Bob asks: Then why don't YOU come up with a rule 2?

General discussion about computer chess...
User avatar
Chris Whittington
Posts: 437
Joined: Wed Jun 09, 2010 6:25 pm

Re: Bob asks: Then why don't YOU come up with a rule 2?

Post by Chris Whittington » Sun May 24, 2015 6:49 pm

Rebel wrote:Chris,

Isn't it fascinating how this search [grab all] vs eval [red flag] line of thinking has developed? When I showed him my 6th row code which is 100% semantical equal with Fruit but then again is also is 100% semantical equal with my "caught bishop on the 7th row" I was merely showing how simple chess knowledge may end up in identical code.

And immeditely after followed by the usual spanking, nuking and bullying.

Applying Bob's logic Fabien is guilty plagiarising my caught bishop on the 7th row I wrote in the early 80's, it's 100% semantical equal. Perhaps you could make the same claim.

It's too silly for words.
for eval - they strip the weights and then compare. if they did that for search (weights like R=2 etc) they (all programs) would be almost completely identical.

why this? partly because they built search idea onto search idea over many years, but basically the whole thing is one big copy.
but the same is true of eval, one new idea added every now and again, eval is mostly one big (ideas) copy. there's more scope for mix and match with eval, and a lot of scope for tinkering. basically, they are claiming their eval mix-n-match plus the tinkering is "original". and they just "forget" about the search similarities.

why this? I guess because they are programmers. not chess players. they "understand" search, its part of the computer science. But they don't really understand the eval, to them the tinkering mix-n-match is an "art", their creative input.

hyatt
Posts: 1242
Joined: Thu Jun 10, 2010 2:13 am
Real Name: Bob Hyatt (Robert M. Hyatt)
Location: University of Alabama at Birmingham
Contact:

Re: Bob asks: Then why don't YOU come up with a rule 2?

Post by hyatt » Sun May 24, 2015 6:57 pm

Chris Whittington wrote:
Rebel wrote:Chris,

Isn't it fascinating how this search [grab all] vs eval [red flag] line of thinking has developed? When I showed him my 6th row code which is 100% semantical equal with Fruit but then again is also is 100% semantical equal with my "caught bishop on the 7th row" I was merely showing how simple chess knowledge may end up in identical code.

And immeditely after followed by the usual spanking, nuking and bullying.

Applying Bob's logic Fabien is guilty plagiarising my caught bishop on the 7th row I wrote in the early 80's, it's 100% semantical equal. Perhaps you could make the same claim.

It's too silly for words.
for eval - they strip the weights and then compare. if they did that for search (weights like R=2 etc) they (all programs) would be almost completely identical.

why this? partly because they built search idea onto search idea over many years, but basically the whole thing is one big copy.
but the same is true of eval, one new idea added every now and again, eval is mostly one big (ideas) copy. there's more scope for mix and match with eval, and a lot of scope for tinkering. basically, they are claiming their eval mix-n-match plus the tinkering is "original". and they just "forget" about the search similarities.

why this? I guess because they are programmers. not chess players. they "understand" search, its part of the computer science. But they don't really understand the eval, to them the tinkering mix-n-match is an "art", their creative input.

That is simply naively wrong. Searches don't look anything like each other. Too many variables. Too many different ways to do things. EVERYBODY does parallel search and that introduces 100X as many different choices to make. All you have to do is compare a couple of open-source programs to see how wrong you are.

hyatt
Posts: 1242
Joined: Thu Jun 10, 2010 2:13 am
Real Name: Bob Hyatt (Robert M. Hyatt)
Location: University of Alabama at Birmingham
Contact:

Re: Bob asks: Then why don't YOU come up with a rule 2?

Post by hyatt » Sun May 24, 2015 7:05 pm

Rebel wrote:Chris,

Isn't it fascinating how this search [grab all] vs eval [red flag] line of thinking has developed? When I showed him my 6th row code which is 100% semantical equal with Fruit but then again is also is 100% semantical equal with my "caught bishop on the 7th row" I was merely showing how simple chess knowledge may end up in identical code.

And immeditely after followed by the usual spanking, nuking and bullying.

Applying Bob's logic Fabien is guilty plagiarising my caught bishop on the 7th row I wrote in the early 80's, it's 100% semantical equal. Perhaps you could make the same claim.

It's too silly for words.

And I explained that "copying code is copying code". But that if you copy ONE eval term, nobody would raise a red flag. BTW I had the trapped bishop code in 1995 before ANY commercial program used it. I got accused of adding a "cheapo" eval term to rack up easy wins by several ICC program operators (Rebel 8 or so, chess tiger, genius, etc). I added the term to stop Crafty from playing a stupid move, but as soon as I did it realized it could stop defending the a-pawn (main pawn here although h pawn on occasion) because the opponent should not capture it. And EVERYBODY did capture it.

Here's the code from the oldest version I have, circa 1995:

if (square == A7) {
if (And(mask_b6b7,BlackPawns) &&
Attacked(B6,0)) score-=BISHOP_TRAPPED;
}

I do not think that if anyone sees THAT term in ANY program it will cause any concern. It is trivial. White bishop at A7, black pawn on B6 or B7, and square B6 attacked by black (so that if black has pawn on B7 and plays B6 the bishop can't take it, or if black has already played B6 it is defended and can't be taken).

I don't know when you added it, but I can tell you for 100% certain that when _I_ added it you were not using it, or it was badly broken. And I can also tell you that if you did the same thing as I did above, nobody would care. UNLESS they looked at your eval and you did the same thing I did in several different terms, in the same order. Then it would become quite suspicious even though by itself it means nothing at all.

You guys need to get off this 0/1 bandwagon. Either you zoom in microscopically, or you zoom out so far "playing chess makes all programs nearly identical." There are other zoom ranges to consider. Ala' the "filtration" the two of you seem to (a) harp on all the time but (b) don't seem to understand in the least...

User avatar
Chris Whittington
Posts: 437
Joined: Wed Jun 09, 2010 6:25 pm

Re: Bob asks: Then why don't YOU come up with a rule 2?

Post by Chris Whittington » Sun May 24, 2015 7:07 pm

hyatt wrote:
Chris Whittington wrote:
Rebel wrote:Chris,

Isn't it fascinating how this search [grab all] vs eval [red flag] line of thinking has developed? When I showed him my 6th row code which is 100% semantical equal with Fruit but then again is also is 100% semantical equal with my "caught bishop on the 7th row" I was merely showing how simple chess knowledge may end up in identical code.

And immeditely after followed by the usual spanking, nuking and bullying.

Applying Bob's logic Fabien is guilty plagiarising my caught bishop on the 7th row I wrote in the early 80's, it's 100% semantical equal. Perhaps you could make the same claim.

It's too silly for words.
for eval - they strip the weights and then compare. if they did that for search (weights like R=2 etc) they (all programs) would be almost completely identical.

why this? partly because they built search idea onto search idea over many years, but basically the whole thing is one big copy.
but the same is true of eval, one new idea added every now and again, eval is mostly one big (ideas) copy. there's more scope for mix and match with eval, and a lot of scope for tinkering. basically, they are claiming their eval mix-n-match plus the tinkering is "original". and they just "forget" about the search similarities.

why this? I guess because they are programmers. not chess players. they "understand" search, its part of the computer science. But they don't really understand the eval, to them the tinkering mix-n-match is an "art", their creative input.

That is simply naively wrong. Searches don't look anything like each other. Too many variables. Too many different ways to do things. EVERYBODY does parallel search and that introduces 100X as many different choices to make. All you have to do is compare a couple of open-source programs to see how wrong you are.
Ideas-wise, most searches are equivalent. Same as evals, except eval has more mix and match possibility, and lots of scope for tinkering.

Hyatt, don't bother answering. Your 'tis, 'tisn't-s are getting far too repetitive over four years, and way too boring. Nobody believes your version of events anymore, however many times you want to deny the obvious. There is NO copied code in Rybka, no copyright violations. Don't bother to reply.

User avatar
Chris Whittington
Posts: 437
Joined: Wed Jun 09, 2010 6:25 pm

Re: Bob asks: Then why don't YOU come up with a rule 2?

Post by Chris Whittington » Sun May 24, 2015 7:10 pm

hyatt wrote:
Rebel wrote:Chris,

Isn't it fascinating how this search [grab all] vs eval [red flag] line of thinking has developed? When I showed him my 6th row code which is 100% semantical equal with Fruit but then again is also is 100% semantical equal with my "caught bishop on the 7th row" I was merely showing how simple chess knowledge may end up in identical code.

And immeditely after followed by the usual spanking, nuking and bullying.

Applying Bob's logic Fabien is guilty plagiarising my caught bishop on the 7th row I wrote in the early 80's, it's 100% semantical equal. Perhaps you could make the same claim.

It's too silly for words.

And I explained that "copying code is copying code". But that if you copy ONE eval term, nobody would raise a red flag. BTW I had the trapped bishop code in 1995 before ANY commercial program used it. I got accused of adding a "cheapo" eval term to rack up easy wins by several ICC program operators (Rebel 8 or so, chess tiger, genius, etc). I added the term to stop Crafty from playing a stupid move, but as soon as I did it realized it could stop defending the a-pawn (main pawn here although h pawn on occasion) because the opponent should not capture it. And EVERYBODY did capture it.

Here's the code from the oldest version I have, circa 1995:

if (square == A7) {
if (And(mask_b6b7,BlackPawns) &&
Attacked(B6,0)) score-=BISHOP_TRAPPED;
}

I do not think that if anyone sees THAT term in ANY program it will cause any concern. It is trivial. White bishop at A7, black pawn on B6 or B7, and square B6 attacked by black (so that if black has pawn on B7 and plays B6 the bishop can't take it, or if black has already played B6 it is defended and can't be taken).

I don't know when you added it, but I can tell you for 100% certain that when _I_ added it you were not using it, or it was badly broken. And I can also tell you that if you did the same thing as I did above, nobody would care. UNLESS they looked at your eval and you did the same thing I did in several different terms, in the same order. Then it would become quite suspicious even though by itself it means nothing at all.

You guys need to get off this 0/1 bandwagon. Either you zoom in microscopically, or you zoom out so far "playing chess makes all programs nearly identical." There are other zoom ranges to consider. Ala' the "filtration" the two of you seem to (a) harp on all the time but (b) don't seem to understand in the least...
You get off your band wagon.

There is NO copied code in Rybka.
There are no copyright violations in Rybka.
The COMP-EVAL comparison completely failed to do ANY filtration. Filtration is required at EVERY abstraction stage.

Hyatt, don't bother to reply. Your denials are unsatisfactory and rejected in advance.

User avatar
Rebel
Posts: 515
Joined: Wed Jun 09, 2010 7:45 pm
Real Name: Ed Schroder

Re: Bob asks: Then why don't YOU come up with a rule 2?

Post by Rebel » Sun May 24, 2015 8:28 pm

hyatt wrote:
Rebel wrote:Chris,

Isn't it fascinating how this search [grab all] vs eval [red flag] line of thinking has developed? When I showed him my 6th row code which is 100% semantical equal with Fruit but then again is also is 100% semantical equal with my "caught bishop on the 7th row" I was merely showing how simple chess knowledge may end up in identical code.

And immeditely after followed by the usual spanking, nuking and bullying.

Applying Bob's logic Fabien is guilty plagiarising my caught bishop on the 7th row I wrote in the early 80's, it's 100% semantical equal. Perhaps you could make the same claim.

It's too silly for words.

And I explained that "copying code is copying code".

But that if you copy ONE eval term, nobody would raise a red flag.
But you did anyway, as bitten by a snake.

Your whole rule #2 and the interpretation as practiced in the R/F case --> one BIG OBSTACLE for progress.

This is my last reply, Mark is no longer interested and I have gathered enough new material. Especially the lines of thinking and the difference therein (pure abstracted code comparison vs looking what the code exactly evaluates) was an unexpected (negative) surprise. Speaking of unexpected negative surprises, the lack of an answer on the double pawn by Mark (apparently) not able to admit an Itsy Bitsy Teenie Weenie mistake I can not understand. And oh, before I forget there was Bob's confession for not allowing Chris to the Panel.

BB+
Posts: 1484
Joined: Thu Jun 10, 2010 4:26 am

Re: Bob asks: Then why don't YOU come up with a rule 2?

Post by BB+ » Sun May 24, 2015 10:24 pm

Rebel (to hyatt) wrote:With your influence you could ask Jaap (since he wrote the contract with Nintendo) for the person to contact at Nintendo and provide Fabien with that information.
Even assuming that Jaap wrote the contract (in 2007), I don't see any reason for him to know "the person to contact" at Nintendo 8 years later. I would suspect that FSF/Letouzey would be quite capable of achieving the desired contact.
Rebel wrote:And when that happens the CC division of the ICGA will be in big trouble.
I am not sure what this "CC division of the ICGA" is? In 2002 the ICCA became the ICGA, and since then has tolerated computer chess having a special role, due to its history with the organisation. But there's no reason for the ICGA to retain this priviliged place for computer chess, other than the necessity of organising the WCCC under Section 1 of Article II of the by-laws. [For that matter, 2002 would have been a logical time for the ICGA to de-affiliate with FIDE, given the broader focus].
Rebel wrote:n a normal world we would fire such as man,
I'm not sure how one would "fire" the president of an assocation, usually you would oust such a person by a vote (eg, Cock de Gorter with the CSVN).
Chris Whittington wrote:You can vote for me!
Section 3 of Article II states: All candidates for election must have been members of the ICGA for no less than two (2) years, and I should think that would mean continuously up to the time of the election.
Chris Whittington wrote:c. make a complete review of the icga-Rybka case
You must mean the Fruit/Rybka case (or Letouzey/Rajlich), as the "Rybka v ICGA" case was something heard by the FIDE Ethics Commission, and I'm not sure how/why the ICGA would review this.
Rebel wrote:Don't try to make Fruit the mother of all chess engines, it for 90% relies on the work, ideas, publication of others.
The reason why Fruit keeps coming up is that it was the "mother" of Rybka, in the sense of eval. I guess one could say that it was nice of Letouzey to "collate" some segment of computer chess knowledge (use this, don't use that), for Rajlich to re-use.
Chris Whittington wrote:Maybe there is a bit of divergence as search has developed over time, but basically all searches are semantically the same, probably in the same order as well.
Chris Whittington wrote:if they did that for search (weights like R=2 etc) they (all programs) would be almost completely identical.
This is complete nonsense. One could compare Stockfish to IvanHoe (for instance), and see the difference. The pruning conditions can differ, the null move conditions (static threat detection), the extension conditions (singular moves, passed pawns, checks), the LMR conditions (history, refutation), how exclusion search is used in determining singularity, qsearch use of hash, ... The order in which one does various prunings/reductions is also variable. There are other methods, such as prob-cut, which haven't caught on, but appear in some engines. Even Rybka and the IPPOLIT had a number of differing search peculiarities, though I'd call them in the same class. Not to mention the implementation details, of splitting functions into pv-search, near-root-search, mid-depth search, low-depth search, qsearch, pv-qsearch, exclusion search, ... or keeping it simple with just search/qsearch and usng if statements.
Rebel wrote:Mark is no longer interested
Hmm, I told you I was going to be busy, and I did other things on the weekend. Looking at another 50+ posts in the other thread, now I know why people tend to hate Mondays.

hyatt
Posts: 1242
Joined: Thu Jun 10, 2010 2:13 am
Real Name: Bob Hyatt (Robert M. Hyatt)
Location: University of Alabama at Birmingham
Contact:

Re: Bob asks: Then why don't YOU come up with a rule 2?

Post by hyatt » Sun May 24, 2015 11:43 pm

Chris Whittington wrote:
hyatt wrote:
Chris Whittington wrote:
Rebel wrote:Chris,

Isn't it fascinating how this search [grab all] vs eval [red flag] line of thinking has developed? When I showed him my 6th row code which is 100% semantical equal with Fruit but then again is also is 100% semantical equal with my "caught bishop on the 7th row" I was merely showing how simple chess knowledge may end up in identical code.

And immeditely after followed by the usual spanking, nuking and bullying.

Applying Bob's logic Fabien is guilty plagiarising my caught bishop on the 7th row I wrote in the early 80's, it's 100% semantical equal. Perhaps you could make the same claim.

It's too silly for words.
for eval - they strip the weights and then compare. if they did that for search (weights like R=2 etc) they (all programs) would be almost completely identical.

why this? partly because they built search idea onto search idea over many years, but basically the whole thing is one big copy.
but the same is true of eval, one new idea added every now and again, eval is mostly one big (ideas) copy. there's more scope for mix and match with eval, and a lot of scope for tinkering. basically, they are claiming their eval mix-n-match plus the tinkering is "original". and they just "forget" about the search similarities.

why this? I guess because they are programmers. not chess players. they "understand" search, its part of the computer science. But they don't really understand the eval, to them the tinkering mix-n-match is an "art", their creative input.

That is simply naively wrong. Searches don't look anything like each other. Too many variables. Too many different ways to do things. EVERYBODY does parallel search and that introduces 100X as many different choices to make. All you have to do is compare a couple of open-source programs to see how wrong you are.
Ideas-wise, most searches are equivalent. Same as evals, except eval has more mix and match possibility, and lots of scope for tinkering.

Hyatt, don't bother answering. Your 'tis, 'tisn't-s are getting far too repetitive over four years, and way too boring. Nobody believes your version of events anymore, however many times you want to deny the obvious. There is NO copied code in Rybka, no copyright violations. Don't bother to reply.
Getting no more irritating than your continual "most searches are equivalent" which is pure nonsense.

Unless you abstract out to the point all you recognize is alpha/beta minimax search. THEN I would agree. but that loses 99% of the important details.

Post Reply