Author: Bob Durrett
Date: 14:20:19 11/04/02
Go up one level in this thread
On October 31, 2002 at 20:17:37, David Hanley wrote: >Actually, as a human player, i found that one more quickly in my puzzle book >than a lot of others-- the bishop on f6 and rooks on the h-file are a common >"pattern" > >So i guess is question is which sacs don't fit the patterns. > >dave My gut feel is that Dave has just expressed a VERY profound observation! Generally, what is expected [by a GM] depends on what one [that GM] has seen before. This assumes that pattern recognition plays a dominant role in GM chess. What is needed is to devise a set of query searches of a large database, such as Megabase, to find successful sacrifices which do NOT occur in positions containing relevant often-recurring position-fragments [i.e. patterns]. Suppose, for example, that the following query search process were followed: (1) A new database is created from Megabase by deleting all games not containing sacrifices. (2) A second database is created from the first database by deleting all games in which the sacrifices were not successful. (3) Next, each position, in the second database, where the sacrifice was played would be examined to identify all familiar patterns in each of those positions. [Determining whether or not a pattern is "familiar" would involve additional searches (and maybe a lot of work) in the original Megabase and subsequent analyses.] (4) Next, a third database would be created from the second database by deleting all games in which the sacrifice(s) occurred in positions containing "relevant" familiar patterns. Note that "relevant" implies that the pattern being considered has something to do with the sacrificial move. (5) Next, a preliminary database of POSITIONS [i.e. not games] involving successful sacrifices could be created. (6) Finally, the preliminary database of positions could be scrubbed to remove uninteresting cases. The above process, or something like it, could then produce positions involving sacrifices which would not be suggested by familiar patterns in the position. The final set of positions might be grouped in some meaningful way to facilitate their subsequent study. 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.