Author: Bob Durrett
Date: 08:59:23 12/29/03
Go up one level in this thread
On December 28, 2003 at 23:31:21, Tom Likens wrote: >On December 28, 2003 at 20:54:36, Bob Durrett wrote: > >>Your bulletin is exciting since it shows there is still room for innovation in >>engine design. As you and others here develop and implement this idea the >>overall level of play of chess engines will go up simply because the >>front-runners will be forced to adopt some of your new ideas. I have often had >>the suspicion that "current wisdom" regarding searching, pruning, and the like >>is holding back the truly innovative people who might feel obligated to cling to >>the old ideas. I think the whole current theory about "extending and reducing" >>needs to be re-thought, although I am definitely no authority on that subject. >>The Earth is not flat after all! >> >>Perhaps the next extension of your idea is to perform and utilize complete >>dynamic position evaluations to refine the searching. >> >>Congratulations to you and the others for corageously pursuing a new idea which >>perhaps may go against "current wisdom." It is only through original innovation >>and invention that advances will be made. Old ideas are not always the best >>ideas. >> >>Bob D. > >Hello Bob, > >I think you might be underestimating the commercial engines. My suspicion >is that they have been using techniques similar to this for years. From my >short experience with the idea it's seems obvious that it has a tremendous >amount of potential but could be getting it right is tricky. My guess is >the commercial guys have struck the right balance and are reaping the >benefits. > >BTW, Tord Romstad deserves the credit for starting my neurons along these >paths. He's been doing this and more for quite awhile now in Gothmog and >has been more than willing to share his ideas with anyone that wanted to >discuss them. > >regards, >--tom Maybe Tord Romstad should be viewed as being like Captain Kirk in Star Trek who "goes where no man has gone before." In your case, maybe you are one who "goes where only one man [Tord Romstad] has gone before." : ) _ _ _ _ As a non-programmer unfamiliar with the details of the inner workings of chess engines [except for what I get out of this bulletin board], I am not caught up in the "tried and true" methods published in the past but am free to think about ideas which may or may not have been tried in practice. I realize that most of my chess engine ideas are probably naive or unworkable, but that doesn't keep me from thinking about such things anyway. Recently, I have been reviewing recently published chess texts in the search for new ideas [i.e. new to me] which might be applicable to chess engine design. It seems, intuitively, that some of the mental processes of the top GMs reported in the literature might be useful in chess engine design and the idea of allowing the position evaluation [of internal nodes] steer the searching seems to be an example. One of the things Morphy was famous for was attacking only when the conditions for a successful attack were present in the position. Morphy would build up the position creating those conditions. Then, and only then, he would attack. I see no reason why chess engines could not do the same thing Morphy did. The position evaluation should be able to detect and quantitatively assess the features of a position which are "preconditions" for an attack. If those preconditions are present and sufficient then the chess engine should prioritize the searching to give priority to looking at attacking continuations consistent with [or which would exploit] the specific preconditions present in the position. This would be an example of the searching [and Hyatt's "search space management"] being impacted by the position evaluation. According to the literature, when a top GM evaluates a position, he evaluates not only the static but also the dynamic features of the position. In fact, the top GM does a "total" evaluation. As noted in one of my other recent bulletins, a top GM is not limited to assessment of a position in material terms but can see and quantify positional factors as well. The top GM can not only see all important factors in a position [not just material] but can also search for and find continuations which exploit the advantages, defend the disadvantages, and generally select continuations which are consistent with the types of advantages/disadvantages in the position. Sometimes positions cannot be assessed by a GM in an intuitive manner but require, instead, quantitative assessment by the GM. The GM examines lines which help him/her to reach a correct assessment of the position. This, again, is a case where the position evaluation suggests lines to be examined but the final product is a better assessment of the position. Not only can a position evaluation help shape the searching strategy but it can work in reverse too. To me, "smart searching" is such that the continuations to be searched are selected based on position evaluation. A refinement for the future might be to revisit positions in a tree after subsequent evaluations provide new information relevant to these positions. Maybe this makes no sense to real chess programmers, who can easily see the flaws in the above. I hope that doesn't happen. Bob D.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.