Fischer Random - Chess960 - Chess XLM

The recent rise in popularity of Fischer Random has got me to thinking. 

What changes (if any) do you think it would require in the evaluation functions used by the top engines that have been developed over the last few decades based upon classical chess and the traditional starting position?

For example, I think the value of the pieces is probably about the same (i.e. pawn - 1 point, Bishop/ Knight - 3 points, etc. ) 

I think having the bishop pair is still valuable as it allows you to control squares of both colors.

I think many of the pawn structure principles still apply - doubled, isolated, backward, hanging, etc. etc. 

One principal that at least has me wondering is the value of the center. While I think the center of board is still an important principal as pieces posted there typically radiate more power as compared to being on the edge or in a corner of the board - I am not convinced that the traditional thinking of opening with a center pawn still applies to Random. One of the reasons for opening with a center pawn is not only to help control and stabilize the center, but is also to move the bishop out quickly so the King can castle early. Depending upon the initial starting position, this may not be a factor at all. 

Another principal that has me scratching my head a little bit, is the classical wisdom in attempting to castle as early as possible. Depending upon the starting position of the king, and the pieces surrounding (ie. protecting it), it may not gain anything to castle in certain positions. 

Another thing that (at least for me) has become a tad harder to evaluate is the concept of an outpost. An outpost is really only meaningful if it is located near the opponents King so as to assist in an attack, or is only useful is there is some other strategic objective that it can assist with. I have gotten pretty accomplished at deciding whether an outpost in a particular sector of the board is useful in classical chess. (For example a Knight in d5 or e5). I have a much harder time deciding about outposts in Fischer Random. I think this concept would be much harder to code into an engine with the possibility of 960 different starting positions. 

Anyway, that is about as far as I have gotten in my thinking on the topic. 

What ideas do you have? 

Unfortunately I have no ideas!

Oh man, I finally get an answer and that is it? Thank you for at least reading my thread!

well, i thought i posted an interesting question. either it is not interesting, or no one has a clue how to answer it. Anyone?

oof

Same principles as in regular chess, but like you said, you're going to have to evaluate during the game and decide if it's worth it.

+1 

Also consider 4 player chess. Bishops are worth 5 here with the huge board

Nice piece ! Agree in principle with what you say.

My initial plan is to castle early on the king side. As you say this can cause probs.

Too many good players [1600+] castle long and I think unsoundly.

The centre and outposts are still important.

I play 3min. arena and look for a configuration that allows a quick B exchange for a R.

Stockfish is already awesome at evaluating 960/FR positions.  Maybe there are some flaws in there that a super GM could expose in long time control games, but remember that stockfish was able to very quickly analyze the super GM games in the recent 960 World Championship.  I am not aware of a single instance in which the engine's analysis of the super GM 960 games was called into question.  Additionally, in the long time control games, the players were allowed to use stockfish to analyze the best first move, and they did so.  They did not always pick the move the engine thought was best, but they generally did, and there was no discussion of flaws -- rather, if they picked another move, they just thought it would lead to more interesting/better play from a human perspective.

I generally disagree. I think the center is extremely important in chess960 (I think Caruana has said the same). Outposts are still important, even if the king is not there (but even more valuable if it is near the king). The only principles I think are different are early castling (I agree with you), and there is no need to wait to develop the rooks and queen like in chess518 - there are times when you want to develop a queen early, or get the rooks active. In this way I think chess518 has misled ppl into how to play an opening, with ideas that are only specific to 1 configuration.

I didn't say the center was not important. I was suggesting that the conventional wisdom of opening with a center pawn may be questionable in fischerrandom.

I sometimes find myself get rid of one of my Bishops early in the game, only to subsequently shape my pawn structure such that my other Bishop will become a strong one.

Interesting idea. Thanks for sharing!