Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / Loop as a Fruit clone
- - By Rebel (****) Date 2013-09-20 06:03
Loop 2007

:wink:
Parent - - By bob (Gold) Date 2013-09-20 14:03
I do not find any source code for loop anywhere.  Any links?
Parent - - By michiguel (***) Date 2013-09-20 20:58

> I do not find any source code for loop anywhere.  Any links?


You can ask the ICGA secretariat, which is investigating/leading the case against it.

Miguel
Parent - - By bob (Gold) Date 2013-09-20 21:26
You know we don't have source?  Can't even locate the author.  Our efforts were done just as they were done for Rybka, disassembly.

Not what I need HERE...
Parent - - By Rebel (****) Date 2013-09-21 00:15
Ask Fabien, it's nearly identical.
Parent - - By bob (Gold) Date 2013-09-21 16:28
How do I get loop source from Fabien?  Can't be "identical" if it uses bit boards.  Has to be modified.

as I said...
Parent - - By Rebel (****) Date 2013-09-21 16:54
How do I get loop source from Fabien?

Because it's Fruit 2.2.1 (closed source)


Can't be "identical" if it uses bit boards.  Has to be modified.

It has some poor bit-board constructions.

Look, I can't give you the hex-rays source code, you will need to contact Fabien.
Parent - By bob (Gold) Date 2013-09-21 16:57
I am not asking for it.  Never did.  However, if you are saying fruit 2.1 is a bitboard program, I'll email him today.  Since it was not open source, I've never paid any attention to any release notes or anything else since anything I use for testing has to compile and run under linux.
Parent - - By Rebel (****) Date 2013-09-21 16:59
Is not going to work Bob, Reul (if you do nothing) gets a free card out of jail.

I recently spoke with Jaap v/d Herik. He told me 2 things:

1. There are not enough programmers to form a Panel to investigate Loop;

2. He met Watkins in Japan during the last WCCC and Watkins told him the evidence against Loop is not so clear.

Now one of the gents is lying, because we know Mark's opinion about Loop.

Postby BB+ » Mon Apr 18, 2011 7:15 am
I made a cursory glance at Loop 2007 (the 64-bit version). It uses exactly the same PST as Fruit 2.1. The pawn eval uses the same scores. It shifts the mobility count as with Fruit (by 4,6,7,13) before multiplying, and then the arithmetic is the same (4 for N, 5 for B). I'm not that interested in going further.

http://www.open-chess.org/viewtopic.php?p=12152#p12152
Parent - - By bob (Gold) Date 2013-09-21 17:08
What is "not going to work"???

I've sent an email to Fabien.  10 minutes ago.  Will wait to see what he says and if he is willing to send me the bitboard fruit source code.  I'll post here to let you know if he sends it (and what I find) or if he declines...

I would not be surprised to see the loop case end up in a "summary judgement" since the author is completely unwilling to cooperate (sound familiar)?
Parent - - By Rebel (****) Date 2013-09-21 17:38
What is "not going to work"???

My comment came with an "if"


I've sent an email to Fabien.  10 minutes ago.

Good.

Be prepared for disappointment.

It's a can of worms.

Key-word --> money.
Parent - - By bob (Gold) Date 2013-09-21 17:58 Edited 2013-09-21 18:05
Here's the quote I was asking about.

Is not going to work Bob, Reul (if you do nothing) gets a free card out of jail.

I didn't understand "what" is not going to work.

As far as "disappointment" goes.  "Nothing ventured, nothing gained."  I asked.  If he says no, he says no.  I'm not sure about the money comment.  2.1 is way old and way weaker than current engines.  2.2 is stronger than 2.1, but I don't think it gets in the same discussion with today's programs.  More if / when I hear anything.

Just got a reply from Fabien.  "No fruit used bit boards, maybe you are thinking 2.3+?"...  He said Ryan's version might have, but he doesn't have that source.  I'll try to contact Ryan next...
Parent - - By Rebel (****) Date 2013-09-21 18:28
It's over 2 year ago that Fabien filed a complaint against Loop, that's what I meant.

Again, you need Fruit 2.2.1, it's topping Adam's clone list BTW.
Parent - - By bob (Gold) Date 2013-09-21 18:31
as a clone of what?  2.1?

I don't see why I need 2.2.1...  not bitboard according to fabien.  My interest is simply in seeing ANOTHER fruit conversion to bit boards...
Parent - - By Adam Hair (**) Date 2013-09-21 19:16
I don't think that you need the source to 2.2.1. However, Loop's evaluation is a clone of Fruit 2.2.1. According to move selections and evaluation scores, Loop (starting with Loop 10.32f) more closely resembles Fruit 2.2.1 than Fruit 2.1 does.
Parent - - By bob (Gold) Date 2013-09-21 19:52
How close are 2.1 and 2.2?  Fabien referred me to Ryan, who did the 2.3 and beyond versions that were bitboard based.  I've emailed Ryan already to see if he is willing to send me the entire source, or at least evaluate.cpp...
Parent - By Adam Hair (**) Date 2013-09-22 02:57
Here is some data:

From the similarity study I conducted and Ed is hosting, Loop 2007 (which is Loop 13.6) and Fruit 2.2.1 have the same move selection 84.73% of the time. That is by far more than any other pair of engines and is over 13 standard deviations from the mean in my study. For Loop and Fruit 2.1, it is 70.81% (about 9 sd from the mean).

When searching a low fixed depth, the percentage rises. At depth 6, the percentage of matched moves for Loop/Fruit 2.2.1 is 89.83% (71.23% for Loop/Fruit 2.1 and 72.26% for Fruit 2.2.1/Fruit 2.1). At depth 2, it rises to 95.97%. Almost all engine pairs I have tested fall well short of these numbers for fixed depth.

Finally, here are some plots comparing the depth 6 scores of the engines for each position in the similarity tool:







Looking at the plots, it can be seen that the magnitude of Loop's scores is higher than Fruit's. But, it can also be seen that the scores in the Loop/Fruit 2.2.1 plot fall more precisely on a straight line than Loop/Fruit 2.1 or even Fruit 2.2.1/Fruit 2.1.

In comparison, here is Shredder 11 and Fruit 2.2.1:



This is much more like the typical plot for unrelated engines. Overall, Loop (starting with at least Loop 10.32f) behaves like it is a minor variation of Fruit 2.2.1.
Parent - - By Rebel (****) Date 2013-09-21 19:44
Loop 2007 is Fruit 2.2.1 + a few bit-board changes.

Loop is vice-world champion 2007 Amsterdam.

Make sure the case isn't wiped under the carpet, as it has every sign of a cover up.
Parent - - By bob (Gold) Date 2013-09-21 19:51
Nothing will be covered up, I can guarantee you that...
Parent - By Rebel (****) Date 2013-09-21 20:08
Nothing will be covered up, I can guarantee you that...

I like to hear you say that Bob.

But have some sort of reality check first.

http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?pid=485282#pid485282
Parent - - By Adam Hair (**) Date 2013-09-21 19:05

> I recently spoke with Jaap v/d Herik. He told me 2 things:
>
> 1. There are not enough programmers to form a Panel to investigate Loop;
>
> 2. He met Watkins in Japan during the last WCCC and Watkins told him the evidence against Loop is not so clear.


The first statement has some validity. The investigation has been dormant for almost a year in part because a lack of interest in continuing the investigation. But I do not believe that MW told Jaap that the evidence against Loop was unclear. The evidence presented so far has been quite clear. Either there has been a misunderstanding or Jaap has not spoken the truth.
Parent - - By bob (Gold) Date 2013-09-21 19:55
The delay has primarily been the result of zero communication from the author.  This won't be allowed to go on much longer.  It was a criticism of the last investigation, and we decided to make every effort to contact the author before proceeding...

Everybody hates this RE-approach to these investigations, if they can be avoided.  If we had rybka's source there would have been no debate at all.
Parent - - By Ugh (*****) Date 2013-09-23 12:16
The delay has primarily been the result of zero communication from the author.  This won't be allowed to go on much longer.  It was a criticism of the last investigation, and we decided to make every effort to contact the author before proceeding...

What exactly are these "every efforts"? I guess more than "send an email to a defunct email address" and "ask a friend" as detailed below about eighteen months ago? A technically qualified young man in his twenties could be anywhere, USA, China. Has an international enquiry agent been used?

I ask because I don't want to imagine it might suit to not find him, delay a long time and kick the case into the long grass. Two years or so does seem rather a long delay so far, and now, according to Ed's communication the enquiry has been more or less parked. This appears at variance with case Rybka where Vas asked for time to prepare and was denied.

Are you able to post the document that tasks you to make this particular enquiry and sets out what you are supposed to do and report about?

Allegations against two more Chess Engines – The LOOP Program
Posted on May 22, 2012 by Joke Hellemons
By David Levy

The International Computer Games Association (ICGA) has received a formal complaint from Fabien Letouzey, the programmer of FRUIT, regarding the program LOOP. In addition the ICGA has received sufficient prima facie evidence to warrant an investigation. The allegation is that LOOP, which has participated in the World Computer Chess Championship, uses code derived from FRUIT. FRUIT placed second at the 2005 World Computer Chess Championship in Reykjavik.

LOOP was developed by Dr Fritz Reul. The ICGA has attempted to contact Dr Reul in order to give him the opportunity of commenting on the allegation prior to the ICGA conducting a technical investigation into LOOP, but it appears to us that Dr Reul does not wish to be found. The email address that had previously been used by Dr Reul is no longer active, so on behalf of the ICGA I have sent him two emails via a friend of his who indicated that he would pass email on to Dr Reul, who did not respond to either of my emails.

The ICGA is now commencing an “in absentia” investigation into LOOP in order to examine the allegation about the use of code derived from FRUIT. If Dr Reul wishes to defend himself against these allegations the ICGA would be pleased to hear from him. He will then be invited to comment on any evidence that the ICGA might consider during the investigation process.

Allegations have also been made concerning the program THINKER, which will also be the subject of an ICGA investigation.
Parent - By bob (Gold) Date 2013-09-23 13:19
"every effort" has included emails, direct contact with known associates, etc.  Harvey has already mentioned that it is time to resolve the issue, since additional time doesn't seem to be worthwhile regarding direct contact.

It is NOT going to just lie dormant, as it has been public information for quite a while.  One criticism of the last investigation was that it was too rushed.  We've tried to avoid that the second time around.
Parent - - By Rebel (****) Date 2013-09-21 20:05
The first statement has some validity. The investigation has been dormant for almost a year in part because a lack of interest in continuing the investigation. But I do not believe that MW told Jaap that the evidence against Loop was unclear. The evidence presented so far has been quite clear. Either there has been a misunderstanding or Jaap has not spoken the truth.

Here are Jaap's words, decide for yourself.

dear ed, i can imagine that it bothers you. I may admit that it bothers me too. you know that i eliminated the program List from the tournament in graz. thereafter we ( ICGA board) had a long discussion and many things happened. The result was that reul was given the opportunity to return. then even more things happened, he came to me to ask me for the possibility of being his supervisor which i did. I spoke in Yokohama with mark watkins and gave him the opportunity to tell his story to the assembled chess porgrammers. Mark did so and informed us on the state of affairs. from his point of view it was not as clear as you stated but there were pointers. His material was not cross checked since the panel members, competent to perform this task had lost interest in doing so. therfore the progress of the investigation was calmed down to almost being stopped.

.....

therefore, i admit i was interested in true facts and i spoke to mark watkins. amir ban went into discussion with mark, and a few other people from the audience too. the oral formulation of a kind of conclusion was that there is not a convincing proof, moreover there was a discussion to what extent ....  . in brief, it was not clear whether the monks were equal, they might be similar, but the weighing of the proportional part was too difficult for the "audience judges".

as seen from your side, i may understand that our perspectives on these clone topics are still different and that my arguments did not make the gap in our minds on this topic smaller. so be it. however, be convinced that my sympathy for your opinion is there but agreement is too great a request which i cannot fulfil

i wish you all the best and look forward to seeing you in future events and be convinced that i certainly will start a conversation with you

best regards,
    jaap
Parent - - By bob (Gold) Date 2013-09-21 21:43
Doesn't bother me at all.  Jaap can't "close the investigation".  I suppose he might try to convince David of something different should the panel decide it was a copy...
Parent - - By Rebel (****) Date 2013-09-21 22:03
Here is the bet, it's above your power.
Parent - By bob (Gold) Date 2013-09-22 17:19
Strange things do happen.  But I suspect this one won't be covered up because too many know about it.

We're still working on it.  Might be news before too much longer, hopefully...
Parent - By user923005 (****) Date 2013-09-20 22:08
The loop source code was posted on a github site some time ago.
However, that version was array based, not bitboard based.
He seems to have removed that site.
Up Topic Rybka Support & Discussion / Rybka Discussion / Loop as a Fruit clone

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill