Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / "Secrecy and exclusion - to control"
- - By Ugh (*****) [fr] Date 2013-10-31 19:53
You don't get it, do you? There is only one excuse for secrecy and exclusion - to control.

Last time you excluded critics, failed to provide an open process, selectively leaked Hyatt opinions and "facts", kept discussions secret, had a secret panel within the panel and failed to provide evidence documents during the process. All designed, including your propagandising outside of the process, to get a guilty verdict. Only later, to accompany the wedding photograph PR shit in the international press, did you part release some documents, still keeping the panel discussion secret and coupled with idiotic PR about FSF  court cases and huge copyright damages claims in Polish courts. To my mind you should be in jail for this.

Open Justice, the default position for ALL justice, would have made all evidence and forum discussions available at the time. Full document disclosure and read capability. Purpose? To keep the "judges" under judgement. If you don't understand or you deny that, you are either pig ignorant or a mafia capo.

This time, you're outdoing even the previous disaster. Total blackout. Why? Control obviously. Ed is asking entirely reasonable questions. But meets a brick wall of silence. We don't trust you. Get it?
Parent - - By bob (Gold) [us] Date 2013-10-31 20:44
Then WHY did you advocate "secrecy and control" as the proper way to conduct the Rybka investigation.  Do you know the meaning of the word "consistency"?  Apparently not.  I talked about Rybka but shouldn't have.  I didn't talk about loop but should have.  Really consistent.

We are in "blackout" because you guys complained bitterly that I discussed the Rybka case before the panel was finished.

Pick a side and stay on it for at least 24 hours...
Parent - - By Ugh (*****) [fr] Date 2013-10-31 21:11
We guys complained bitterly because you EXPRESSED OPINIONS before the panel was finished. You should have PUBLISHED DATA, for example, the Watkins and Zach documents, at the same time they were released into the panel. The panel deliberations should have been readable. The transparency would have been some sort of check on your biased behaviour.

In other words, you were propagandising for a verdict of guilty. Not seeking truth. That's what we complained about. No open justice. That's what we complained about.

But since you fail to understand (either pig ignorant or mafia capo) you fail to address the actual problem. Going into total blackout doesn't solve the problem, which is two fold. First it increases the concern expressed by Ed that the Loop case has elements of the ICGA investigating the ICGA, and in secrecy. Second, that you and the other secretariat member are demonstrated not fit for purpose, and that YOU need to be observed by all and given supervision, not possible under your blackout conditions.
Parent - - By bob (Gold) [us] Date 2013-10-31 21:22
Sorry.  I published a LOT of data during the panel investigation, because I had discovered a LOT of data before the panel investigation was started.  Ditto for Zach's stuff which I had seen many times prior to the protest.

The loop case is ongoing.  I really don't give a rat's a$$ as to what Ed thinks about the investigation.  He did not initiate it, he is not taking part in it, so he can wait to see the outcome when it is made public.  End of that story.  All the speculation, guessing, assuming and other nonsense will have absolutely no effect on the final resolution.  The old expression "like it or lump it" comes to mind.  We are following what we believe to be the correct approach.  I don't see where the ICGA charter says we have to satisfy either you or Ed here...
Parent - - By Ugh (*****) [fr] Date 2013-10-31 21:49
You published a mass of PROPAGANDA during the panel investigation. Zach's paper was not published until afterwards. The panel discussions on Zach's paper were not published at all, and still are not published. It is clear to me you never read Zach's paper, else you would not have tried to persuade me it contained 6x64 identical PST values, when even a cursory read shows 11x64 values, not identical. You hadn't even registered 11 tables and thought it had 6. You repeated your 6 several times, 6 is the natural number to assume, but it is incorrect. For the person supposed to be supervising the report, that showed an astonishing lack of knowledge of it, don't you think? Your demonstrable lack of knowledge, and thus of having read it, gets worse, as you know.

We know you don't give a rat's arse as to what other computer chess community people think. So you operate without consensus, without agreement, having screwed up the time before, with dubious mixed messages and several years of delay, and all in secret.

Pig ignorance or mafia capo?
Parent - - By bob (Gold) [us] Date 2013-10-31 22:01
Sorry, Zach's stuff was up on a website for years.

There was a LONG discussion on CCC about this when it first started.  Just because you are not aware of what was publicly available doesn't mean nothing was available...

Pig ignorance or just plain old-fashioned ignorance?

BTW, if you want to quote me, quote me CORRECTLY.  I said "I don't give a rat's a$$ about what Ed thinks.  Nowhere in that did I mention "what other computer chess community people think", just "one specific one."
Parent - - By Ugh (*****) [fr] Date 2013-11-01 17:45
The paper was not published and you never read it. It wasn't published until afterwards, and you only read it after it came under criticism. How else do you explain your repeatedly telling me there were 6*64 identical values in the PST's, when it isn't 6, it's 11 and they were not identical anyway. You'ld need to have read the paper to know that though, would you not? Pretty fundamental.

*if* you had used open justice, this kind of incompetence would have been picked up at the time, but, instead, you were allowed to blunder onwards with no oversight.

Open justice is an important safeguard against judicial bias, unfairness and incompetence, ensuring that judges are accountable in the performance of their judicial duties.

And now, second time round, total secrecy. And, having made such a clown of yourself last time, incompetent, biased, not fit for purpose - you are it again, second time, even worse. How is anybody to take your secret Secretariat seriously?
Parent - - By bob (Gold) [us] Date 2013-11-01 19:02
Continue to show your ignorance.  Zach's work has been public for several years..  WAY before the rybka investigation.  If you read Fabien's letter, HE posted this on January 23:

A few things I noticed yesterday, can you confirm?
- Rybka search info was obfuscated in some way (like displaying depth-3 or something), any pointers on details please?
- Vasik claimed that Strelka 2.0 is a clone of Rybka 1.0 (and you know what that would imply!)
- Zach Wegner found many Fruit ideas (and nearly identical code) in Rybka 1.0; I think someone else did, too
- Some even stronger open-source program appeared as a decompilation of Rybka (with own ideas, sounds familiar), what came up of looking at those?


So he was asking about Zach's efforts which WERE PUBLICLY AVAILABLE. 

He wrote this in his formal protest to the ICGA:

Since then it has emerged from highly respected sources like Zach Wegner, Bob Hyatt and others that there is a lot of evidence that has been accumulated over the last few years that Rybka 1.0 beta is a derivative of Fruit 2.1.

Zach Wegner has presented evidence of alleged significant copied/derived Fruit evaluations in Rybka 1.0 beta here: https://webspace.utexas.edu/zzw57/rtc/eval/eval.html


Follow that link to see what was available, you might be surprised, you WILL be educated.

Zach's stuff was around a LONG time before Fabien filed his protest.  A LONG time before.  I read it as it was written, because I was involved with Christophe and Zach (and others) when this first popped up.  The original discussions were on CCC in fact, and can easily be found.

Whether you take it seriously or not is irrelevant, IMHO.  You just want to complain, so that no one takes you seriously at all.
Parent - - By Ugh (*****) [fr] Date 2013-11-01 19:33
Ah, if I hadn't been banned from reading your panel stuff then I might have noticed. Nevertheless you did NOT operate to Open Justice last time, critics were kept in the dark, nobody still knows what the panel discussed and you clearly had not read that Zach document (your claim of 6 PST tables being the first sign of null-reading).

Your Secretariat, last time, was not fit for purpose, yet you are still in position. And now, the second time, the same incompetents, operating in total secrecy, telling critics "you don't give a rat's arse".

Pig ignorance or mafia capo. Why else have you gone secret?
Parent - - By bob (Gold) [us] Date 2013-11-01 21:09
You didn't have to be on the panel to know about that web site.  EVERYBODY knew about it.  The discussion went on for a couple of years as the Rybka evidence was compiled by Zach...

As for secrecy, you criticized me for saying too much during the investigation.  This is what you get in response, EXACTLY what you claimed you wanted back then...
Parent - - By Ugh (*****) [fr] Date 2013-11-01 23:00
My criticism was your SELECTIVE leaking of your biased opinion and factoids (designed to convict Vas) whilst keeping other material secret. My criticism was the lack of Open Justice.

My solution is Open Justice.

Your solution is more secrecy.

We don't use secret justice in fair and democratic systems. Dictatorships and mafias use secret justice. So, I repeat, you are either pig ignorant of how to behave fairly, or, you are behaving as a mafia capo.
Parent - - By bob (Gold) [us] Date 2013-11-02 02:16
As opposed to leaking what YOU selectively want to know.

Real consistent.

There is PLENTY of "secret justice" in the world today.  In the US.  In the UK.  Etc.

My solution is simply to find out what happened, nothing more, nothing less.

cheating and justice are not exactly related.  Cheating doesn't violate criminal law in the case of ICGA rule 2, necessarily.   It is just cheating.
Parent - - By Rebel (****) Date 2013-11-02 06:56 Edited 2013-11-02 06:59
We know of course your definition of cheating, 2 recent quotes from you.

Bob - According to ME, taking a source program, making detailed notes about everything that is done, and then using that to "write" your own code is NOT ORIGINAL.

Bob - Taking notes CAN capture exactly what the code does. That IS "copying code".

Then we read.

Bob - I've tested everything in Glaurung 2, fruit 2, toga 2, and a lot of other ideas as well

You went forwards and backwards through Glaurung, Fruit and Toga, figured out what the code does and tried it in Crafty.
Parent - - By bob (Gold) [us] Date 2013-11-02 15:49
To correct your gross dishonesty, here is the COMPLETE post:



PostPost subject: LMR revisited.    Posted: Wed Apr 01, 2009 11:08 am  Reply to topic Reply with quote Edit/Delete this post  Report Post
I have been experimenting with LMR and various ways one might do it more intelligently. And after about 4 weeks of solid testing, I have produced exactly -nothing- that helps. I have tried things from using a complex ordering early in the tree, and then using either one or two "late" counters, one for PV nodes, one for non-PV nodes (ala glaurung). Doesn't help one bit. Just doing the counters without the ordering. nada. I've tried (again) fruit-like history counters. Either no better or much worse depending on the threshold set. I've tried a more localized history counter, like failhi[ply][from-to] and faillo[ply][from-to] so that the values don't get completely scrambled. Again, either no better or worse depending on how those values are used.

I'm still testing, but so far, looking at several programs that do LMR, absolutely none of the ideas I have seen make any difference at all in testing, unless they make the program weaker. I have tested with very fast games, 1+1 games, 5+5 games, and even a few 60+60 tests (those turn into 3 hours + per game, which means I can play about 256 games every 3 hours, which is really too slow to be usable. But I wanted to make sure that the things I was trying were not just tested on very fast games where they might not matter.

I've tested everything in Glaurung 2, fruit 2, toga 2, and a lot of other ideas as well (some given above, not all). This appears to be an idea that either works, or it doesn't. Tuning seems to either hurt or have no impact.


That was about LMR, it was ONLY about LMR, you know that, but you want to dishonestly IMPLY that it means I went through everything.  I looked at EVERYONE's LMR code.  At the time all were using history counters.  To this day I do not use history counters with LMR.  You are one giant fraud, snipping things out of context and trying to paste them together to say something you want them to say, as opposed to providing the entire quote which is not so helpful in your arguments.

Keep trying.  You continue to be a fraud, and many realize this...
Parent - - By Ugh (*****) [fr] Date 2013-11-02 16:10
hey, hey, hey, you're going slightly nuts here ....

Ed quoted from your post "I've tested everything in Glaurung 2, fruit 2, toga 2, and a lot of other ideas as well" and provided a link to the entire post (in blue).

What exactly is "fraudulent" about that? Nothing. I think the point he is making is that you go grubbing around in other sources looking for ideas and testing them, Vas also did, presumably everybody at the leading edge of chess program development does, and why not? It's not illegal and you indeed encourage using ideas. I think you wrote somewhere that RE is fine to discover ideas also. Or am I misremembering?

Thus, Ed's complaint is the dual-standard used. Why single out Vas, when you yourself do much the same ....
Parent - - By bob (Gold) [us] Date 2013-11-02 17:59
The IMPLICATION is fraudulent.  I did not test ALL the ideas in glaring, fruit, etc.  I looked EXPLICITLY at their LMR stuff since I could find no history implementation that worked.  I eventually proved to myself that the history part of Fruit's LMR code did not work either.  I commented out just that part of fruit and it played exactly at the same Elo on cluster testing.    I've not tested "everything" in glaurung, fruit or toga, JUST the LMR ideas they used relative ONLY to history counters.

I don't do "much the same thing" as Vas.  I didn't go "grubbing around in other sources".  The discussions were ongoing at the time in CCC, and Tord (and others) mentioned the things they were doing, which were very close to what I was doing (and which was, therefore, very close to that which was not working."  WITH regard to LMR/history.

This is completely unrelated to the "I went forward and backward through fruit" as stated by Vas...

It was, quite simply, another example of dishonesty that he CLAIMS to dislike, yet uses all the time.
Parent - - By Ugh (*****) [fr] Date 2013-11-02 18:09
"The IMPLICATION is fraudulent.  I did not test ALL the ideas in glaring, fruit, etc."

hey, Don Quixote, that windmill isn't there. He never said you did.

The post went like this:

Are you denying you wrote this?
Bob - I've tested everything in Glaurung 2, fruit 2, toga 2, and a lot of other ideas as well

In what sense is this then, not true?
Ed - You went forwards and backwards through Glaurung, Fruit and Toga, figured out what the code does and tried it in Crafty.

I guess you are just trying the derail the thread as per usual, by filling it with pointless arguments about nothing. Don't you get told off for that on CCC? I think so.
Parent - - By bob (Gold) [us] Date 2013-11-02 19:46
Believe what you want.  NO LMR idea from fruit or glaurung is in Crafty.  Tord posted that he took my final history idea and used that in Glaurung, even though I removed even that from Crafty since it didn't work.

Totally irrelevant to rybka/fruit case.
Parent - - By Ugh (*****) [fr] Date 2013-11-02 20:03
Did either Ed or I ever say there was LMR Fruit or Glaurung in Crafty? Well, for sure, I never even commented and Ed said you TESTED ideas. Another windmill that isn't there. You are deliberately making arguments where none exist. Why is that?
Parent - - By bob (Gold) [us] Date 2013-11-02 20:08
Ed said I copied ideas, which he claims is cheating by my definition (that is NOT my definition, by the way).  No ideas I tried remain, they did not work.

Another lie, of course...
Parent - - By Ugh (*****) [fr] Date 2013-11-02 20:40
You're just being silly now. Ed has NOT accused you of cheating. He is using the discrepancy between your two statements:

Vas copied ideas therefore he cheated.
Bob copied ideas but did not cheat.

to demonstrate that, by your own reasoning, the first statement (Vas cheated) can't be true. For some reason you are trying to twist this into Ed saying that Bob cheated. Presumably the usual thread derail by pointless argument that you are so fond of.
Parent - - By bob (Gold) [us] Date 2013-11-03 02:13
First, here is MY statement, from the post I quoted above:

I'm still testing, but so far, looking at several programs that do LMR, absolutely none of the ideas I have seen make any difference at all in testing, unless they make the program weaker.

Now please feel free to tell me what "ideas" I copied.  Seems like the statement is pretty clear. "Absolutely none of the ideas I have seen make any difference." 

So

(a) what ideas did I copy?  Precisely which ones?  That would be "none".  So his claim is 100% false.

(b) where has ANYONE said copying an idea is cheating?  Ideas != code.

Please read before you post nonsense.  And that is all your post is in this case.
Parent - - By Ugh (*****) [fr] Date 2013-11-04 12:20
Yes, we all know ideas are not code. What exactly are you whingeing about? Ideas are information and information moves freely. You have a philosophical disagreement with Ed about the ethics and lawfullness of reverse engineering, but that's another matter not particularly relevant to Rybka/Fruit or Fruit/Loop/Toga which is all about open source.

So where is your problem? You, like everybody, looks at and tests ideas found elsewhere. Sometimes you use the idea and sometimes not. That's ok.

What Ed is complaining about is that you made Vas into a special case. He has somehow done something with ideas, according to you, that is ethically different to what you have done with ideas. It seems to be that your claimed "difference" is that he didn't understand the idea(s) like you do, and somehow (well, by copy/modify) went from Fabien written code to Vas written code. Using the process of taking the entire Fruit 2.1 source and modifying it. Whereas you go from code to idea back to your own written code.

You seem to be saying you "understand" what you are doing, but Vas, who was a "newbie" didn't .....
Parent - By bob (Gold) [us] Date 2013-11-04 14:27
You are absolutely wrong.  Feel free to show me anywhere I have claimed that Vas broke rule two by copying ideas.  Never once said that.  I said he broke rule two by copying code.  Zach spent his time on the eval, which involved non-literal copying since Fruit was mailbox and rybka was bitboard.  I initially looked at the older versions and Crafty where literal copying was done.  Now we have the literally copied hash table store code that bypasses the non-literal copying discussion completely.

And again, there is NO history "idea" from Fruit present in Crafty.  There is no history idea in crafty whatsoever.

And no, it is not about my understanding what I am doing while Vas was a newbie.  It is about Vas copying the work of others, modifying it as necessary to fit within the bitboard framework, and then calling that original.  It is anything but original.
Parent - - By Rebel (****) Date 2013-11-02 20:46

> I think you wrote somewhere that RE is fine to discover ideas also. Or am I misremembering?


You are not misremembering, Bob is not against RE.

>Thus, Ed's complaint is the dual-standard used. Why single out Vas, when you yourself do much the same ....


Exactly.
Parent - - By Ugh (*****) [fr] Date 2013-11-02 21:05
I approve of Bob's concepts of total freedom of ideas and information. Read it, Reverse engineer it, discuss it, whatever.

But he gets into difficulties when it comes to then using the idea. It seems one Bob rule is, code it so it looks different enough. Another Bob rule is don't use the ideas in the same order. And another rule is sprinkle some ideas from other sources around a bit. Than it's called original, even though no idea within is original.
Parent - - By Rebel (****) Date 2013-11-03 00:37
I approve of Bob's concepts of total freedom of ideas and information. Read it, Reverse engineer it, discuss it, whatever.

Bob's rules are totally unclear to me.

Take RE for example, he supports it, then he says:

Bob - Taking notes CAN capture exactly what the code does. That IS "copying code".

What else is RE than figuring out what the code does?
Parent - By Ugh (*****) [fr] Date 2013-11-03 20:19
I think he has "taking notes" on some sort of grey scale. Black (bad) note taking involves writing down or memorising everything. White (good) notetaking involves extracting the idea and writing down or remembering the idea only.

In the white version, when you implement the idea, you need to have forgotten everything else, whilst also simultaneously remembering everything else to make sure your code is not the same as what you're supposed to have forgotten.
Parent - - By bob (Gold) [us] Date 2013-11-04 02:13
I don't see what is unclear.  Using an idea from someone else is perfectly acceptable.  Unless you take it to the extreme where you take their code, take detailed notes about EVERY idea, how they are put together, how they work together, etc, and then use THAT to write a program that matches your notes accurately.  That is NOT original.  If you take a single idea that someone else discovers, it is often quite difficult to use in your existing program.  Too many differences.  If you take ALL the ideas and the structure, then that problem is not there.

I personally believe you understand this exactly.  It's not rocket science at all, it is simply "ethics 101".  Even if none of your ideas are original, if one comes from A, another from B, another from C, you STILL have a lot of work to make them work together effectively.  As opposed to taking ALL ideas from one program where you would implement them in the same way, called in the same way, etc.  That is NOT so hard to do and is certainly not original as required by rule 2.

Can we get off of this nonsensical discussion?  It is going nowhere.  You are completely and intentionally trying to distort what I have written to further your agenda.  I've not varied one bit in my stance on what is original and what is not.  There is most definitely a grey area in the middle.  But Rybka is not a grey area.  It is WAY over the line.  Ditto for Houdini regarding the robolito family.  Etc.  Should we find one that is "grey" it would be problematic.  That's not happened here.  Just the hash code would violate rule 2 without any of the eval code that has been presented.  And that's not "grey" at all, it is pure black and white.
Parent - - By Ugh (*****) [fr] Date 2013-11-04 10:48
Basically, your whole argument depends depends on the theory that Vas took Fruit 2.1, methodically converted it to bitboards, thoroughly tested that he had a completely accurate bitboard copy (same moves, same scores etc etc) and then modified, added, deleted chunks of it, and released it as Rybka in December 2005.

To do this he would need to have created a bitboard data structure on top of Fruit with the original mailbox data structure still in place, written test routines to check for integrity of the two data systems operating side by side, with two makemove's and undomove's and two genmove's and two movelists and two evaluations, each pair being integrity checked as the conversion took place. The search function is not so bitboard dependent, so that might have been possible to bodge with its own internal tests as the conversion proceeded, or just rewritten at the end. On completion of the first stage (identical bitboard copy), he throws away the old functions.

I'm guessing you would consider the other conversion option, of converting line by line, destroying the old version gradually in the process as too difficult because of the inability to thoroughly test each stage.

The problems with your argument are multifold.

First, the asm (or derived source) code evidence for copy/modify source is very weak. The >=0.0, the "quirk" of OPENFILE-HALFOPENFILE, and much later the unresolved hash code. That's it. As both Zach and Watkins stated there is no direct copied code.

Second, the time frame. Five and a half months cannot be enough to make this conversion AND add substantially more ELO.

Third, given the time constraints, you must consider the alternative, which is the existence of a prior developed engine with "ideas" coded in from various origins, including Fruit.

Fourth, Fruit "ideas" are not original. If Rybka uses a similar "idea" there is no way to demonstrate in the absence of matching code (asm or derived source) that the similar idea came from Fruit. For example, the iconic mobility function.

Essentially there is no direct opyright infringement found and no sufficient proof of copy/modify. Maybe Vas did, and maybe Vas didn't. For some people the 0.0 is overwhelming, for others it is a "so what?". My opinion, only an opinion of course, is that there is no way enough for a legal case (no FSF, no punitive damages a la Watkins Polish court case, no Fabien case, even after several years), that the "plagiarism" charge is just a fanciful piece of nonsense designed to get round the lack of copyright infringement.

Feel free to ramble, but in my opinion, again, the best anyone on your side can say is "we're suspicious, but case not proven", insufficient compelling evidence despite five years of hunting. And, if a "guilty" Vas gets off free, in your opinion, then that is the price you have to pay to live in a fair and civilised world.
Parent - By bob (Gold) [us] Date 2013-11-04 14:38
Please don't lecture me on what Vas must have done to convert fruit to bitboard.  I've already done that.  Cray Blitz -> Crafty.  I know EXACTLY what has to be done.  And I know that the new program is not original at that point.  However, over time, Crafty evolved to have nothing of Cray Blitz left, other than perhaps the FEN parsing code.  Took 10 years to get to that point.  I'd assume Rybka might reach that point as well, since I did.  But not with 1.0 beta through 2.3.2 at least.  Since 2.3.2 has been shown to have directly copied, literally copied code that comes from Fruit.

I don't consider the "other option" to be too difficult, as it is exactly how I developed Crafty.  You can go back to late 1994 rec.games.chess.computer and find discussions about bitboards as I first wrote the move generator, non-rotated bitboards, following the explanation given by Slate in "Chess Skill in Man and Machine".  I worked on nothing but the move generator for 2-3 months, posting the code in r.g.c.c several times for feedback.  I tested it by itself.  When it was tested and optimized, I moved on to the next piece, the search.  Again a complete rewrite because CB used an iterated search, but I wanted to use a recursive search since the code is easier to read and is cleaner.  That approach DOES work, as it worked for me.

What do you mean "unresolved hash code?"  Does "unresolved" mean "copied" or "unacceptable"?  If so, I agree, since that code, by itself, violates ICGA rule 2 directly, literally.

Your 5 1/2 month quote is wrong.  We had the 1994 ACM event in November.  By the end of December Crafty was actually playing chess on ICC.  By June 1995 it was beating Grandmasters at blitz regularly.  5.5 months is a long time to do this.

An individual "idea" is not a problem.  The problem becomes when two supposedly original programs use the same ideas, in the same ways, in the same order, with the same quirks.  That crosses the line, and that is what we found in Rybka.

No direct copyright infringement?  Can you look at the hash code and say with a straight face "This was not copied"???
Parent - By bob (Gold) [us] Date 2013-11-03 02:14
There is a difference between "ideas" and "code".
Parent - - By Rebel (****) Date 2013-11-02 19:34

>I looked at EVERYONE's LMR code.


Yepper, and by your own definition that's copying code.

LMR is worth 100 elo (see Don in the same thread and confirmed by myself) and more provided you have the formula well for single, double and triple reductions. And you did not get LMR to work by yourself so you decided to scrutinize Fruit, Glaurung and Toga.

That's cheating according your own given (above) definition.
Parent - By bob (Gold) [us] Date 2013-11-02 19:43
I had LMR working, ON MY OWN.  I simply had no luck getting the history counters to work with it.  Are you REALLY that thick?  Read the thread you quoted me from.  I started on LMR due to discussions on CCC, not on someone's code.  Sorry.

First, I don't do triple deductions.  Just 1 or 2.  3 has not worked for me.  I believe _I_ posted the LMR gain, not you or Don.  I ran the tests on my cluster.   And it wasn't 100 Elo.  If you don't do null-move OR LMR, if you add either, you get +80.  When you add the other, you get +40. +120 for BOTH, +80 for just one (doesn't matter which one). 

That's not cheating according to my definition.  The HISTORY stuff I looked at did not work, and is not now nor has it been in Crafty.  Sorry.

Try again.

And by all means, continue the intentional mis-quoting...
Parent - By bob (Gold) [us] Date 2013-11-03 02:16
then choke on this for a bit:

I'm still testing, but so far, looking at several programs that do LMR, absolutely none of the ideas I have seen make any difference at all in testing, unless they make the program weaker.

Absolutely no ideas nor code were copied.  Why don't you try reading FIRST???

And absolutely no cheating either.  Continue the fallacious arguments however, it is amusing to point out the contradictions and/or false statements,
Parent - By bob (Gold) [us] Date 2013-11-04 02:17
Sorry, but what _I_ got to work was 100% my original code.  Not one piece of an idea from Glaurung, fruit or toga.  And I have told you this many times.  Any idiot can look at Crafty's search.c, this specific code:

/*
************************************************************
*                                                          *
*   Now it's time to try to reduce the search depth if the *
*   move appears to be "poor".  To reduce the search, the  *
*   following requirements must be met:                    *
*                                                          *
*   (1) We must be in the REMAINING_MOVES part of the move *
*       ordering, so that we have nearly given up on       *
*       failing high on any move.                          *
*                                                          *
*   (2) We must not be too close to the horizon (this is   *
*       the LMR_remaining_depth value).                    *
*                                                          *
*   (3) The current move must not be a checking move and   *
*       the side to move can not be in check.              *
*                                                          *
************************************************************
*/
          if (Piece(tree->curmv[ply]) != pawn ||
              !Passed(To(tree->curmv[ply]), wtm)
              || rankflip[wtm][Rank(To(tree->curmv[ply]))] < RANK6) {
            extensions = -LMR_min_reduction;
            if (moves_searched > 3)
              extensions = -LMR_max_reduction;
            max_extensions = Max(depth - 1 - LMR_remaining_depth, 0);
            if (extensions < -max_extensions)
              extensions = -max_extensions;
            if (extensions)
              tree->reductions_done++;
          }
        }

BTW the comment above is not quite accurate, I DO reduce checking moves if SEE < 0 for that move.  Been doing that for quite a while.  Please point out the history counter usage above.  Please find ANY history counters in Crafty anywhere.  Otherwise your argument is simply a lie, because no idea NOR code was copied.  Doesn't exist in Crafty.

Another discussion that should die the death it deserves, you are simply wrong, again.  Since I didn't copy ANY idea from those programs, I also didn't (using your distorted definition) copy any code either...
Parent - - By Ugh (*****) [fr] Date 2013-11-02 18:20
Bob Hyatt:
As opposed to leaking what YOU selectively want to know.
Real consistent.
There is PLENTY of "secret justice" in the world today.  In the US.  In the UK.  Etc.
My solution is simply to find out what happened, nothing more, nothing less.
cheating and justice are not exactly related.  Cheating doesn't violate criminal law in the case of ICGA rule 2, necessarily.   It is just cheating.


Yup, plenty of secret justice. Guantanamo. Special renditions. Espionage cases. Charles I Star Chamber. You're in good company.

Sometimes secret courts are used to deal with Mafia, but it isn't the case this time, is it? Or is it in reverse, with you as the mafia capo?
Parent - - By bob (Gold) [us] Date 2013-11-02 19:44
No mafia at all.  YOU complained about my discussing things during the Rybka case.  Hence forth, I won't discuss anything (except within the panel) until the final verdict is rendered and the result made public.  Like it or not, that's the way it is.  Blame yourself.
Parent - - By Ugh (*****) [fr] Date 2013-11-02 20:07
Final verdict about what? You investigation is so secret, I don't actually know what you are investigating. Was Fabien's ACTUAL complaint letter ever published? Two days ago you were a grand jury (which is not a final verdict making process), now you're back to court case and final verdict. It's a Schroedinger investigation!
Parent - - By bob (Gold) [us] Date 2013-11-02 20:10
So you know we are investigating something, but you don't know WHAT we are investigating, yet you are complaining about it nonetheless???  :)

You know the topic.  Ed has mentioned it multiple times...
Parent - - By Ugh (*****) [fr] Date 2013-11-02 20:31
As far as I know the only information is that Fabien made a formal complaint to the ICGA containing an allegation that LOOP contained code from Fruit. But I don't think the actual complaint letter has ever been published (in contrary to case Rybka).

So, no, I don't know what the actual complaint is, just a part of the complaint, called an "allegation". There may be more parts. Have you seen the actual complaint letter/email? Sorry for being so suspicious, but I learnt never to assume and always to check the raw data.
Parent - - By bob (Gold) [us] Date 2013-11-03 02:22
All I'm willing to say, based on past criticism, is that there is an investigation concerning loop/fruit...  We've tried to contact the author repeatedly, through several different ways, with zero success.  We are now moving along toward a determination.
Parent - - By Rebel (****) Date 2013-11-03 08:17
That's no answer to Chris question.

Have you seen the Fabien complaint about LOOP or not?
Parent - - By bob (Gold) [us] Date 2013-11-03 15:52
Why would I answer any question about an ongoing investigation?  You'd just complain more.
Parent - - By Ugh (*****) [fr] Date 2013-11-03 16:25
A little "poetically licensed" intermezzo for you, Bob. History repeats or rhymes or something .....

Mr C: I worked on the XYZChessProgram. Mr A was not a programmer, he was the chess expert who provided some of the heuristics. The program was developed by me using mailbox but the final product had to use bitboards. The conversion was done by Mr B.

Interviewer: Does it mean, that the later XYZChessProgram_2 was programmed by someone else? Perhaps Mr B, but based on your earlier work?

Mr C: I worked on the XYZChessProgram_2 as well as the XYZChessProgram.

A different interview:

Interviewer: You Mr A, had an essential part in the XYZChessProgram project. XYZChessProgram was not only very well playing and winning the World Championship, he was also the best chess .. for many years. What was the secret?

Mr A: As I said, Mr B. was a genius programmer. I do not recall the mailbox-bitboard conversion that Mr C. mentioned. I am not saying he is wrong, but:
a. converting code between the two is enormously difficult - it is easier to write something from scratch
b. Mr B. was very much his own man, and went about his programming tasks in his own way. So although it is quite possible that Mr B. looked at Mr C's code and discussed the structure of the program, I have always felt that the XYZChessProgram was very much Mr B's original work.
Parent - - By bob (Gold) [us] Date 2013-11-04 02:05
I have no idea what that is supposed to parody.  Not a clue.
Parent - - By Ugh (*****) [fr] Date 2013-11-04 13:21
Well, apart from the poetic license of mailbox/bitboard, this is a David Levy quote concerning a slightly parallel situation in the 1980's:

As I said, Mr B. was a genius programmer. I do not recall the ....... conversion that Mr C. mentioned. I am not saying he is wrong, but:

a. converting code between the two is enormously difficult - it is easier to write something from scratch

b. Mr B. was very much his own man, and went about his programming tasks in his own way. So although it is quite possible that Mr B. looked at Mr C's code and discussed the structure of the program, I have always felt that the XYZChessProgram was very much Mr B's original work.


which is kind-of amusing, because "our" side uses that argument about Rybka and "your" side says not. Situation itself is no big deal, a far as I know which is very little of course, but the argument used about the originality is relevant, I think.

Mr B = Mark Taylor
Mr C = David Broughton
XYW Chess program = Chess Champion Mark V

There was no internet back then, so this info is from wiki's and reported interviews .....

references:
https://chessprogramming.wikispaces.com/Intelligent+Software
http://www.schach-computer.info/wiki/index.php/Levy,_David
https://chessprogramming.wikispaces.com/David+Broughton

Wiki,  David C. Broughton,
a British computer chess programmer. His first chess program Vega, written on the Z80 North Star Horizon machine, had finished 2nd behind Sargon at the 2nd PCW-MCC tournament held in London in November 1979, and he was soon recruited by David Levy and Kevin O’Connell to work for their companies Philidor Software and since 1981 Intelligent Software. He worked on the project Philidor which 8086 port emerged to Parker Chess, and to the SciSys Chess Champion Mark V. Broughton’s work for the Mark V was in the form of Z80 assembly. As the Mark V had a 6502 processor, the program had to be extensively translated and rewritten. So credit for the Mark V program has to go both to Mark Taylor and David Broughton, though David Levy is known to credit the Mark V to Taylor.

Interview, David Broughton: I worked on the Chess Champion Mark V (manufactured by Scisys) under the chess direction of David Levy and Kevin O'Connell. David Levy was not a programmer: he was the chess expert who provided some of the heuristics. The program was developed by me using my own home computer which had a Z80 processor but the final product had to use a 6502 CPU. The conversion was done by Mark Taylor. The openings book was provided by David Levy.

Interview, David Levy: As I said, Mark Taylor was a genius programmer.
I do not recall the Z80-6502 conversion that David Broughton mentioned. I am not saying he is wrong, but
a. converting code between the two is enormously difficult - it is easier to write something from scratch; and
b. Mark Taylor was very much his own man, and went about his programming tasks in his own way. So although it is quite possible that Mark looked at David's code and discussed the structure of the program, I have always felt that the Mk V was very much Mark's original work.
Parent - By bob (Gold) [us] Date 2013-11-04 14:24
I don't see what converting from one assembly language to another has to do with this.  If it was converted, it would be non-original.  If it were written from scratch, but carefully following the original program's data flow, organization and such, it would be non-original.  If I threw Crafty out and started from scratch, the new program would look similar to Crafty, but there would be thousands of differences, and almost certainly no identical code, assuming I did not intentionally copy anything.

So what does this have to do with the current discussion?
Up Topic Rybka Support & Discussion / Rybka Discussion / "Secrecy and exclusion - to control"

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill