Would You Recommend How to Reassess Your Chess by Silman?

Thank you for the links.

I like this:

It uses the fact that chess is a symmetric game, and that therefore the analysis function must give symmetric scoring. That is to say that at any point, the score for white is exactly minus the score for black, or equivalently the sum of the two scores always equals zero.

This is quite straightforward to understand. If white is winning by one pawn, then clearly black is losing by the same amount. This principal can be extended to positional advantages, i.e. if white has doubled rooks on one file, then white has a bonus score, whereas black's position is weaker by the same amount because of this.

---

I should look for opportunities to get bonus scores on every move.

What is the bonus score for a lead in development by just one tempo? What is the bonus score for doubled rooks?

What is the numerical value for a three tempi lead in development? I have this situation in one of my games. What is that worth! 

These symbols are in all of my chess books: =, +/-, -/+.  Even though they are good enough for my level of play, I prefer numbers, like (+ 0.70).

Thanks, sarsaila.

It's hard to say. The values used are experience-based (connected with chess games statistics), and are not calculated in certain formulas.

Being a tempo ahead allows better development, so being a tempo ahead adds points to the attacked and defended squares, and to a certain piece's placement points. So, if you have opened up a good diagonal with the extra tempo  or have deployed a piece to the center, I guess the position evaluation difference is how much the tempo is worth (in the second site there are some piece evaluation tables; if you have moved Bc1-Be3 you should have gained 0,10-(-0,40)=0,50 points for example). But if you have already developed well, it may turn out to be a minor advantage.

It may turn out to be useful to read some books (or sites) about chess programming. They are very straightforward. A look through them can be informative about what is valuable in chess, how to choose a move, which moves to consider first, how to manage your time, how to decide upon calculation depth, etc, etc. Few typical chess books deal with those topics. Improving a chess program is connected with better time management, better (wider and more exact) view over the table, improvement at tactical calculation algorithms, implementing endgame (or solved position) knowledge, etc. - all of which are problems of improving human play too.

I've always wondered how you could truly program strategy in an effective way. For us humans, the concept of using points to try to generalize things (pawns are worth 1, knights are worth 3, castling is worth .8, an outpost is worth a third of a pawn, etc) is badly criticized and for good reason. A computer may understand attacked and weak squares, but how does it know if they are significant? There are positions where one side has a totally open long diaonal for the fianchettoed bishop, but it doesn't matter since nothing is happening along that particular diagonal. This is very interesting to me

But I guess when you combine at least a hint of strategical skill with the incredible calculation, it doesn't matter

Check out my short review:

http://blog.chess.com/SimonSeirup/book-reivew-how-to-reassess-your-chess-4th-edition

The computer can't guess whether there will be any play going on over certain squares. But the very control of these lines / files / squares is what is useful, because attacking them restricts the opponent from placing unthreatened pieces there, and it also allows own pieces to be maneuvered along them. It may turn out that the reason the diagonal from the example is not used is just because it's being controlled by one of the sides, and the second finds it not favorable to place pieces there. (In the sense of the Negamax search used by computers, that's a good advantage for the first side, because it disallows many moves for the other side, which would otherwise be improving development therefore would gain points; in ordinary chess reasoning, you don't let the opponent execute plans containing control of that diagonal, and if you restrict him/her from using a few more lines he/she will be totally strangled.)

I guess that works out well for the computer, then. There are, of course exceptions -- when I think of fianchetto on an "empty" diagonal, I think of benoni structures where black's pieces are cramped and the only thing he has is a bishop on g7 on a wide open diagonal. But since it doesn't work with the other pieces, it may not be useful. (But then, I guess if there were no active pieces working with the bishop it would factor that in too )

What is the difference between strategy and tactics? I think that tactics require particularity; temporatity, circumstance dependance; progress (action); a limit in view length;  tangibility and certain inevitability. Strategy requires integrity; permanency; evaluation; farsightedness; generalization.

That's whyI think that matters which concern position evaluation are strategical, including piece placement, attacked and defended pieces / squares, en prise pieces, pawn structure strengths and weaknesses - doubled / tripled / backward / isolated / passed, etc. pawns, and many others which are included in the analysis.

Inbalances are checked both through the position (and pieces) evaluation and through specific move sequence analysis, similar to the example you gave in the brackets. If you mean that humans tend to have a look at certain more promising lines first, there are algorithms for move alalysis priotitization at chess prorgams too, and there are some examples of them in those two sites (alpha-beta pruning, principal variation search, killer moves).

It seems that although a computer can't exactly "plan," it can positionally evaluate, just in a, naturally, very primitive way compared to humans.

Supposedly Botvinnik worked for many years on a chess program that was meant to implement the kind of "thought processes" and reasoning that humans employ. Afaik, there isn't much to show for all that effort - though I'd love to have a look through his notes.

I gave the "Killer moves" as an example of move choice prioritization, not about usage of strategy. My puspose was to show that there are mechanics in chess programs that imitate human preference for exploring promising lines first (trying to find a good plan by exploiting obvious strengths / weaknesses and other features).

Computers are better (at least more efficient) than humans in calculation. But that doesn't itself mean they don't use strategy. What is important is whether there are strategical concepts involved in chess programs (according to me there are), not how good computers perform in comparison to humans. The difference in calculation skills is much more obvious than the difference in strategy (as I wrote, tactics are particular, tangible and inevitable, so their usage by humans and computers is easy for a side watcher to spot and compare). What catches the eye is the computer's tactical skill, not its ability to evaluate the position, since one can easily see the particular effect of a tactical sequence within 2-3 moves, but not the overall reasoning and farsightedness, which is strategy. The latter is hardly perceptible, but it is present.

A chess game can be seen in two equivalent ways: an initial board position followed by a sequence of moves, or an initial board position followed by a sequence of board positions.

To decide whether tactics or strategy are more important, used, preferred, etc. we need to compare which of those views is used. Computers use both, because they both generate possible move branches and evaluate static positions at their ends.

If you don't agree, think about the following problem: How could a computer decide whether a move is good without evaluating the resulting position difference?

No matter how it calculates, it should either be able to evaluate the resulting position, or come up with a position with known evaluation (such as a predefined position (usually an endgame one), a checkmate or a draw). The latter is much more resource consuming than the first one, not to mention it is not always possible.

Well, I believe computers are actually really inefficient calculators -- they'll look at the most obviously bad lines sometimes -- it's just that they do this calculation so fast it simply doesn't matter.

And although a computer has its evaluation system -- just as a human would -- that doesn't mean we plan the same way -- a computer "plans" by calculating 6 moves ahead, then evaluating afterwards; a human probably wouldn't bother with precise calculation and instead come up with some general ideas in his head. Basically, computers don't work with general ideas; always precise moves.

However, they could be made to if someone came up with a scheme to codify and model those general ideas in mathematical or heuristic ways. This is what Botvinnik was working on. Evidently he didn't get very far, but there is nothing inherent in the problem to rule out this approach on theoretical grounds, and technology has moved on since then. Why don't people make the effort to do this? Well, maybe someone is trying. But why attempt something harder when the tried and tested way works well enough and can be improved by tweaking evaluation functions etc. and with faster hardware?

We are not computers therefore positional chess does exist.  davidegpc has once again proven his own arguments false.

I'd imagine it to be really, really difficult though, because the brain is just so ridiculously complex. In fact, if we try to program these vague "general ideas" that work well for us humans, it might actually make the engines play their own game worse.

"But why attempt something harder when the tried and tested way works well enough and can be improved by tweaking evaluation functions etc. and with faster hardware?"

Exactly. The way engines think is what got them so far in the first place. Personally, I think that's the only practical way to go. After all, an engine's brute force method should, eventually, solve chess!

While I do agree with much of the above, I have to concede some arguments to the positional players' position...namely that because we humans cannot calculate as well as a computer, once our powers of calculation are over-extended and we have no exact "memorized"/"prepared" knowledge of the lines that show up in our games, our next best bet is perhaps "positional guess"***.  Perhaps I will write a book about it to go with my very own original concept of "positional inequities"***, since it has been established that you don't need to be a super-GM to write a beloved book or even be a master to write a cult classic or good seller (MDLM, Hortillosa, Albertson).   

*** copyright 2011 dannyhume. All rights reserved.

Saying there is no such thing as positional chess is a weak and pointless argument over semantics.

problem with computer do they have plans or ideas ? nor they can tell you what you are doing wrong.