Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / A Big Thanks and a Small Update
1 24 5 6 7 820 21 Previous Next  
Parent - By Banned for Life (Gold) Date 2012-01-09 06:25
There is crafty rotated bitboard code.

Nope. Vas wrote his own bitboard code prior to releasing Rybka.

And fruit code.

Nope. You're batting zero so far.

But maybe if you keep making these claims we'll all join you in your alternate reality field!
Parent - - By Vasik Rajlich (Silver) Date 2012-01-09 08:01

> There is crafty rotated bitboard code.  Always has been.  And fruit code.  Always has been too.

This is false.

Parent - - By Venator (Silver) Date 2012-01-10 16:06
This is false.

Oops, this must have come as a big blow for Bob :lol:
Parent - - By bob (Gold) Date 2012-01-10 18:12
Sure, just like his claim that no fruit was copied.  That ALL rybka versions were 100% original.  That Rybka was developed incrementally.  To name just a few.
Parent - - By Uly (Gold) Date 2012-01-10 22:41
I'm glad you're finally seeing the light :razz:
Parent - - By bob (Gold) Date 2012-01-10 22:50
Right.  Oops, wait.  Rybka 1.6.1 and earlier were NOT original. Neither was 1.0 beta through 2.3.2a.  So perhaps ONE of us is NOT "getting the point"???
Parent - By Venator (Silver) Date 2012-01-11 10:12
(1) Copied rotated bit boards code? WRONG.

(2) Copied Fruit PST's? WRONG.

(3) Pages of identical Fruit-Rybka code? WRONG.

(4) Copied Fruit bishop PST in Crafty? RIGHT.

It is clear you haven't seen the light yet.
Parent - By Venator (Silver) Date 2012-01-11 10:19
* No Fruit was copied: correct, Fruit ideas were used, the source code is original.

* All Rybka versions were 100% original: correct, from R1 beta all Rybka versions are original at the source code level.

* Rybka was developed incrementally: once again correct, I have even witnessed this myself when I visited Vas.

This still leaves the annoying fact for you that when I add 8 to the Fruit bishop PST, I get the Crafty bishop PST.
Parent - - By RFK (Gold) Date 2012-01-08 22:00
I wonder if any previous girlfriends of, Bob  have any private love letters of his,  that they would care to share with is wife? That is assuming he had any girlfriends prior to the relationship he had with his wife before they married. :wink:
Parent - - By bob (Gold) Date 2012-01-08 22:08 Edited 2012-01-08 22:10
And this has what to do with anything?  Had several girl friends.  Did not mail them.  I grew up in a small town.  Was easier and much faster to either walk over or call...

But again, the point would be???

BTW it is rather difficult to think of something as private to him when it so obviously contains MY code, don't you think.  I have not seen anyone even ADDRESS that part of the ICGA report as it really is irrefutable.
Parent - - By RFK (Gold) Date 2012-01-08 22:16
But it wouldn't  matter to your wife what they had to say, now would it! Because she is the genuine copy -as they say! eh! The very idea of anyone thinking of soliciting any previous thoughts you may have had toward them to your wife would be unconscionable.  You  could argue they were just practice for recognizing and interacting with the real thing.
Parent - By RFK (Gold) Date 2012-01-08 22:19
Are you getting my drift now Bob?!
Parent - - By bob (Gold) Date 2012-01-08 22:47
And that has what to do with the current discussion?
Parent - - By RFK (Gold) Date 2012-01-08 23:00
Don't play dense! Rybka 1.6.1 was not your love letter to read!
Parent - - By bob (Gold) Date 2012-01-08 23:11
Sure it was.  _I_ wrote MOST of it.

Hard to forget that small fly in the ointment, eh?  And it was PROVEN that I wrote it.
Parent - By RFK (Gold) Date 2012-01-08 23:34
Refer to my last post to you!
Parent - - By zwegner (***) Date 2012-01-08 20:13
OK, sounds fine. We can also discuss privately if you like.

Also, just in case you didn't see it, I posted another question a bit below. I had meant to post it in reply to you but replied to myself instead...

Parent - By Vasik Rajlich (Silver) Date 2012-01-08 20:25
My screen is a bit crowded. If I miss something please bump it.

Parent - - By AWRIST (****) Date 2012-01-08 19:44
In what traditions do you see that the private tournament director Deville sent your private pre-versions to the ICGA panel as a sort of evidence for copying guilt that they tried to prove?
On a ranking of evil (from 1 up to 10) where do you see Olivier Deville with this publication 5 years later?:wink:
Parent - By Vasik Rajlich (Silver) Date 2012-01-08 20:04
I don't really have a problem with that. It's quite normal to commit a small infraction to try to catch a bigger one, that has a wide ranging precedent.

Parent - - By Ray (****) Date 2012-01-08 22:02
Olivier committed a gross breach of trust by giving an author's private version to other people. He has shown that he is not to be trusted in future.
Parent - - By bob (Gold) Date 2012-01-08 22:13
If he violated ANYONE's trust, it would be mine.  That was MY code he sent us, you DO realize?

But since I have always released my source, I won't hold it against him at all.  He was trying to help the ICGA resolve the issue.  And he did help.
Parent - - By Ray (****) Date 2012-01-08 22:19
We have been here before... Those early versions did not compete in ICGA tournaments so were irrelevant. Doesn't matter of they were clean, dirty of whatever. They were of no concern to the ICGA, and Olivier was wrong to give them out. It is not as if a court served him with a court order or anything. It was merely the ICGA, an amateur organisation. Engine authors should be very wary of sending Olivier anything at all. He probably regrets his actions now, or at least he should.
Parent - - By bob (Gold) Date 2012-01-08 22:44
They were distributed, which violates Crafty's license.  They were entered in tournaments that violated rule 2 which all computer chess tournaments in the CCT/ACCA family have followed.  And they show a distinct propensity the author has to copy rather than create from scratch.

I realize you don't like it and want to dismiss it.  Because there is absolutely no way to refute it.  So you resort to the childish "it doesn't count."  I suppose I could contact Peter Skinner, who has been the CCT tournament director for a long time, and file a protest, and a brief rehash of the crafty/rybka case would THEN be highly relevant, if that would make you happier.  You can hardly argue that it can't be investigated since it DID play in CCT.  Or I could ask Olivier if he would like an investigation because of the ChessWar participation.  Either would again make it instantly relevant.

Find some other ploy than "it just doesn't count."  "prior bad acts used to show a pattern of behavior" are most definitely allowed, particularly in legal proceedings.

Olivier did NOTHING wrong.  Turns out he sent the thing to the original author.  That would be me.
Parent - - By Ugh (*****) Date 2012-01-08 23:27
if you "got" Vas on 2.3.2 you would not need to keep bringing up the old 1.6

now quit with being prissy and concentrate on ICGA violation ONLY

do you have a stand-alone case for rule #2 or not, if you do 1.6.1 is a sideshow, if not then we understand your desire to fly off at tangents
Parent - - By bob (Gold) Date 2012-01-08 23:34
I have said this previously.  1.6.1 is a "slam dunk".  1.0 beta is less of a slam dunk, but it is still a massively high probability it was copied.   2.3.2a is a little less certain.

Why?  Because 1.6.1 is a dead copy in all regards.  1.0 beta is a copy, but then there was the necessary translation to bitboards, so one has to go to the abstraction/filtration/comparison approach that is not needed in 1.6.1.  2.3.2a represents a significantly modified 1.0 beta, so the AFC becomes a little less clear.  Not that ALL are not absolutely convincing.  But 1.6.1 has zero chance for error.  1.0 beta is in the 99% range.  2.3.2a is lower, maybe 90%.  But together, they paint a clear and damning picture.

1.6.1 is NOT a tangent.  Just to those of you that don't want to have to deal with it...
Parent - - By RFK (Gold) Date 2012-01-08 23:35
Rybka 1.6.1 has no place in any discussion.
Parent - - By bob (Gold) Date 2012-01-09 05:45
Only for those that think copyright is meaningless, and honesty can fly out the window...
Parent - - By Albert Silver (***) Date 2012-01-09 14:16
I'm confused. I thought you would be bringing up arguments on what you felt were violations by Vas in regards to his participation in ICGA events. After all, let us suppose, just for argument's sake, that Rybka 1.6 was a verbatim copy of Crafty from A to Z, but that all of the Rybkas that actually participated in ICGA events did so legally, what then? You might have a bone to grind, but the ICGA investigation and condemnation would still be a sham.

Inferring he might have done something wrong now because of something in the past is all nice and well, but unless you actually have something concrete to discuss on the Rybkas that participated in the ICGA events, all this Rybka 1.6 talk just looks like a giant smoke screen to cover up a lack of evidence with relevant versions.
Parent - - By bob (Gold) Date 2012-01-09 17:28 Edited 2012-01-09 17:33
1.  Rybka 1.6.1 was a dead copy of MY code.   In fact, 1.4 through 1.6.1 were.

2.  Rybka 1.x (pre 1.0 beta versions) participated in CCT and ChessWar to name just TWO public uses that violate Crafty's license agreement.

3.  1.4 through 1.6.1 show a CLEAR start on the idea that "copying code is OK" and was therefore relevant to the ICGA investigation.  Particularly considering Vas' repeated claim that ALL versions of Rybka, including the pre-fruit versions, were 100% original.

4.  1.0 beta uses Crafty's rotated bitboard code verbatim.  THAT does violate ICGA rule 2 because according to Vas, the rotated bitboard code remained in place until he replaced it with Pradu's magic bitboard stuff in version 4.0. 

5.  This clear and continual copying is, simply, unacceptable.  We see a pattern, dating right back to the beginning of Rybka, that shows a clear disregard for rules, laws, licenses, and ethical behavior.

6.  We DID present evidence showing that at the very least, ALL rybka versions through 2.3.2a were not compliant with ICGA rule 2.  And 2.3.2a DID compete in an ICGA event...

Finally, it would seem that to you (and a few others) the fact that he copied code right and left, violated licenses, violated copyright law, would all be perfectly OK if he did not use the code in an ICGA event.  But just sold it and stayed out of ICGA tournaments?  Better give that "thought process" due consideration, because it is an incredibly unethical way of accepting bad behavior.  However, Rybka DID compete in the ICGA events, which is what triggered the ICGA becoming involved.  Even if we had found NO evidence to show copying in versions that participated in ICGA events, don't you think what happened would STILL be unethical?  I find this kind of "but if he didn't participate, it wouldn't matter."  Unethical behavior ALWAYS matters to people that have ethics.
Parent - - By Banned for Life (Gold) Date 2012-01-09 19:22
4.  1.0 beta uses Crafty's rotated bitboard code verbatim.  THAT does violate ICGA rule 2 because according to Vas, the rotated bitboard code remained in place until he replaced it with Pradu's magic bitboard stuff in version 4.0. 

This is not the case. Vas developed his own rotated bitboard code in 2004, and used this in R1B and later versions of Rybka. As you've pointed out, he does use your Hankel matrix based indexing system, but this idea is not protected by copyright or license.
Parent - - By bob (Gold) Date 2012-01-10 17:54
If he uses THAT, he uses the code.  It won't work with ANY OTHER code.  Get it?  Why don't you do some reading to see how rotated bitboard work, before chiming in without having a clue?  I can point you to a free technical paper on my web page if you don't have access to the ICCA journal issue where it was published...
Parent - - By Ugh (*****) Date 2012-01-10 18:10
well, it seems to me that there are really only two realistic ways to store, let's say the shift counts for a 45 degree rotated matrix.

Let's say you store the diagonal matrix SW-NE into 64 bits, starting with the first diagonal (1 bit only in bottom right corner), then the next, working your way across the board, then you get one series of shift patterns. Or you might start at the top left corner. That's 2 patterns.

True if you obfuscated the storage and inserted the diagonals in random order, you'ld get many more patterns, but who would want to try that? Just confusing and pointless.

Occams razor says two realistic ways to store, so of course, comparing two programs is evens chance those patterns match anyway. Compare a pool of 20 programs and there will be matches all over the place, not connected to "copying" in any way. Simply there's a limited way to do it. And your abstraction-filtration-comparison procedure would throw the match out at filtration stage.

No copying proven. QED.
Parent - - By bob (Gold) Date 2012-01-10 18:17
Are you conversant in rotated bitboards?  One doesn't JUST rotate left or right 45 degrees for bishops.  Do some research on your own.  Look up "Dark Thought" and see how they did it.  Dark Thought actually did what you said would not be done, BTW, because they could then use a cute indexing formula to set/clear a bit in the "rotated" (I called them "scrambled" but the idea is exactly the same) bitboard without needing the 64-word array to do this.

So much you DON'T know.  Big surprise, there.

And, BTW, some rotate +45, -45, some use +45, +135, if they use traditional rotation as I defined in Crafty.

And then there are rooks.
Parent - - By Ugh (*****) Date 2012-01-10 18:40
Yes I am conversant in bitboards thank you very much. Experimental internal version of CSTal III was entirely bitboard, I quite enjoyed writing it. And yes it did work.

You are just trying to confuse everybody as usual. Obviously there's more than one bitboard, you also need to do the opposite rotations, but the idea of storing the diagonals (whichever rotation direction and there are not many) sequentially is clearly going to be done one-by-one, in obvious sequence and that leads to TWO obvious possibilities per rotated bitboard.

Thus two obvious shift pattern possibilities per bitboard. And almost guaranteed matches if you compare early stage bitboard programs. True that third or fourth generation bitboard developments might have thought up some different ways, but you are comparing Vas early efforts with your early efforts. Bingo - they match. No copying involved.

And then there are rooks.

hahahahahaha!!! desperate to throw in some irrelevant confusion!!!

Yes, there will be some more identical shift patterns for those too.
Parent - By bob (Gold) Date 2012-01-10 19:51
Sorry.  2005 was NOT my early effort.  Crafty was bitboard in 1994 when it started.  Rotated came the following Summer.  And MY rotated bitboard code has been rewritten MANY times.  None of which looked similar.  Each new revision was done to make it faster, or more efficient in terms of memory usage, or more amenable to both 32 and 64 bit architectures.  For a diagonal lookup, how many bits do you use for the index?  The correct answer can be zero, 1, up to 8.  Currently I use a max of 6.  And previous versions squeezed out the bit the bishop stood on to cut it to 5.  Or less for shorter diagonals.  There are MANY ways to do the indexing.  Based on your comments, you've not given it any thought at all.
Parent - By Rebel (****) Date 2012-01-10 19:56 Edited 2012-01-10 20:00
You might want to learn to apologize first when you accuse FALSELY.

That table is not R1
Parent - - By Rebel (****) Date 2012-01-09 20:41
ICGA HYATT - 1.0 beta uses Crafty's rotated bitboard code verbatim.

Like the alleged table you claimed in :

Surprise, it's not there.
Parent - - By Ugh (*****) Date 2012-01-09 20:49
Bob's problem is that when he believes something strong enough he imagines it is ok to invent the data to support it
Parent - - By Rebel (****) Date 2012-01-09 21:36
Like the PST case.

IMO he got the point just recently, the other option as he claimed is much worse.

I am in a mild mood today.
Parent - - By bob (Gold) Date 2012-01-10 17:53
Nothing has changed in the PST case at all.
Parent - - By Rebel (****) Date 2012-01-10 19:33
Except your understanding of it.
Parent - - By bob (Gold) Date 2012-01-10 19:41
My understanding has not changed at all.  As I said, I RAN Zach's code to verify that the numbers matched Rybka or Fruit perfectly, depending on which set of constants were used.  When this came up, I already had the code handy and posted in almost instantly.  Took 30 seconds to modify fruit to dump it's PSTs to show that they matched the numbers from Zach's code.  Etc.

My understanding of what was in the report has not changed.  Perhaps YOU finally understood what "exact match" meant?  As in "The code zach provided produces an EXACT match for the rybka numbers when using the rybka constants, and an EXACT match for the fruit numbers when using the fruit constants?"  Appears you finally DID get that, after a few months.
Parent - - By Rebel (****) Date 2012-01-10 20:47 Edited 2012-01-10 20:56

> My understanding has not changed at all.

1. Then why all your ramblings about equal PST's especially in the beginning of the PST debate (July 2011) ?

2. Why did you leave Rybka forum in an embarrassed state of mind 2 for days (see the CB graph) when Miguel found a semantically equal Fruit PST in Crafty to come up 2 days later with a lame explanation when there was absolutely no need for you explain anything at all?

Try to explain that in the light you fully understood the PST issue.

You have seen the confusion you created among the good Rybka folks and how most of them BELIEVED the Ryba PST equalled the Fruit PST's since YOU promoted that.

It's all here on the first part of:

An apology would be the appropriate thing to do.

Until then that text stays there.
Parent - - By bob (Gold) Date 2012-01-10 22:23
2. Why did you leave Rybka forum in an embarrassed state of mind 2 for days (see the CB graph) when Miguel found a semantically equal Fruit PST in Crafty to come up 2 days later with a lame explanation when there was absolutely no need for you explain anything at all?

Simple question:  Liar or stupid?

Why?  Original post was made by user "Michiguel"  at Date 2011-08-11 23:19.  Note.  8-11 is ALMOST over.  My next post on the subject was the next morning, 8-12 at 11:45am.  There were several others that same day.  So I "ran and hid in embarrassment"?  You REALLY should, since this is simply yet another false statement on a large pile of existing false statements.  BTW there were 26 posts made by me on the day I was "hiding".  Care to explain that?  Your typical "sloppy research" where you trust some silly graph by someone ELSE rather than doing as the panel did and looking at the data for ourselves?  I did only post in maybe two "batches" that day, once around lunch, once later.  So not even 12 hours missing. rather than your 48 hours?  And you really wonder why I keep using the "liar" keyword?  Use the forum search, user=bob, min age=140, max age=150, and those 26 posts will be on the first two pages that supposedly do not exist there.  And there are posts on the 13th as well.

Sorry to once again rain on your parade, but I am SURE you are getting used to it?

Facts speak louder than bullsh**...

So, waiting pointlessly for a reply...

BTW, only ONE of us "runs and hides."  That would NOT be me.
Parent - - By Rebel (****) Date 2012-01-10 22:31
So you challenge the 48 hour aspect and leave the real issue untouched ?

Coming from you that's almost a confession :razz::razz::razz::razz:
Parent - - By bob (Gold) Date 2012-01-10 22:49
What is the "real issue".  You asked one point I have answered a hundred times by now about the "equal PSTs".  I did not say the VALUES matched exactly.  In fact, I ORIGINALLY made an incorrect statement, saying one could take the Rybka PSTs and divide an entire table by the same single constant to produce Fruit's numbers.  I quickly CORRECTED that and have not said otherwise, since.

The other issue is you claim I ran and hid and did not post for two days.  Absolutely false.  That is not a "point"?  Not to you, of course.  Any inaccuracies from you are acceptable, but I (and any other ICGA person) have to spell every comment carefully using only words with two syllables or less?  So, in your peculiar method of debate, "you ignore your second (incorrect) point completely and leave that untouched?

Coming from YOU, that is almost expected behavior.

In any case, for the record, did I "run and hide in embarrassment" or was that false?
Parent - - By Rebel (****) Date 2012-01-11 00:14
It's really simple, why did you felt the need to defend and explain yourself how that Fruit PST Bishop table ended up in Crafty, you went through your notes of years ago and came up with a long explanation only to show why the values in that PST matched semantically.

Why did you do that?

It can only be explained one way, that you at that time did not understand what the documents stated about the PST's.

It further explains all your wrong statements promoting Fruit PST values = Rybka PST values.

The story and explanation semantically fit so to say.

Agree ?

I am happy to remove that page.
Parent - - By bob (Gold) Date 2012-01-11 00:22
Answer:  someone raised the question.  I had to dig back thru a zillion archives to find the code from when we tested the bishop PSTs.  I was a bit sloppy and don't always remember to include a "reason" for a specific version, so I had to hunt.  Took me a day and about 40 DVDs worth of stuff to remember what we were doing.  I responded because someone asked the question. 

I do not have any idea what you mean by "my wrong statements promoting Fruit PST values = Rybka PST values"  I can't think of any single post where I have EVER said that If you examine any Fruit and Rybka PST for the same piece, all values are identical.  I have said many times, that the fruit PST initialization code will produce either Fruit or Rybka PST values EXACTLY with the constant changes Zach gave.

I will say it again.  Before the ICGA report was made public, I had CAREFULLY verified Zach's statements by running the code that I posted immediately after that particular question was raised.  That was before any ICGA decision or anything.  I ALWAYS knew what the report from Zach, and Mark W, and the final report said.  Why you want to suggest otherwise is beyond me.  Sort of like that silly RE challenge that proved exactly nothing, and wasted about 30 minutes of my time to find what you wanted, then another hour of my time to interleave the C and ASM to show how it matched EXACTLY...
Parent - - By Rebel (****) Date 2012-01-11 00:47
Not correct, here is another one, July 27, 2011

There was NO CODE COPYING for the PST issue.  NONE.  NADA.  ZILCH.  ZIPPO.  It is the PST VALUES themselves that were copied.

How can you possibly say that while the document is about copying code ?

And the good folks at Rybka forum believed you that Rybka PST values = Fruit PST values, literally as stated by you above.

And when they later found that this was not the case they were mad at you.

Can you blame them ?

I don't know why you want to stick with it, it only damages your case and it makes mine stronger.
Up Topic Rybka Support & Discussion / Rybka Discussion / A Big Thanks and a Small Update
1 24 5 6 7 820 21 Previous Next  

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill