Chess960 FAQ

Chessgames.com's page titled FISCHERANDOM CHESS: Random Position Generator is much more than what the title states. It's a discussion group (aka 'Kibitizer's Corner', aka 'Reader Commentary') that started in January 2003. Although the number of posts has declined in the last year or so, it is a valuable repository of ideas touching on all aspects of chess960.

While browsing through 6+ years of posts, I noticed several themes that repeated frequently. These make a good basis for a chess960 FAQ:-

• Where can I play chess960?
  
• How do you castle? (often with reference to a specific position)
  
• Is chess960 really chess or is it some kind of chess-like variant?
  
• Aren't some chess960 positions forced wins for White?
  
• Why can't chess and chess960 co-exist?
  
• Does Fischer deserve credit for the chess960 idea?
  
• Isn't chess960 mainly for strong players who want to avoid opening theory?

Do you have another question that you think should be in a chess960 FAQ or a good answer (not necessarily original) to one of the questions?

***

Here's a FAQ-like answer from Chessgames.com itself: (Q) Why does CG.com spell 'Fischerandom' with one 'r'? (A) Because that's how Fischer spelled it. • This begs the question: when did Fischer ever spell it? He wasn't known for committing his thoughts to writing. • Here's another: (Q) Why does the CG.com 'Random Position Generator' use nonstandard numbering for the start positions? The (A) is mentioned many times in the discussion.

Populating the Database

I executed the plan described in A Simple Chess960 Database for Tracking SPs by creating an entry for each SP that I've already discussed elsewhere. A few small design issues cropped up that will need a little more thought.

How should I handle twins? Some discussions compare and contrast the differences. I did nothing special to flag them, so I'll document them here:

• SP034 - BNNQRBKR
  
• SP448 - BBRNNKQR
  
• SP534 - RNBKQBNR

How should I treat multiple posts on the same SP?

• SP059 - NNRQBKRB : Separate posts on the same day?, or
  
• SP534 - RNBKQBNR : Multiple links in one post?, or
  
• A combination of both techniques?

How much should I say about the traditional start position (and its twin)?

• SP518 - RNBQKBNR
  
• SP534 - RNBKQBNR

Before doing any extra work, I'll run for a while with what I have and see how useful it is (or isn't).

First Move Advantage in Chess960

A recent article on Chessbase.com, Refined Chess - a new proposal to combat draws (26 April 2009), another viewpoint in an ongoing dialog about the draw problem in chess, mentioned chess960 in passing.

Random Start Position: This was Fischer's proposal, which enjoyed a modest success. The problem is, not all starting positions offer the same chances to both players. Some gives white a huge advantage, some are too drawish.

This is a common criticism of chess960, where, as here, the critics always present the argument as a fact known to everyone, so examples are not required. I raised the issue on the SchemingMind.com chess960 forum (see Problems with certain start positions), and asked

I'm a newcomer to chess960, have played only a handful of games, and haven't seen this problem of start positions where 'some give White a huge advantage, some are too drawish'. Are there any positions that are known to be problematic?

Here are some of the comments I received.

QuantumLoop: 'In my experience so far I have not found any case that I felt offered one side or the other an advantage let alone a HUGE advantage.'

ichabod801: 'I recently read Gene Milener's book on FRC. He presents some data in there that seems to show a white advantage in some positions. However, it's from one tournament (Mainz), so it's not like it's a big enough sample to really draw any conclusions.'

Austin: 'I think the best way to find an answer would be to organise a correspondence tournament and invite the top CC players of the day to participate under full ICCF conditions. Such a tournament should really expose the 'truth' in each of the positions played.'

heuschrecke: 'I annotated the first 100 starting positions of Chess960 here on SM and I am actually playing in the German Chess960 Championship. [...] There are some starting positions where the right to move first leads to a few forced moves: White threatens to win a pawn with his first move. Black has only one answer and White with his 2nd move threatens to win another pawn. That's it and there's no advantage due to the right to make the first move. A big advantage for White and drawish starting positions - I don't know any of them. • And you may run Rybka for the next 50 years to find out if White has in any of the starting positions a big advantage of if any of the starting positions is drawish from the first move on. The development of opening theory in one and only starting position (the standard one) took 100 years. Even taking into account the tremendous acceleration in hard- and software development - how long would it take to come to a satisfying assessment of any of the 959 starting positions of Chess960?

QuantumLoop: 'World Champion Bobby Fischer,chess genius and originator, spent the years after his brilliant conception playing only Chess960. Bobby was a deeply honest man. If there were something unfair with his conception I am certain he would not of promoted it. If this World Champion, one of the greatest chess players in history, thought Chess960 superior to position 518 [the traditional start position] alone, enough so as to abandon playing "the old chess" surely this must be the strongest testimonial you will ever hear!

Along with those perceptive comments was a link to an older forum thread on the same subject -- Advantageous positions -- where Chess Tiger wrote,

The Chess Tigers offer a reward of Euro 5.000,- to anybody finding a forced lost starting position. You naturally have to bring the evidence!

I asked Chess Tiger if the offer was still valid. He replied,

Yes, it is! But I guess you will agree that we are not really in danger to pay this out.

By introducing chess960, the author of the Chessbase article on 'Refined Chess' was relying on a straw man defense to make his case about drawn games. The purpose of chess960 is to eliminate the impact of opening theory in chess, not to reduce the number of draws. If it does both, that's fine, but draw reduction is not its primary selling point.

***

Wikipedia has an excellent article on First-move advantage in chess. Many of its points -- winning percentages for both sides, drawn with best play, symmetrical openings -- are also relevant to chess960, even if the variant is not (yet) cited.

Small Opening Moves Can Have Big Consequences

I advanced from the first round of the 2009 Chess960 Dropout Tournament (see Pyramids and Dropouts for an explanation of how the format works) by winning both games. I discussed the start of one game in my post on Chess960 Twins, while the other game started with the position shown in the diagram.

Start Position 482

The line-up of the h-side pieces is the same as in the traditional start position (SP518). The initial moves were 1.c4 c5 2.b3 b6 3.Nc3 Nf6 4.Nf3. The first observation to make is the symmetry of the position. Chess960 games often start with symmetrical moves or transpose into symmetrical positions, more often than from SP518. There are several reasons why this happens, and I'll discuss them in a future post.

At this point in the game I judged that multiple piece exchanges on f3/f6 were inevitable, making the development of the h-side Bishop a key decision. Instead of continuing 4...Nc6, maintaining the symmetry, I decided to give my opponent the chance to break the symmetry, thereby going astray. I chose to capture toward the center by playing 4...e6 followed by a later ...gxf6. My opponent chose 5.g3, capturing away from the center with a later e2xf3. These decisions had a major impact on the evolution of the game.

After the inevitable exchanges, I had a mass of central Pawns, partly unopposed. I worked these into a steamroller, gained a big advantage in space that let me maneuver on both sides of the board, and eventually won the game. You can see how the idea worked by playing through the full game, PGN courtesy of SchemingMind.com:

[Event "2009 Chess960 Dropout Tournament, Round 1"]
[Site "SchemingMind.com"]
[Date "2008.12.16"]
[Round "1"]
[White "Chessic"]
[Black "bemweeks"]
[Result "0-1"]
[SetUp "1"]
[FEN "bqrnkbnr/pppppppp/8/8/8/8/PPPPPPPP/BQRNKBNR w KQkq - 0 1"]

1.c4 c5 2.b3 b6 3.Nc3 Nf6 4.Nf3 e6 5.g3 Nc6 6.Bg2 Nd4 7.O-O Nxf3+ 8.Bxf3 Bxf3 9.exf3 Be7 10.Ne4 O-O 11.Nxf6+ Bxf6 12.Bxf6 gxf6 13.Qe4 Qd6 14.Rcd1 f5 15.Qh4 Qd4 16.Qxd4 cxd4 17.f4 f6 18.Rfe1 Kf7 19.Rc1 d6 20.d3 Rfe8 21.f3 Re7 22.Re2 e5 23.Rce1 Ke6 24.fxe5 dxe5 25.f4 Kd6 26.Kf2 Rce8 27.Kf3 Kc5 28.a3 e4+ 29.Kf2 a5 30.Kf1 h5 31.Kf2 h4 32.gxh4 Kd6 33.b4 axb4 34.axb4 e3+ 35.Kf1 Rh7 36.Kg2 Rxh4 37.Kg3 Reh8 38.Kf3 Rh3+ 0-1

Just as in traditional chess, opening ideas can easily carry into later phases of a chess960 game.

A Simple Chess960 Database for Tracking SPs

As outlined in Blog Posts Cross Referenced to Start Positions, I initialized a database of chess960 start positions (SPs) by creating an anchor for the first SP of each month. The posts are accessible from the 'Blog Archive' in the sidebar. The image shows what the database currently looks like for the last three months of 2001, where 1 December 2001 corresponds to SP699, etc.

Not much to write home about, is it? Here's the same table from the previous post pegged to specific dates. It lists SPs about which I've already written plus a mention of why I used it.

• 2000-01-03: SP002 - BQNNRBKR • An Exercise in Chess960 Positional Thinking • considerations on the first few moves
  
  
• 2000-01-25: SP024 - NBQNBRKR • A Followup, an Error, and an Insight • a position which might be much better for White
  
  
• 2000-02-04: SP034 - BNNQRBKR • How Top Players Treat the Same Chess960 Position • also SP919 RKBRQNNB: (like SP034, but with pieces in reverse order); 10 different sequences of opening moves from the same SP
  
  
• 2000-03-01: SP059 - NNRQBKRB • Undefended Pawns in Chess960 Start Positions • three undefended Pawns
  
  
• 2000-12-28: SP361 - NRKBBRQN • Pyramids and Dropouts • initial moves explained
  
  
• 2001-02-21: SP416 - BBRNQNKR • Chess Engines Don't Always Get It • making a plan for the initial development
  
  
• 2001-03-25: SP448 - BBRNNKQR • Chess960 Twins • also SP703, its twin; making a plan for the initial development
  
  
• 2001-06-19: SP534 - RNBKQBNR • A Database of Chess960 Start Positions • symmetrical SPs w/ K & Q in the center
  
  
• 2001-06-19: SP534 - RNBKQBNR • Chess960 Opening Theory • also SP518; initial moves for the tradtional SP as determined by chess engines
  
  
• 2002-02-27: SP787 - BRQKNRNB • Chess960? I'm Hooked! • initial observations on chess960
  
  
• 2002-06-20: SP900 - RBBKQRNN • Lesson in Chess960 Opening Patterns • watch out for early blunders

One issue is how to treat posts that discuss twin SPs: link them once or twice? Another is how much material to reference for SP518 (the traditional start position), which comes up frequently for the purpose of comparison.

Note that this structure offers two ways to use posts tied to a specific day. A single day can have an unlimited number of posts (although there might be a practical limit that I'm not aware of) -or- a day can have one running post with separators for different topics related to that day. Also possible is a mix of the two structures.

Blog Posts Cross Referenced to Start Positions

One of the reasons to create a blog specifically for chess960 (see First Post, New Blog) is to cross reference blog posts with discussions on specific start positions (SPs). For example, here are posts where I've already discussed a specific SP. There are two posts that discuss SP534, but they are not linked together.

• SP361 - Pyramids and Dropouts
  
• SP416 - Chess Engines Don't Always Get It
  
• SP034 - How Top Players Treat the Same Chess960 Position
  
• SP448 - Chess960 Twins
  
• SP534 - Chess960 Opening Theory
  
• SP059 - Undefended Pawns in Chess960 Start Positions
  
• SP024 - A Followup, an Error, and an Insight
  
• SP534 - A Database of Chess960 Start Positions
  
• SP002 - An Exercise in Chess960 Positional Thinking
  
• SP900 - Lesson in Chess960 Opening Patterns
  
• SP787 - Chess960? I'm Hooked!

Using a blog as a database to manage content isn't obvious, but I hit on the idea of assigning specific historical dates to specific SPs. For example, 1 January in the year 1 AD (or any day in that year) could be SP001, a day in the year 2 AD could be SP002, and so on through 1 January 960 for SP960. (In fact, chess960 SPs are numbered 0 to 959, but that's a detail that doesn't change the concept.)

A key question to use this scheme is how far back I can pre-date blog posts. I tested the possibilities in Find Earliest Possible Post, and discovered that Blogger.com -- the software service domain for blogs published at Blogspot.com -- allows only the dates 1 January 2000 or later. This is a limitation for blogs that treat historical subjects. I can easily imagine that it would be useful for a WWII blog to pin posts against specific dates in the 1930s and 1940s, or for an American Civil War blog to tie posts to dates in the 1860s. Since Blogspot doesn't work that way, I had to find another method.

Another idea would be to use the category concept to tie posts together. For example, a category called 'SP518' would pull together all posts referencing the traditional start position (RNBQKBNR), while category 'SP534' would do the same for its twin (RNBKQBNR). This might in fact be the best method to keep track of SPs. One drawback is that with 960 categories, the length of the category sidebar would become a nuisance to scroll.

Next I hit on the idea of assigning SPs to specific dates starting 1 January 2000. That date would be used for SP000, 2 January would be SP001, and so on. It's a little clumsy to keep track of which date corresponds to which SP, but it's easy to invent navigational tools that locate SPs quickly. The category system isn't excluded for important positions like SP518.

The start of each month maps to specific SPs according to the following table. Using this scheme the post 'SP787 - Chess960? I'm Hooked!' would be referenced under 27 February 2002.

2000-01-01 (31 days) : SP000 BBQNNRKR
2000-02-01 (29 days) : SP031 NQNRKRBB
2000-03-01 (31 days) : SP059 NNRQBKRB
2000-04-01 (30 days) : SP090 NNRKBBRQ
2000-05-01 (31 days) : SP120 NBQRBNKR
2000-06-01 (30 days) : SP151 NRBNQKRB
2000-07-01 (31 days) : SP181 NRBBNKRQ
2000-08-01 (31 days) : SP212 NBBQRKNR
2000-09-01 (30 days) : SP243 BNRKQNRB
2000-10-01 (31 days) : SP273 BNRBKNRQ
2000-11-01 (30 days) : SP304 BBNQRKRN
2000-12-01 (31 days) : SP334 NRQKRBBN
2001-01-01 (31 days) : SP365 NRKBRQBN
2001-02-01 (28 days) : SP396 QBRNNKBR
2001-03-01 (31 days) : SP424 RBNQBNKR
2001-04-01 (30 days) : SP455 RNBNKQRB
2001-05-01 (31 days) : SP485 QRBBNKNR
2001-06-01 (30 days) : SP516 RBBNQKNR
2001-07-01 (31 days) : SP546 BRNKNBQR
2001-08-01 (31 days) : SP577 BQRBNKRN
2001-09-01 (30 days) : SP608 BBRNQKRN
2001-10-01 (31 days) : SP638 RNKQRBBN
2001-11-01 (30 days) : SP669 RNKBRNBQ
2001-12-01 (31 days) : SP699 RQKNBNRB
2002-01-01 (31 days) : SP730 RKNQBBNR
2002-02-01 (28 days) : SP761 RKNBBNRQ
2002-03-01 (31 days) : SP789 RQBBKNRN
2002-04-01 (30 days) : SP820 RBBKNQRN
2002-05-01 (31 days) : SP850 BRKNRBNQ
2002-06-01 (30 days) : SP881 BRQBKRNN
2002-07-01 (31 days) : SP911 RKQRNNBB
2002-08-01 (31 days) : SP942 RKRNQBBN

Note that I've excluded 29 February 2000 from the table. This keeps the numbering consistent across Februaries in different years. I'll introduce this scheme over the next few days to see how it works in practice.

First Post, New Blog

This blog is an evolution of the posts that I've been running since August 2008 on my Chess for All Ages blog: Showing posts with label Chess960. For now, I'll continue writing about chess960 at the rhythm of one post per week.