Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / A Big Thanks and a Small Update
1 2 3 4 5 620 21 Previous Next  
Parent - - By Vasik Rajlich (Silver) Date 2012-01-08 20:20

> The question is, can you expand your "is" by "is and always was" ?


Can you rephrase this?

> Do you consider that Houdini is a clone (or a derived clone)? Do you think Houdini is an original engine that doesnt have any rybka code?


I don't know. I haven't investigated any of the closed source programs (and I am not planning to).

Vas
Parent - - By cipri (**) Date 2012-01-08 21:02
I'm very happy that you said that you dont know if Houdini is clone or not (or if it has any kind of rybka code),
because some people here in the forum are praising Houdini saying that it is for sure original code without any rybka code (e.g. Arrière Pensée).

I have an interesting question.

If you would now at that moment, know nothing about chess-programming, but having the intention to write a chess engine. How would you start?
Would you be looking very closely at the source code of all strong open source engines and try to combine as much as possible all of their strong points?
Or would you be reading books/articles where just the main ideas are described but not the full source code?
Parent - - By Vasik Rajlich (Silver) Date 2012-01-08 21:18
I always like to try to learn things from a fundamental point of view. For example at the moment I am studying poker. For me this means learning basic equilibria in "endgame" situations (ie. short stacks, short handed) and then expanding from there. Most players are quite different. They like to just dive in and go from there.

Computer chess is really quite a unique beast though. There aren't that many general principles and it's probably not worth it to spend too much time on them. If I was starting from scratch today, I'd probably take Stockfish or Robbolito and just pick them apart.

Vas
Parent - - By Banned for Life (Gold) Date 2012-01-08 21:58
For example at the moment I am studying poker.

Excellent! No doubt you will be competing in the World Series of Poker in the near future...
Parent - - By Vasik Rajlich (Silver) Date 2012-01-08 22:08
There aren't many poker players who don't play the WSOP! :smile:

Vas
Parent - By OleM (**) Date 2012-01-10 14:58
Well, there are poker players, and people who play poker. Those that don't play the WSOP probably fall in the second category I guess.
Parent - - By Venator (Silver) Date 2012-01-11 09:48
at the moment I am studying poker.

I have told Hans, he surely will be interested in this development!
Parent - - By Vasik Rajlich (Silver) Date 2012-01-11 16:14
Maybe I'll finally come visit you guys for some tournaments. Poland isn't exactly a poker hotbed. :smile:

Vas
Parent - - By Venator (Silver) Date 2012-01-11 16:39
Ah, that would be cool!
Parent - - By Vasik Rajlich (Silver) Date 2012-01-11 17:01
Although Hans might not think so when he can't pay rent next month. :smile:

Vas
Parent - - By Alkelele (****) Date 2012-01-11 17:13
Guys, come to Copenhagen! The girls are blonde and ****, and we could be playing for a currency that's actually worth something :-)

That's 3-0 in my book (but we must insist on Hans buying a return ticket in advance).
Parent - By Vasik Rajlich (Silver) Date 2012-01-11 17:27
That sounds like a plan! I was never really into the blonde thing, but 2-0 is good enough. Send me an email. :smile:

Vas
Parent - - By Venator (Silver) Date 2012-01-11 17:35
The girls are blonde and ****

What does **** mean, 'ugly' !?  :lol: :lol: :lol:
Parent - - By Vasik Rajlich (Silver) Date 2012-01-11 17:38
:lol:

Vas
Parent - - By Venator (Silver) Date 2012-01-11 18:09
I hope **** means 'rich', so we can have a decent payroll  :lol:
Parent - - By Vasik Rajlich (Silver) Date 2012-01-11 18:31
That has its pros and cons. Nothing in this world is free. Maybe Dagh meant "easy".

Vas

ps. The term is "bankroll". You've been using corporate lingo for too long. :smile:
Parent - By Venator (Silver) Date 2012-01-11 18:57
The term is "bankroll".

Ah yes, you are right. Too much daydreaming about my payroll, I guess!

Maybe Dagh meant "easy".

Would be perfect for me :wink:
Parent - - By Hans van der Zijden (*) Date 2012-01-11 17:46
Hi Vas,

Don't worry about me not being able to pay my rent. You are not going to play me poor. I am too good for that. Besides I would stop playing well before the bottom of my bankroll would be in sight. Here you have a schedule for tournaments in the Amsterdam Casino:

WEEKLY POKER TOURNAMENTS

There is a maximum of 50 participants for all tournaments.
Monday         € 110 NLH FREEZE-OUT
                        startstack 2.500
                        levels 20 minutes
                        draw 20.00 h - start 20.15 h
Tuesday        € 110 NLH UNL REBUYS € 50 + ADD-ON € 50
                        startstack 1.000 - rebuy 1.000 - add-on 1.500
                        levels 20 minutes - rebuy period 3 levels
                        draw 20.00 h - start 20.15 h
Wednesday € 165 NLH 1 REBUY € 150
                        startstack 2.500 - rebuy 2.500
                        levels 20 minutes - rebuy period 3 levels
                        draw 20.00 h - start 20.15 h
Thursday      € 60 PLO UNL REBUYS € 50 + ADD-ON € 50
                        startstack 1.000 - rebuy 1.000 - add-on 1.500
                        levels 20 minutes - rebuy period 3 levels
                        draw 20.00 h - start 20.15 h

On sundays they have different tournaments. Every day they also have cashgames naturally. But the rake is rather high.

Usually somewhere in november they have big tournament with a few hundred players and buy-ins varying from 100 to 5000 euro. A lot of foreign (professional) players enter. You should buy-in several months in advance to be sure of a place, except maybe for the really expensive ones.

You are more than welcome. Not sure if I will have a room available for you, but there are lots of hotels also.

Best regards, Hans.
Parent - - By Vasik Rajlich (Silver) Date 2012-01-11 17:58
Hi Hans,

thanks for the info! What is the blind structure? Is it 25/50 for the first level? How many players are playing in these tournaments?

For what it's worth, in East Europe Prague is really great, Budapest is decent (with a few law enforcement problems recently :smile:), and Poland is kind of a desert.

I'm not a big fan of cash games in casinos.

Best regards,
Vas
Parent - - By Hans van der Zijden (*) Date 2012-01-11 18:04
I will have to look that up. I haven't been in the casino for 2 years now. I only play online. If I find something I will let you know.
Parent - By Vasik Rajlich (Silver) Date 2012-01-11 18:09
Ok. It's probably 25/50. In which case the Wednesday tournaments sound pretty good (and the others are very fast).

Vas
Parent - By OleM (**) Date 2012-01-17 21:45
This question has a flaw, if it presumes no knowledge, then how would he know what is an engines strong point?
Parent - - By Banned for Life (Gold) Date 2012-01-09 05:08
4.1. PST. To pre-empt Riis (as he diverges rapidly), the evidence here is clear. As
I stated above, the Rybka 1.0 Beta PST can be generated from the Fruit 2.1 code
by changing 17-18 numbers (tuning), and 4-6 lines of code (amongst about 50).


This is a rather devious attempt to claim that that Rybka's use of the same polynomial generator functions as fruit implies that it also uses the same generation code. there is no evidence at all, none whatsoever, that this actually occurred. Please feel free to provide any evidence to the contrary. Note that this baseless assertion is referred to as the strongest evidence that Rybka contains copied Fruit's code.

Rybka does reuse the Fruit PST generation polynomials, albeit with different coefficients. This has never been denied, and no protection is ever afforded to mathematical functions in other endeavors. Please note that the tables themselves, after being normalized to centipawns in both cases, are not particularly similar.

I challenged anyone to find any other engine (not Fruit-related) whose PST could be derived from such
a small set of changes from the Fruit 2.1 code. There has been no answer.


This is a trivial challenge. The Crafty bishop PST can be derived exactly with no changes to the Fruit 2.1 bishop PST generation code, by increasing each of the polynomial coefficients by 2. The laughable assertion has been made that this is a coincidence...

5- Do you know of any other engine (not Fruit-related) whose PST's can be derived by a small set of changes form Fruit code?

Question 5 is certainly the wrong question. It is designed to focus on the fact that the Rybka uses a similar polynomial with different coefficients to generate its PSTs, even though these same generator polynomials can generate a very large range of PSTs and the Fruit and Rybka tables themselves are significantly different. Is anyone really claiming that there is something wrong with reusing the same third, fourth, or fifth order polynomials to generate PST values? Suppose it could be shown that any PST could be closely modeled with the same eighth order polynomial generating function. Would this disqualify all programs? One should also note that PSTs represent chess knowledge and should be expected to become more similar over time, as they are subject to additional tuning.

On the other hand, if one does want to measure similarity in PSTs, a much better test would measure the correlation between the corresponding tables in different engines. If this test were conducted, it would be noted that Rybka's PSTs are NOT more similar to Fruits than those from other engines. Correlation is an excellent measure of this similarity in that it is scale invariant, so it would not be influenced by whether an engine provided generally low or high evaluations, or even by whether the PSTs were given in centipawns or with values 32 times greater (as in Rybka).

Just for grins, next week I'll put together a little MATLAB routine for performing a calculation of distance between PSTs, and publish it so that anyone who is interested in how similar two PSTs really are can come up with a representative number.
Parent - - By Loboestepario (****) Date 2012-01-09 06:54
I appreciate your comments and I'm sure that carrying a conversation with you would make me learn a great deal but this is about Vas and rybka, and not you or me.
If Vas continues to avoid answering questions any attempt of yours to make him appear more credible simply falls short.
Parent - - By Banned for Life (Gold) Date 2012-01-09 07:15
There is a broader question. Vas has repeatedly stated that his code is original. His detractors have come back saying, why are you using the same formulas for PST generation? It is a valid question as to whether this question is relevant or not. It is even more important to realize that this question may be defined by the ICGA as an originality issue (they can define originality any way they want), but it is not indicative of copied code. Claiming copied code based on reuse of formulas isn't valid here, or anywhere else.
Parent - - By Vasik Rajlich (Silver) Date 2012-01-09 07:59
Yes, I agree. I don't know exactly how similar my PST generation formulas are to Fruit but in general I would have nothing against having them be very similar.

There isn't much Elo at stake in PSTs, but for other aspects of the program there can be significant Elo at stake and my answer is the same.

Loboestepario: If I missed something, please bump or repost. This thread is kind of messy (but I also think the mods shouldn't be too aggressive in cleaning it up).

Vas
Parent - - By Loboestepario (****) Date 2012-01-11 04:12

> Loboestepario: If I missed something, please bump or repost. This thread is kind of messy (but I also think the mods shouldn't be too aggressive in cleaning it up).


Thanks for your time, I don't have any more questions. Honestly, I am not closer to understanding this issue after reading your answers multiple times but I'm sure that's my problem.
Parent - By Vasik Rajlich (Silver) Date 2012-01-11 07:25
No problem. :smile:

Vas
Parent - By David Evans (****) Date 2012-01-08 18:15
The technical questions are at the crux of the matter Vas i dont care for the rest of the nonsense.

Can you answer the technical questions and as they say the rest is just hot air..

Gl
Parent - - By Hannibal (**) Date 2012-01-09 18:33 Edited 2012-01-09 18:41
1) By what standard did you judge Ippolit and Strelka to be clones (or illegal) of your own engines?

- Of Ippolit: "These are all just decompiled Rybka 3 clones. It's pretty obvious from the Ippolit sources, any programmer will tell you the same thing."
- Of Strelka: "These changes are extensive and no doubt lead to differences in playing style and perhaps a useful engine for users to have, but they do not change the illegality of the code base."

2) How does that standard differ from the standard that was used by the ICGA in defining originality?
Parent - - By Vasik Rajlich (Silver) Date 2012-01-09 21:26
Ok, this could be a long discussion. :smile: For me the term "clone" refers to a program whose source code isn't original.

Vas
Parent - - By sockmonkey (***) Date 2012-01-09 21:48
And, just to be sure that we understand one another, a "program whose source code is original" is a program that someone typed him/herself, whether or not he/she was using someone else's code as the basis for that typing. Or how would you define it?

Thanks for the clarification, it's helpful.

Best, Jeremy
Parent - - By Vasik Rajlich (Silver) Date 2012-01-09 22:01
Well, that's not quite right. I am not sure how exactly I would formulate that.

Vas
Parent - - By sockmonkey (***) Date 2012-01-09 22:44
Sorry to push, but I think it's important to understand how you view this, not only for helping to clarify the current ICGA situation, but also to explain your attitude toward programs that you've called "clones" in the past. That is, you can't determine that something is "unoriginal" without having a clear idea of what "original" is, can you?

Let's take an example. Ippolit may have been the result of machine-assisted, human-guided disassembly of Rybka 3 (let's say it was, just for the purposes of argument -- I don't know that to be the case). The definition that you've offered so far would typically mean that Robbolito (a completely re-typed Ippolit, if you will, in which a human went through the program module for module and cleaned up/standardized/"idiomaticized" the original machine-assisted code), is an original program. Would you agree on a theoretical level? Understanding that we're speaking of a hypothetical Ippolit/Robbolito here, which conform to the descriptions I've offered.

Thank you, Jeremy
Parent - - By Vasik Rajlich (Silver) Date 2012-01-09 23:26
I am not sure if Robbolito is an entirely re-typed Ippolit or just a massively cleaned-up one. If it's entirely re-typed, then I think this is a kind of edge case. You could go either way on that. AFAIK the changes going from Ippolit to Robbolito were pretty big.

Anyway, this is really just a matter of terminology.

Vas
Parent - - By sockmonkey (***) Date 2012-01-09 23:48
I tend to think that it's a much more important distinction than a simple matter of terminology, particularly given that you have been accused of plagiarism by the ICGA, and have yourself accused others (the authors of Strelka, Ippolit, presumably Robbolito and the gang as well) of same. I would certainly say that my definition of "original", which I consider fairly commonsensical, is significantly stricter than yours.

But I would further note that, taking the evidence presented by the ICGA at face value, Rybka 1 is quite likely, at least in parts, and using your own definition, "an edge case". I'm quite sure that you disagree, but I don't honestly see how you can, on the one hand, make accusations against software that used Rybka as a (too-)strong model, while on the other hand disputing the ICGA's accusations that you used Fruit as a (too-)strong model. I wonder if you could comment on that.

Thanks, Jeremy
Parent - - By Vasik Rajlich (Silver) Date 2012-01-10 08:20
My comments about Rybka are simply correct. There is no edge area there. I pick terms which describe the situation properly and accurately, and I avoid terms which don't.

My comments about other engines involve some speculation. I think on the balance that these are not edge cases either, but they might be.

It makes some sense for organizers to try to go through all of the scenarios and define what exactly is ok and what is not ok. I am not in a position of authority, so I am quite happy to just say that something is for me a borderline case and leave it at that.

Vas
Parent - - By sockmonkey (***) Date 2012-01-10 10:23
I'll close by observing that this ultimately comes down to "Rybka is original because I say so, despite many indications to the contrary". I am personally not overwhelmed by that response, but maybe it's enough for others.

Thanks (very much) for your comments and time.

Jeremy
Parent - - By Ray (****) Date 2012-01-10 10:30

> I'll close by observing that this ultimately comes down to "Rybka is original because I say so, despite many indications to the contrary".


Dr Soren Riis also says that Vas is innocent of the charges against him.
Parent - - By sockmonkey (***) Date 2012-01-10 10:45
Ray, I'm actually talking with Vas (well, actually, I'm done now). If I needed commentary from the peanut gallery, I'd solicit it. Thank you for your understanding.
Parent - - By Rebel (****) Date 2012-01-10 13:16

> If I needed commentary from the peanut gallery


I once respected you, glad I left your forum.
Parent - By Ugh (*****) Date 2012-01-10 13:20
+1
Parent - By Venator (Silver) Date 2012-01-10 16:10
+1

Glad I never attended to his forum.
Parent - - By RFK (Gold) Date 2012-01-10 13:25
Why do you slip so easily into unwarranted ridicule -who is responsible for having taught you this behavior ?:sad:
Parent - - By sockmonkey (***) Date 2012-01-10 13:35
Pots... kettles...
Parent - By RFK (Gold) Date 2012-01-10 13:37
And here comes the knee jerk reaction! :razz:
Parent - By Vasik Rajlich (Silver) Date 2012-01-10 11:39
Sure thing, thanks for the questions. :smile:

Vas
Parent - By Rebel (****) Date 2012-01-09 22:04
Here is the deal, provided a programmer already has an engine and he wants to copy code from another engine:

Copying code in most cases makes no sense because the variable names, data structure bite each other, so he will have to do a conversion to his own variable names, data structures.

Is that own written code ?

Data is another matter of course.
Parent - - By tomgdrums (****) Date 2012-01-09 21:52
Hi Vas,

So this is all rather sticky and I am not even sure I have a question.  I would say your silence throughout all of this made this customer sad.  (sadder than watching Andy Reid coach!)  And a tad bit betrayed.

It is amazing how polarized the whole thing has become.  Worse than congress!

Either way good luck with your further endeavors.  I would say that silence is not always the best policy.
Parent - - By Vasik Rajlich (Silver) Date 2012-01-09 22:03
Well, maybe you're right. I don't know. I did give an interview to Nelson which you can find on Youtube. :smile:

Vas
Up Topic Rybka Support & Discussion / Rybka Discussion / A Big Thanks and a Small Update
1 2 3 4 5 620 21 Previous Next  

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill