Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / Bit masks and copyright
- - By bob (Gold) Date 2013-09-16 18:57
You are the one claiming that the bit mask that contains 1's for any possible friendly pawn square not BEHIND the rook is copyrightable.  That is the ONLY difference in the rybka and fruit rook evaluations.  The executable code for Rybka does not change one scintilla to add that bit to that "uncopyrightable table" as you call it.

Learn to read and follow the discussion.  You are arguing with your OWN point.  Not with one I made.  I quote:

PST tables are not an issue for the same reason that the piece values are not an issue. They represent pure chess knowledge and the less creative input from the developer, the better.

The bitmask is used to implement a different algorithm (i.e. exclude pawns behind the rook).

These are very different vis-a-vis copyright law...


My response is "pure bullshit."  They are equivalent, not "very different".  Only a REAL idiot would try to make that argument.
Parent - - By Banned for Life (Gold) Date 2013-09-16 19:07
You are the one claiming that the bit mask that contains 1's for any possible friendly pawn square not BEHIND the rook is copyrightable.

Don't be an idiot. I would never claim that a bit mask can be protected by copyright. However, use of different bit masks in a short code sequence implementing a different algorithm would weaken the claim of copyright infringement by another party. It's not my fault you're too stupid to see the difference.

My statements are correct and you are a moron.
Parent - - By bob (Gold) Date 2013-09-16 22:52
You said "they are very different vis a vis copyright law."  They are NOT "very different".  They are identical.
Parent - - By Banned for Life (Gold) Date 2013-09-16 23:04
Wrong. They are very different. One is a set of natural values representing chess knowledge, similar to the digits of pi, or temperature values. The other is representation of an algorithm. Neither can be protected by copyright, but they are not identical. They are not even related...
Parent - - By bob (Gold) Date 2013-09-17 05:48
they are IDENTICAL.

EACH contains data for 64 squares.  One contains 0's and 1's to indicate where pawns must be or must not be.  One contains integers to indicate where pawns must be or must not be.  They are absolutely identical.  Only you would want to argue that ridiculous point.  NEITHER is "the representation of an algorithm."
Parent - - By Banned for Life (Gold) Date 2013-09-17 07:07
A PST is a STATIC representation of chess knowledge. What is the average value (usually given as a differential value) of a piece being on a particular square? If 32-bit tablebases were available, these values could be determined using a multiple regression approach. This could be done with much less accuracy by doing a regression analysis on millions of positions, where each position has an evaluated score. Or this can be done the way Larry does it, using his subject matter expertise. Or this can be done the way you do it, by just guessing. In any event, there is absolutely nothing creative about this table. The better it represents the underlying chess knowledge, the stronger the engine will be, and vice versa. The bottom line is that a PST represents a closed form estimate for the differential value of a particular piece on a particular square. Once this table is created, it never changes other than possibly being blended with another static table. The blending constitutes a simple algorithm, but the PSTs are generated a priori and don't change during a game, or during a sequence of games. As engines progress, PST values will converge, just as piece values have.

The pawn bitmask for awarding rook mobility values is used to generate a DYNAMIC representation of the value of the rook. It is NOT part of a closed form solution, but is a small part of an algorithm for evaluating the position. The algorithm for when to award a bonus for a half open file, e.g. whether to do this when the rook is in front of the pawns or not, is clearly algorithmic, and clearly dependent on the position in the game. Ideally the bit mask would change to reflect the current position. The fact that this isn't currently being done doesn't detract from the bitmask's algorithmic nature. It only indicates the algorithm might not be optimal. Unlike with PSTs, there is no reason to believe that there is only one optimal style for playing chess, and there is also no reason to believe that there is an optimal method of setting mobility.
Parent - - By bob (Gold) Date 2013-09-17 17:10
Pawn masks are NOT "dynamic".  They are exactly as static as PSTs.  And in fact, PSTs are not NECESSARILY static, many of us in the past had dynamic PST values that were adjusted before the start of a search.  Both are static evaluation components however, not a dynamic thing about either.

BTW there is no "rook mobility" in this discussion.  Mobility is quite a bit different from a simple "rook on open or half-open file".
Parent - - By Ugh (*****) Date 2013-09-17 17:19
He didn't say pawn masks are dynamic, he said pawn masks are used to generate a dynamic representation of value. I'm not sure where mobility crept in from, probably he meant open file evaluation, but typed mobility.

That's one each for you.

1-1 draw.

Meanwhile in the other match, Team Ed 17 (Hyatt own goal 12) Team Bob 0
Parent - By bob (Gold) Date 2013-09-17 18:34
Totally disagree.  Do not PST values do EXACTLY the same thing?   used to generate a dynamic representation of value?

There is absolutely no difference, except bitmaps are 1 bit per square of information, where a PST can be arbitrarily large numbers of bits per square...
Up Topic Rybka Support & Discussion / Rybka Discussion / Bit masks and copyright

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill