Monday, October 7, 2013

Chess960: What is a logical first move?

In the previous post on viable first moves for SP395 Mark offered to play a match to try and find out the truth about what is a viable first move.Thanks for the match sounds like fun!

But before we start I have to clarify that the post was about viable first moves not logical first moves!

So if I reduce the viable list down to logical first moves, what are the list of logical first moves?

I think to decide that, we need to agree on why the center is so important in Chess960......

The center is fundamentally important in all 960 positions because of strategic flexibility.....the problem with Chess is that for at least a move you do not know what your opponent will commit to strategically. Therefore it is illogical to commit unnecessarily on either wing.

So the center (or possibly a flank) is the only place left to play....

Definition of wings, flanks and center:
Wings = a,b or g,h
flanks = c or f
center = d,e

So using that, a logical first move:
  1. Doesn't break any tactical truths about the start (there usually aren't any)
  2. Doesn't break any technical truths (as in moves that excessively slow down development)
  3. Prefers playing out in the center for the sake of flexible strategy. That can include ANY move that controls the center either directly or indirectly. 
  4. Doesn't play out knights unless they take some ownership of the center and/or there is a clear tactical theme that is generally beneficial in material or tempo.
  5. Avoids committing to the wings because it commits to a specific strategy too soon. There may be a few exceptions to this where committing to a wing satisfactorily restricts the opponent in a very few start positions.
  6. Doesn't castle straight away - there is no point revealing intentions so soon.

SP395: Eight logical first moves!?

SP395 is a great example because it includes 100% of all possible first moves in Chess960. So, despite that there are 16 viable moves, according to the above criterion for logic we get these logical moves (not ranked in any particular order):
  1. c4 - yes because it takes some direct center control
  2. d3 - yes because it is a direct center move
  3. d4 - yes because it is a direct center move
  4. e3 - yes because it is a direct center move enabling Nc1-e2
  5. e4 - yes because it is a direct center move
  6. g3 - yes because it is an direct center move
  7. Nc3 - yes because it is a direct center move
  8. Ne3 - yes because it is a direct center move
Rejected moves:
  1. a3 - no because it commits to one wing too soon
  2. a4 - no because it commits to one wing too soon
  3. b3 - no because it is too vague (doesn't commit to the center either directly or indirectly)
  4. b4 - no because it commits to one wing too soon
  5. f3?! - no because it traps the bishop and weakens the king
  6. g4 - no because it commits to one wing too soon.
  7. Nb3 - no because it is too vague (doesn't commit to the center either directly or indirectly)
  8. h3?! - no because it achieves too little
  9. h4?! - no because it commits to one wing too soon
  10. O-O?! - no because it makes little sense to reveal castling intentions so soon
Contested moves on technical grounds:
  1. c3?! - problem is that it is technically not that good although it is an indirect center move. It is technically weak because it doesn't develop any pieces and probably commits to developmental congestion because keeping the c3 square clear is important developmentally.
  2. f4?! - problem is that it is technically not that good although it is an indirect center move. It is technically weak because the only ownership right it makes to the center is to e5, a square that the opponent has no need to own.
  3. Nd3?! - problem is that is technically not good. It takes control of center squares that the opponent need not own, and causes a potentially congested development.
Using the same logic we can compare SP395 with standard chess SP518:

SP518 {c4, d3, d4, e3, e4, f4, Nc3, Nf3} = 8 logical moves
SP395 {c4, d3, d4, e3, e4, g3, Nc3, Ne3} = 8 logical moves

2 comments:

  1. Good post, Harry. Are you warming to the realization that engines don't play the initial moves very well, especially when there are no tactics to consider? Note that I'm not really expecting to play a match with you. The match idea was more of a thought experiment to explore the concept of 'viable moves'. It gave me the foundation for a series of posts comparing first-class moves with second-class moves. I'm going to continue with that theme for a few more posts. - Mark

    ReplyDelete
  2. And a good thought experiment it was thanks Mark. Honestly, I had not thought of the fundamental reason why we play out into the center until I wrote this post. I just assumed that the center was necessary to control for vague developmental reasons.

    But actually, there is no-where else to play by logical elimination.

    I'd say if an engine had it's first move screened for an algorithm something like I have suggested above, they would go up in ELO and beat the other engines in the CCRL competition more often.

    The most difficult step in the above algorithm is what is a good technical move that promotes good development? Funnily enough I think that most engines these days can actually answer that question, yet cannot answer the much more simpler questions due to a lack of programming on the first move choice.

    My two cents.

    ReplyDelete