Critter 1.2 update + new homepage

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

BubbaTough
Posts: 1154
Joined: Fri Jun 23, 2006 5:18 am

Re: Critter 1.2 update + new homepage

Post by BubbaTough »

Don wrote:
One thing that is perhaps not given enough credit is the huge importance of massive automated testing - in some sense I think this has been the major breakthrough of the past few years. It has not really been possible to do this well within a reasonable budget until multi-core chips have become ubiquitous.
I strongly agree with this. I think most authors generate ideas faster than they can test them. Faster and more plentiful cores mean more ideas can be tested, and the bar lowered for how good an idea has to be before it can be confirmed or denied how good it is. There is no doubt having a pool of vetted ideas to try in open source is very nice, but the ability to test these + variations of them + ideas they inspire + ones own ideas + retune the system after change....well, that is the real game changer in my opinion.

-Sam
User avatar
Don
Posts: 5106
Joined: Tue Apr 29, 2008 4:27 pm

Re: Critter 1.2 update + new homepage

Post by Don »

kranium wrote:
mcostalba wrote:
zamar wrote:Congrats for the new release Richard!

Now there are 4 engines in IPON list with almost equal strength:

Rybka 4 2954
Komodo 2952
Critter 2948
Stockfish 2942

Only Houdini has a clear lead.

Cheers,
Joona
I fully second Joona, it is impressive to see 3 engines on par with Rybka 4 when until a couple of years ago the previous version Rybka 3 was practically unchallenged at the top.

Publishing of Ippo sources for sure has greatly changed the landscape in the last two years and I like to think that also SF has contributed a little in helping the fast raise of today's very strong engines.

Congratulations Richard !
yes, the raping and plundering of the misbegotten Ippolit source code is nearly complete now.
most top engines have benefited sufficiently, albeit with blood on their hands.
How is there blood on your hands for using good ideas?



in addition, these same top program authors and the establishment 'good old boys' have managed to discredit it enough to keep it off most major rating lists.
yes, job well done...
pats on the back, a big thanks to StockFish!?, and congrats all around!
User avatar
fern
Posts: 8755
Joined: Sun Feb 26, 2006 4:07 pm

Re: Critter 1.2 update + new homepage

Post by fern »

Perhaps so, but you are a genius as a programmer.
Fern
kranium
Posts: 2129
Joined: Thu May 29, 2008 10:43 am

Re: Critter 1.2 update + new homepage

Post by kranium »

Don wrote:
kranium wrote:
mcostalba wrote:
zamar wrote:Congrats for the new release Richard!

Now there are 4 engines in IPON list with almost equal strength:

Rybka 4 2954
Komodo 2952
Critter 2948
Stockfish 2942

Only Houdini has a clear lead.

Cheers,
Joona
I fully second Joona, it is impressive to see 3 engines on par with Rybka 4 when until a couple of years ago the previous version Rybka 3 was practically unchallenged at the top.

Publishing of Ippo sources for sure has greatly changed the landscape in the last two years and I like to think that also SF has contributed a little in helping the fast raise of today's very strong engines.

Congratulations Richard !
yes, the raping and plundering of the misbegotten Ippolit source code is nearly complete now.
most top engines have benefited sufficiently, albeit with blood on their hands.
How is there blood on your hands for using good ideas?



in addition, these same top program authors and the establishment 'good old boys' have managed to discredit it enough to keep it off most major rating lists.
yes, job well done...
pats on the back, a big thanks to StockFish!?, and congrats all around!
i seem to remember you (and Larry) being one of the loudest voices calling for it's complete banning...
discrediting it as wrong, unethical, and illegitimate at every opportunity...
all the while studying it's code for your own benefit...?

doesn't that seem a bit hypocritical?

if it's illegitimate, then all engines using even 'ideas' from it are likewise stained.

well you've gotten your way...it's been banned from most major lists, cannot enter tournaments, etc.
and Komodo, Stockfish, etc. have had benefited enormously from it's publication...
shoot the messenger! (while raping and plundering it for anything beneficial?)

congrats! pats on the back all around for a job well done.
now if you can get it wiped off the planet completely you'll really be sitting pretty!
User avatar
Don
Posts: 5106
Joined: Tue Apr 29, 2008 4:27 pm

Re: Critter 1.2 update + new homepage

Post by Don »

tomgdrums wrote:
kranium wrote:
mcostalba wrote:
zamar wrote:Congrats for the new release Richard!

Now there are 4 engines in IPON list with almost equal strength:

Rybka 4 2954
Komodo 2952
Critter 2948
Stockfish 2942

Only Houdini has a clear lead.

Cheers,
Joona
I fully second Joona, it is impressive to see 3 engines on par with Rybka 4 when until a couple of years ago the previous version Rybka 3 was practically unchallenged at the top.

Publishing of Ippo sources for sure has greatly changed the landscape in the last two years and I like to think that also SF has contributed a little in helping the fast raise of today's very strong engines.

Congratulations Richard !
yes, the raping and plundering of Ippolit source code is nearly complete now.
most top engines have benefited sufficiently.
in addition, top program authors have managed to discredit it enough to keep it off most major rating lists.
yes, job well done...
pats on the back, a big thanks to StockFish!?, and congrats all around!
I must say I agree with Norman on this!

How is it that the author of Komodo can openly admit to looking at using ideas of Ippolits in his engine and yet the ippolits are not allowed on rating lists??
It's stupid not to use good ideas in a chess program, regardless of where they came from. I'm not sure ANYONE here has ever said anything to the contrary - but certainly I have not.

I am no fan of cloning or derivatives. I have yet to use Houdini or Fire etc. etc. (I used Robbolito in Scid for android and then deleted because I felt bad)

But again Norman is correct about the rating list hypocrisy!! If Houdini and Komodo can be be on the lists then so can Fire and Robbolito!
There are 2 issues. The first I won't argue about, but it's WHETHER these programs are just versions of other programs. You can refer to other threads and have several days of enjoyable reading if you want to get yourself all stirred up again.

However the second and primary issue is something that you have to agree with if you have a scientific mind at all. Most of the people here who don't understand it are probably non-technical users who love computer chess but do not understand statistics or just are not thinking about things from the point of view of fairness to the authors providing these great programs. Here is how it works:

If you have a tournament with say 9 programs competing that are all the same strength (this point doesn't matter but makes it a little easier to understand) then each program has about 1/9 of winning the tournament. Let's say that my program Komodo is one of the players in this tournament. Let's say the tournament director says, "we need 10 programs so nobody has to have a bye." So I tell the TD I have an experimental version of Komodo that can play in the 10th spot. It's basically Komodo with a few minor changes and plays about the same strength as any of the other programs.

So now there are 2 Komodo's and 8 other programs. The odds that ONE version of Komodo will win the tournament is now about twice that of any other program. Now if the other 8 authors decide to walk out of the tournament would you blame them?

This is why tournaments that are well organized do not let program authors enter multiple versions in their tournaments, even if there is a fee and they are willing to pay it for each program.

Let's say that Houdart decides to release 20 different versions of Houdini, and he makes up 20 fake names for the programs and 20 fake author names and presents them anonymously to the world. Let's also say that he uses this technique to get others to test his changes for him, so really each one is different version of Houdini, but it is in fact Houdini.

Now you look on the rating lists to select a program to either purchase (or get free) for your gitfed nephew who wants to become a Grandmaster. Number 21 on the list is a program called Stockfish 2.2 and you think, why should I care about some crappy program that is number 21 when I have this huge selection of great programs that are top 10?

There are many who argue, who cares? Why not just put every program out there and stop obsessing over this? My answer is that if you really believe this makes any sense, then why can't I release many versions of Komodo and I will give them different names if people are stupid enough to think that makes them different and unique? Just to keep them happy I can make them somewhat different and make some a few ELO weaker than others. That's the situation we have with ippo, robbo, firebird, and any number of very close derivatives.

There is also the expression, "don't bite the hand the feeds you" and this applies. There are a lot of original programs our there and most of them are labors of love and they are given away for free despite the fact that they involved huge amounts of effort on the part of the authors. Computer chess is a like a drug for many of us and the program authors are you "suppliers." Some of the "suppliers" are very competitive and want to have a top 10 program so the gross unfairness of some top program essentially getting to claim several spots on the top 10 list, or playing in a tournament where a very good but not best program has no chance whatsoever because someone else gets to have several entries (like 3 or 4 ippo clones) in the tournament is not only grossly unfair, but has a negative impact on computer chess in general as it diminishes a lot of very fine programs that actually are the real McCoys.
User avatar
Don
Posts: 5106
Joined: Tue Apr 29, 2008 4:27 pm

Re: Critter 1.2 update + new homepage

Post by Don »

kranium wrote:
Don wrote:
kranium wrote:
mcostalba wrote:
zamar wrote:Congrats for the new release Richard!

Now there are 4 engines in IPON list with almost equal strength:

Rybka 4 2954
Komodo 2952
Critter 2948
Stockfish 2942

Only Houdini has a clear lead.

Cheers,
Joona
I fully second Joona, it is impressive to see 3 engines on par with Rybka 4 when until a couple of years ago the previous version Rybka 3 was practically unchallenged at the top.

Publishing of Ippo sources for sure has greatly changed the landscape in the last two years and I like to think that also SF has contributed a little in helping the fast raise of today's very strong engines.

Congratulations Richard !
yes, the raping and plundering of the misbegotten Ippolit source code is nearly complete now.
most top engines have benefited sufficiently, albeit with blood on their hands.
How is there blood on your hands for using good ideas?



in addition, these same top program authors and the establishment 'good old boys' have managed to discredit it enough to keep it off most major rating lists.
yes, job well done...
pats on the back, a big thanks to StockFish!?, and congrats all around!
i seem to remember you (and Larry) being one of the loudest voices calling for it's complete banning...
discrediting it as wrong, unethical, and illegitimate at every opportunity...
all the while studying it's code for your own benefit...?

doesn't that seem a bit hypocritical?
No. I do not allow myself to be a hostage. If I rejected every idea that I did not think up myself wouldn't that be pretty ridiculous?

In fact you are being ridiculous. Not only do I advocate using any good idea in a chess program, I also have no problem with actively seeking them out. I would even reverse engineer a program if I enjoyed that type of thing (I don't) to find good ideas.

I'm not going to discuss this with you if you are not even interesting in trying to understand my point of view and after everything I've written you still completely misunderstand my feelings on this.

A few minutes ago I posted on this, trying to illustrate what the problem is any why the majority of program authors abhor what is happening.

if it's illegitimate, then all engines using even 'ideas' from it are likewise stained.

well you've gotten your way...it's been banned from most major lists, cannot enter tournaments, etc.
and Komodo, Stockfish, etc. have had benefited enormously from it's publication...
shoot the messenger! (while raping and plundering it for anything beneficial?)
You are being really silly here. Read the post I recently wrote on why several versions of the same program should not be competing in the same forum.

congrats! pats on the back all around for a job well done.
now if you can get it wiped off the planet completely you'll really be sitting pretty!
tomgdrums
Posts: 736
Joined: Wed Dec 02, 2009 9:48 am

Re: Critter 1.2 update + new homepage

Post by tomgdrums »

Don wrote:
tomgdrums wrote:
kranium wrote:
mcostalba wrote:
zamar wrote:Congrats for the new release Richard!

Now there are 4 engines in IPON list with almost equal strength:

Rybka 4 2954
Komodo 2952
Critter 2948
Stockfish 2942

Only Houdini has a clear lead.

Cheers,
Joona
I fully second Joona, it is impressive to see 3 engines on par with Rybka 4 when until a couple of years ago the previous version Rybka 3 was practically unchallenged at the top.

Publishing of Ippo sources for sure has greatly changed the landscape in the last two years and I like to think that also SF has contributed a little in helping the fast raise of today's very strong engines.

Congratulations Richard !
yes, the raping and plundering of Ippolit source code is nearly complete now.
most top engines have benefited sufficiently.
in addition, top program authors have managed to discredit it enough to keep it off most major rating lists.
yes, job well done...
pats on the back, a big thanks to StockFish!?, and congrats all around!
I must say I agree with Norman on this!

How is it that the author of Komodo can openly admit to looking at using ideas of Ippolits in his engine and yet the ippolits are not allowed on rating lists??
It's stupid not to use good ideas in a chess program, regardless of where they came from. I'm not sure ANYONE here has ever said anything to the contrary - but certainly I have not.

I am no fan of cloning or derivatives. I have yet to use Houdini or Fire etc. etc. (I used Robbolito in Scid for android and then deleted because I felt bad)

But again Norman is correct about the rating list hypocrisy!! If Houdini and Komodo can be be on the lists then so can Fire and Robbolito!
There are 2 issues. The first I won't argue about, but it's WHETHER these programs are just versions of other programs. You can refer to other threads and have several days of enjoyable reading if you want to get yourself all stirred up again.

However the second and primary issue is something that you have to agree with if you have a scientific mind at all. Most of the people here who don't understand it are probably non-technical users who love computer chess but do not understand statistics or just are not thinking about things from the point of view of fairness to the authors providing these great programs. Here is how it works:

If you have a tournament with say 9 programs competing that are all the same strength (this point doesn't matter but makes it a little easier to understand) then each program has about 1/9 of winning the tournament. Let's say that my program Komodo is one of the players in this tournament. Let's say the tournament director says, "we need 10 programs so nobody has to have a bye." So I tell the TD I have an experimental version of Komodo that can play in the 10th spot. It's basically Komodo with a few minor changes and plays about the same strength as any of the other programs.

So now there are 2 Komodo's and 8 other programs. The odds that ONE version of Komodo will win the tournament is now about twice that of any other program. Now if the other 8 authors decide to walk out of the tournament would you blame them?

This is why tournaments that are well organized do not let program authors enter multiple versions in their tournaments, even if there is a fee and they are willing to pay it for each program.

Let's say that Houdart decides to release 20 different versions of Houdini, and he makes up 20 fake names for the programs and 20 fake author names and presents them anonymously to the world. Let's also say that he uses this technique to get others to test his changes for him, so really each one is different version of Houdini, but it is in fact Houdini.

Now you look on the rating lists to select a program to either purchase (or get free) for your gitfed nephew who wants to become a Grandmaster. Number 21 on the list is a program called Stockfish 2.2 and you think, why should I care about some crappy program that is number 21 when I have this huge selection of great programs that are top 10?

There are many who argue, who cares? Why not just put every program out there and stop obsessing over this? My answer is that if you really believe this makes any sense, then why can't I release many versions of Komodo and I will give them different names if people are stupid enough to think that makes them different and unique? Just to keep them happy I can make them somewhat different and make some a few ELO weaker than others. That's the situation we have with ippo, robbo, firebird, and any number of very close derivatives.

There is also the expression, "don't bite the hand the feeds you" and this applies. There are a lot of original programs our there and most of them are labors of love and they are given away for free despite the fact that they involved huge amounts of effort on the part of the authors. Computer chess is a like a drug for many of us and the program authors are you "suppliers." Some of the "suppliers" are very competitive and want to have a top 10 program so the gross unfairness of some top program essentially getting to claim several spots on the top 10 list, or playing in a tournament where a very good but not best program has no chance whatsoever because someone else gets to have several entries (like 3 or 4 ippo clones) in the tournament is not only grossly unfair, but has a negative impact on computer chess in general as it diminishes a lot of very fine programs that actually are the real McCoys.

Hi Don,

1)

I never said it was stupid to use good ideas. But considering how much you protest against the alleged clones (one of the reasons I really like you by the way!), it does seem rather hypocritical to use ideas from those same engines. Sorry but that IS how it is.

2)

Yes ONLY ONE version of an engine should be enrolled in a tournament. I agree with this 100%. BUT Norman (and now I) are referring to ratings lists.

Check any list and you will Rybka 2, 3 and 4. Houdini 1.3 and 1.5.
Junior 10 and Junior 12. Need I go on?

And if you still want to debate your now flawed "one version" theory then go ahead and pick Robolitto as a Single processor version and then either Fire, Ivanhoe OR Ippolit as the MP version. And that solves the Multiple Version problem for lists (which doesn't really exist anyway).

If Houdini is allowed on the ratings lists AND Komodo with ideas from the very engines you railed against is allowed on the lists, then at least one of "those" engines should be on the ratings lists.

And remember I don't LIKE or USE the clones or derivatives. (partly based on your arguments, I may add).

But I don't like hypocrisy and that is what this smells like to me.
Dann Corbit
Posts: 12777
Joined: Wed Mar 08, 2006 8:57 pm
Location: Redmond, WA USA

Re: Critter 1.2 update + new homepage

Post by Dann Corbit »

Games at 2 minutes plus 1 second Fischer increment.
I7 machine, 8 threads no ponder except Komodo, which was limited to 1 thread.

Code: Select all

  Program                    Elo    +   -   Games   Score   Av.Op.  Draws         
1 Critter_1.2_64bit_SSE4   : 3085   26  26   400    65.2 %   2975   42.0 %        
2 Deep Rybka 4.1 SSE42 x64 : 2976   23  23   576    43.3 %   3023   37.0 %        
3 Komodo-203-64-SSE42-dc   : 2885   45  46   176    37.2 %   2975   25.6 %        
There were actually 2 contests, but I forgot to change the name of the target PGN file.
User avatar
notyetagm
Posts: 253
Joined: Mon Jan 25, 2010 3:11 am

Re: Critter 1.2 update + new homepage

Post by notyetagm »

Dann Corbit wrote:Games at 2 minutes plus 1 second Fischer increment.
I7 machine, 8 threads no ponder except Komodo, which was limited to 1 thread.

Code: Select all

  Program                    Elo    +   -   Games   Score   Av.Op.  Draws         
1 Critter_1.2_64bit_SSE4   : 3085   26  26   400    65.2 %   2975   42.0 %        
2 Deep Rybka 4.1 SSE42 x64 : 2976   23  23   576    43.3 %   3023   37.0 %        
3 Komodo-203-64-SSE42-dc   : 2885   45  46   176    37.2 %   2975   25.6 %        
There were actually 2 contests, but I forgot to change the name of the target PGN file.
Damn, that is some performance by Critter 1.2! Need more games! :-)
Dann Corbit
Posts: 12777
Joined: Wed Mar 08, 2006 8:57 pm
Location: Redmond, WA USA

Re: Critter 1.2 update + new homepage

Post by Dann Corbit »

I'll try Houdini next.
I would be keen to see some long time control games with Critter, but you would need a large bank of good hardware to get there.