Round "?"]
[White "Aronian, Levon"]
[Black "Rosa, Mike"]
[Result "0-1"]
[SetUp "1"]
[FEN "bbrkqnrn/pppppppp/8/8/8/8/PPPPPPPP/BBRKQNRN w GCgc - 0 0"]

1.c4 e5 2.Nhg3 Nhg6 3.b3 f6 4.e3 b6 5.Qe2 Ne6 6.Qh5 Rh8 7.Nf5 Ne7 8.Qxe8+ Kxe8 9.N1g3 h5 10.Nxe7 Kxe7 11.d4 d6 12.h4 Kf7 13.d5 Nf8 14.f4 c6 15.fxe5 dxe5 16.e4 Bd6 17.Bd3 Ng6 18.O-O Nxh4 19.Be2 Ng6 20.Nf5 Bc5+ 21.Kh2 Nf4 22.Rc2 cxd5 23.exd5 h4 24.Bg4 Rce8 25.Bb2 g6 26.Nd4 exd4 27.Rxf4 Bd6 0-1

Rosa also managed to draw against Leko in the 2003 chess960 simul. If he plays later this year in the simul against GM Alexandra Kosteniuk, the Women's World Champion should know that she's up against a tiger.

22 May 2010

Chess960 Simuls @ Mainz

In (Almost) No Chess960 @ CCM10, I reported the unwelcome announcement that chess960 at Mainz 2010 would be limited to a single simul by GM Alexandra Kosteniuk; then I wondered about the ground rules of chess960 simuls. As a follow-up, I sent an email on the subject to GM Kosteniuk, who replied that this will be the first time she conducts a chess960 simul, and other than having White in all games, she is not sure how the start positions will be chosen, whether there will be different start positions in every game, or how experienced her opponents will be.

When I sent the email, I recalled that several chess960 simuls had already been played at Mainz. My overview of past years, Chess960 @ Chess Classic Mainz, points to two pages: Playing Chess960 Simuls on 20 boards: Leko & Svidler (2003) and Chess Classic Mainz CCM6 (2006). On the page for 2006 there was no info about Aronian's 20 board simul, but it did get a mention on the corresponding report by Chessbase.com: Chess Classic: Chess960 results, Anand and Aronian simuls.

After a little more digging, I discovered that 59 of the 60 simul games have been preserved. The missing game appears to be Aronian's win against celebrity 'German soap star Vaile', mentioned in the Chessbase report. I also discovered that, in More on the Concept of Distance, I had already commented on an Aronian simul game with celebrity supermodel Carmen Kass.

From those game scores, I can answer two of my original questions. Many of the opponents for both Leko in 2003 and Aronian in 2006 are listed with their ratings. There were a fair number of 2000+ players in both simuls and of the other players with ratings, all were club players or better (>1400). Of course I can't tell if they were experienced chess960 players, but in 2003 I would guess that most were inexperienced. There were also a few chess celebrities among the GMs' opponents. Leko played against Albert Vasse of DGT, while Svidler played Carsten Hensel, Kramnik's former manager.

The PGN records show that the GMs had to contend with a different start position in every game. This was confirmed by the 2003 page:

It was not an easy task for the two top-players, because they could not make use of their enormous opening knowledge. However, their general understanding of the game was enough to win the simul comfortably against 20 ambitious club players.

Now if I can determine how the start positions will be selected, I'll know (almost) everything there is to know about chess960 simuls at Mainz. Perhaps they will use the DGT960 clock, 'the only chess clock worldwide that automatically generates the starting positions of chess960'.

16 May 2010

Chess960 'as a Spectator's Sport'

Continuing with 'Hardly Ever Played Chess960 Before', we saw that Anand qualified along with Aronian into the final match for the FiNet Chess960 Rapid World Championship at Chess Classic Mainz in 2007. I once used the second game of that match to illustrate a post on Castling: The Longest Possibility. A full account of the match can be found at Chesstigers.de: Chess as a spectator's sport, 'Aronian wins Chess960 World Championship in dramatic match against Anand'. The first game ended abnormally.
Aronian fell behind on the clock and around move 20 had only 20 seconds left while Anand still had five minutes on the clock. But Aronian fought back, played quickly and used the five seconds increment to master all tactical tricks. Meanwhile Anand tried to find a clear way to advantage, spending more and more time on every move. Then, suddenly, it was all over: Aronian pointed to Anand's clock, the players shook hands and left the stage. The spectators could hardly believe it: Anand had lost on time, after his loss against Kamsky in the final of the PCA World Championship in Las Palmas 1995, [it was] the second serious game in his 25 years long career he had lost on time.

There followed two draws in the four game match.

Anand had to win the fourth game to tie the match, and this game turned into a fascinating struggle, giving the spectators everything they looked for. Despite needing only a draw, Aronian sacrificed a pawn on the second move which led to an exchange of blows and counterblows that lasted until the end of the game. If one player seemed to get an advantage, the other came up with an amazing resource only to be countered by an even more amazing move. After an early exchange of queens, both sides sacrificed material repeatedly and tried to mate their opponents. In the end Anand came out on top: He managed to stop Aronian's passed pawn and won material. However, even with only a few pieces left Aronian conjured up dangerous counterplay, creating mating threats out of nowhere. But Anand coolly defended and forced the match into a tie-break.

Aronian's Pawn sacrifice is shown in the following diagram. Note how closely the pieces match the traditional start position.

Mainz 2007, Final Match, Game 4
Aronian, Levon

Anand, Viswanathan
(After 2...d7-d5)

Anand accepted the sacrifice with 3.cxd5. After 3...Qa4+, he was forced to keep his King in the center with 4.Kd2, since 4.b3? loses to 4... Qd4. The PGN score for the complete game is given below (SP535 RNBKQNRB).


[Event "FiNet Chess960 Rapid World Championship"]
[Site "Chess Classic Mainz"]
[Date "2007/8/16"]
[Round "10"]
[White "Anand"]
[Black "Aronian"]
[Result "1-0"]
[SetUp "1"]
[FEN "rnbkqnrb/pppppppp/8/8/8/8/PPPPPPPP/RNBKQNRB w GAga - 0 1"]

1.c4 g6 2.d3 d5 3.cxd5 Qa4+ 4.Kd2 c6 5.Nc3 Qa5 6.e4 Nfd7 7.Ne3 b5 8.Ke2 Ba6 9.Ncd1 Qc7 10.g3 b4 11.f4 cxd5 12.Nxd5 Qc2+ 13.Qd2 Bxd3+ 14.Ke1 a5 15.e5 Ra6 16.Rg2 Qxd2+ 17.Rxd2 Bb5 18.a3 e6 19.axb4 axb4 20.Rxa6 Nxa6 21. Nb6 Ke7 22.Nxd7 Rc8 23.Nb6 Rxc1 24.Bb7 Nc5 25.Nc8+ Kf8 26.Rd8+ Kg7 27.Bf3 Rc2 28.Nf2 Rxb2 29.Nd6 Rb1+ 30.Nd1 Ba4 31.Kd2 Rb3 32.Be2 Ra3 33.Bc4 Ra1 34.Nb2 Bc6 35.Rc8 Ne4+ 36.Kc2 Bd5 37.Bxd5 exd5 38.Rc7 Ra3 39.Rxf7+ Kg8 40.Rd7 b3+ 41.Kb1 Nc3+ 42.Kc1 Ra2 43.Rc7 d4 44.Ndc4 Bg7 45.Rc8+ Bf8 46. Rb8 Kg7 47.Rxb3 Nd5 48.Rb5 Ne3 49.Kb1 Ra4 50.Nxa4 Nxc4 51.Rb7+ Kh6 52.Kc2 Na5 53.Rb5 1-0

To break the tie after four games, the players continued with a pair of blitz games. Aronian won the first game with the White pieces and held the second with Black to retain the title he had first won the previous year.

15 May 2010

'Hardly Ever Played Chess960 Before'

In addition to his frequent participation in events for the traditional World Championship title -- the 1995 Kasparov - Anand PCA Title Match, the 1998 Karpov - Anand FIDE Title Match, the 2000 FIDE Knockout Matches, 2005 San Luis, 2007 Mexico City, the 2008 Anand - Kramnik Match, and 2010 Anand - Topalov -- Anand played for the chess960 World Championship in 2007 at Mainz:-
  • Aronian Shines on First Day of Chess960 FiNet World Championship • 'How would Vishy Anand play Chess960? The many fans of the nine times Chess Classic winner were keen to find out -- particularly because the World’s number one revealed before the tournament that he had hardly ever played this form of chess before. He himself said in the press conference after the first three rounds: "I think I coped reasonably well. However, you have to be careful because h- and a-pawns can be without protection right from the start and you have to overcome a mental block because you just cannot imagine that some squares are not protected."'

  • Tigers on the Stage • 'Anand and Aronian were favorites to win the tournament and after the first three rounds most people also expected both players to qualify for the final. However, Bacrot and Kasimdzhanov certainly did not want to go down without a fight, and after all, with 1.5 points from 3 games Kasimdzhanov shared second place with Anand. But their crucial encounter right at the start of day two turned out to be rather one-sided. Anand seemed to have absorbed the lessons from his first three serious Chess960 games quickly and after an opening blunder by Kasimdzhanov, which cost a vital pawn, Anand quickly won.'

That 'opening blunder' occurred after some provocation by Anand.

Mainz 2007, Preliminary Event
Kasimdzhanov, Rustam

Anand, Viswanathan
(After 8...Qd8-d7)

In the diagrammed position, White played 9.f3. There is more to this move than opening the e1-h4 diagonal for the Bishop. Black answered 9...Re8, when White continued 10.e4, threatening a Pawn fork of the Black Knights. Now Black's best was to move one of the Knight's out of danger, but he blundered with 10...dxe4. After 11.fxe4 e5 12.Bg3, Black lost the e-Pawn, because 12...exd4 permits 13.e5.

Here is the PGN score for the game (SP827 RKNQBRNB).

[Event "FiNet Chess960 Rapid World Championship"]
[Site "Chess Classic Mainz"]
[Date "2007/8/15"]
[Round "4"]
[White "Anand"]
[Black "Kasimdzhanov"]
[Result "1-0"]
[Variant "chess 960"]
[SetUp "1"]
[FEN "rknqbrnb/pppppppp/8/8/8/8/PPPPPPPP/RKNQBRNB w FAfa - 0 1"]

1.d4 d5 2.Nd3 Nf6 3.e3 Nd6 4.g4 h6 5.h3 Bb5 6.Ne2 e6 7.a4 Bxd3 8.Qxd3 Qd7 9.f3 Re8 10.e4 dxe4 11.fxe4 e5 12.Bg3 Nc8 13.dxe5 Qxd3 14.cxd3 Nh7 15.Rxf7 Ng5 16.Rf1 Nxh3 17.d4 Nb6 18.Rc1 a5 19.e6 Rxe6 20.Rxc7 Ka7 21.e5 1-0

In the next two rounds, Anand beat Bacrot and drew with Aronian. His score of +2-0=4 was good enough to earn him a place in the final match against Aronian.

09 May 2010

CCRL Discussion Board

While I'm on the topic of Label CCRL, I should mention the CCRL forum: CCRL Public. There isn't a lot of activity for chess960 -- 'Search found 13 matches: chess960' and 'Search found 63 matches: frc' -- but there are some interesting threads.

One thread from 2008, FRC Openings Statistics, complements a recent post, Four Weak Pawns (SP115), where I wrote about the value of the CCRL games in real play.

After some practical experience with CCRL games, I'm less enthusiastic about their value to chess960 opening knowledge. The engines don't play with a book (meaning they don't 'learn') and they have no notion of what it means to follow a strategic opening plan. In my own games I have often found an initial move that none of the machines had played and that seemed like a reasonable move to me. After playing my own moves a few times and suffering no unexpected drawbacks, I stopped looking at the CCRL collection for ideas.

The CCRL thread discusses practical problems of engine vs. engine testing. For example,

I don't think you can get meaningful statistics by repeating the same starting positions with the same engines 100 times. Because there is a large chance of repeating the whole game or the beginning of the game. The engine you are using has particular understanding of particular positions. Even if there is sometimes variation in the chosen moves, the results are not statistically reliable to say something about the position itself. All you can say is that in [a certain engine's] understanding the position is usually won by White.

It's a good point. See the thread for more good points.

08 May 2010

Label 'CCRL'

I reference the CCRL (Computer Chess Rating Lists; see the link under 'Resources' to the right) so often that it deserves to have its own label: Posts with label CCRL. A disadvantage of having spawned this Chess960 (FRC) blog from my main Chess for All Ages blog (the posts Chess960 Blogging Leaves Home and First Post, New Blog mark the transition) is that old posts on the previous blog can't be tagged with labels from the current blog. To rectify that, here is a list of CCRL posts from the previous blog.

The CCRL games database has its flaws, but it's still the best available source of master level chess960 games.

02 May 2010

Four Weak Pawns (SP835)

In my previous post, Four Weak Pawns (SP115), I looked at the first of two chess960 start positions that feature four underprotected Pawns. The second of the positions is shown in the diagram below. Mirroring SP115, its twin, SP835 starts with the a- & h-Pawns undefended, while the c- & d-Pawns are protected only by the King.


SP835: BRKNRQNB

Once again, the four weak Pawns aren't particularly vulnerable to any immediate attack. Of the 52 games in the CCRL collection, the following first moves were played by White...

21 1.b4
11 1.b3
7 1.Nf3
7 1.e4
3 1.g3
3 1.Nc3

...In this position the players don't have the option of castling on the first move, but the other moves are mainly mirrors of the most popular moves for SP115, for the same reasons. I can't explain the overwhelming preference for a push of the b-Pawn, especially 1.b4, where the equivalent SP115 move (1.g4) was one of three equally popular openings. Perhaps it's just a statistical quirk of the small CCRL samples. After 1.b4, the most popular responses for Black were...

8 1...b5, and
6 1...e5

...All six games that started 1.b4 e5 ended in victory for White. Is this another statistical quirk or is there a structural problem with Black's position that needs special attention?

My favorite method to start evaluating a new chess960 start position is to first look at the castling options. In SP835, both players can castle O-O-O by simply moving the d-Knight out of the way. Castling O-O requires more preparation. Both Knights have to move off their start squares and the Queen has to move somewhere. The best square for the Queen is far from obvious and will likely depend on the opponent's scheme of initial development.

My first candidate move here mirrors my choice for SP115: 1.e4. It again stakes a claim in the center, opens a diagonal for the Queen, creates a strong point for the Knight on e3, and is protected by the Rook. That's a lot for a single move to accomplish.

01 May 2010

Four Weak Pawns (SP115)

In the post Four Weak Pawns, where I discovered twin positions with four weak Pawns: SP115 BNQRNKRB and SP835 BRKNRQNB, I noted,
In SP115 the a- & h-Pawns are undefended, while the e- & f-Pawns are protected only by the King. This is quite extraordinary and it would definitely be worth seeing how master level players have coped with these two SPs in practice.

I don't have a large collection of master level chess960 games. My ~1200 games from past Mainz events cover about 25% of the 960 start positions and neither of the two SPs is present. The CCRL collection (see the link under 'Resources' on the right) currently has 60 examples of SP115 and 52 of SP835, so I downloaded them to take a look.

After some practical experience with CCRL games, I'm less enthusiastic about their value to chess960 opening knowledge. The engines don't play with a book (meaning they don't 'learn') and they have no notion of what it means to follow a strategic opening plan. In my own games I have often found an initial move that none of the machines had played and that seemed like a reasonable move to me. After playing my own moves a few times and suffering no unexpected drawbacks, I stopped looking at the CCRL collection for ideas. With that in mind, let's look at SP115, pictured below.


SP115: BNQRNKRB

A quick analysis of the position reveals that none of the four weak pawns are susceptible to an immediate attack. The a- and h-Pawns are not easily attacked by any diagonal piece. The a-Pawn might be captured in three moves by a Knight, but in the time it takes to execute that plan, many countermeasures are possible. The moves most favored by the CCRL engines were...

12 1.Nc3
11 1.g3
11 1.g4
9 1.O-O
6 1.d4
5 1.b3
4 1.b4

...where the plan to develop one of the cornered Bishops by moving the b- or g-Pawn was used in half of the games. Whether to move the chosen Pawn to the third rank or to the fourth requires further consideration.

The move 1.Nc3 is first on the list, although I can't see a convincing reason for it. Is it with the idea of continuing Nc3-b5, attacking the a-Pawn? The most popular responses...

6 1...b6
3 1...O-O
2 1...d5

...indicate that Black isn't particularly concerned with rushing to the defense of the a-Pawn. Another popular move is 1.O-O, which I hadn't noticed until I looked at the CCRL games. It immediately transforms the 'weak Pawn' status of the e-, f-, & h-Pawns: the e-Pawn is now undefended, the f-Pawn is protected twice, and the h-Pawn is defended only by the King. In general, I'm not a big fan of castling on the first move. I prefer to see how my opponent develops a few pieces before I commit to castling on either side.

The other move on the preferred list, 1.d4, would be my personal choice for further inspection. It stakes a claim in the center, opens a diagonal for the Queen, creates a strong point for the Knight on d3, and is protected by the Rook. It also blocks the a1-h8 diagonal, allowing the Bishops on that diagonal to enter the game without being swapped immediately, a natural hazard of games which start with all Bishops in the corner.

All in all, the four weak Pawns don't play much of a role in the initial considerations. It's useful to know that they can't be easily attacked, but don't really serve as a guide for developing the other pieces.