Why are the Ippo derivative stronger than Stockfish?

Discussion of chess software programming and technical issues.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
michiguel
Posts: 6388
Joined: Thu Mar 09, 2006 7:30 pm
Location: Chicago, Illinois, USA
Contact:

Re: Why are the Ippo derivative stronger than Stockfish?

Post by michiguel » Mon Feb 28, 2011 6:02 am

benstoker wrote:
mcostalba wrote:
Houdini wrote: Again, I didn't blame, I suggested giving credit.
Robert, I think you've done a good job with Houdini, actually you are the only one that was able to obtain such a big boost out of Ippo. So I have already written and I have no problem to write again that I think you deserve sure credit for the improving you got with Houdini.

But I don't think would be correct to align all the developers on a single line: "let everybody give credit to Ippo becuase everybody took ideas from that !".

I think is not correct doing so because we blur the line between taking ideas and taking more than that.

If Richard says that Houdini was easy to disassemble because he already got 90% of source code, this for me is an important fact, also because I trust a lot Richard and I know he is not the guy that gives random numbers....

Now I stop because I think you already got the message. Anyhow let's make a funny deal: you write on your site that you started out of Ippo sources and we write on our that we have took ideas from Ippo :-) What do you think ?
Until such time as YOU or the ICGA panel fashions a sufficient definition of an impermissible derivative based on a set of specific engineering criteria, this casting of petty apsersions is just so much bullshit. You know nothing. You don't know what it means to say "got 90% of source code ...". It means absolutely nothing. If you cannot provide a definition and set of criteria for determining an impermissible derivative, then shut up. If you can provide such, then offer it to the ICGA to be vetted.
In the previous post, you taunted a GM saying the he is jealous. Here, you tell one of the authors of strongest GPL engine of the world that he knows nothing and that he should shut up. In both cases, it is a conversation between programmers and technical experts in the technical subforum. This is as subforum in which you previously asked questions and the people graciously answered you. Your lack of manners is unbearable, but that is not critical. The main problem is this is how a cesspool of trolling battles start. I am going to ask you a favor. Could you please redirect all this lack of substance to another subforum?

Miguel

Roger Brown
Posts: 782
Joined: Wed Mar 08, 2006 8:22 pm

Re: Why are the Ippo derivative stronger than Stockfish?

Post by Roger Brown » Mon Feb 28, 2011 7:19 am

michiguel wrote:
In the previous post, you taunted a GM saying the he is jealous. Here, you tell one of the authors of strongest GPL engine of the world that he knows nothing and that he should shut up. In both cases, it is a conversation between programmers and technical experts in the technical subforum. This is as subforum in which you previously asked questions and the people graciously answered you. Your lack of manners is unbearable, but that is not critical. The main problem is this is how a cesspool of trolling battles start. I am going to ask you a favor. Could you please redirect all this lack of substance to another subforum?

Miguel


Hello Miguel,

No, no, no, no!

There will be no lack of substance redirected anywhere.

A lack of substance should not be allowed to exist anywhere on this forum.

I will ask that there be no responses to trigger words and that the technical discussions continue.

I believe that Ben may have issues with how Robert Houdart is described which is provoking a response.

Later.

mcostalba
Posts: 2684
Joined: Sat Jun 14, 2008 7:17 pm

Re: Why are the Ippo derivative stronger than Stockfish?

Post by mcostalba » Mon Feb 28, 2011 12:39 pm

Roger Brown wrote: There will be no lack of substance redirected anywhere.

A lack of substance should not be allowed to exist anywhere on this forum.

I will ask that there be no responses to trigger words and that the technical discussions continue.
I was starting to answer to that idiot but then I read your post and I will follow your suggestion to avoid garbaging this interesting thread with main forum crap-style material. I understand that you said:"A lack of substance should not be allowed to exist anywhere on this forum." but is a fact that reality is different and some subforums are more prone to trolling than others ;-)

It would be a pity if not-developers people do start to pollute this technical threads for no reason at all BTW.

Richard made a series of _very_ interesting posts, for instance I didn't know it exsist an horizont effect exstension (as Don says it was already known to him) and I would like to ask someone to point me to some literature on the subject.

Michel
Posts: 2047
Joined: Sun Sep 28, 2008 11:50 pm

Re: Why are the Ippo derivative stronger than Stockfish?

Post by Michel » Mon Feb 28, 2011 1:20 pm

Even poor GnuChess has a horizon extension. I never understood what it is supposed to do and (foolishly?) disabled it in my branch.

User avatar
Don
Posts: 5106
Joined: Tue Apr 29, 2008 2:27 pm

Re: Why are the Ippo derivative stronger than Stockfish?

Post by Don » Mon Feb 28, 2011 1:36 pm

mcostalba wrote:
Roger Brown wrote: There will be no lack of substance redirected anywhere.

A lack of substance should not be allowed to exist anywhere on this forum.

I will ask that there be no responses to trigger words and that the technical discussions continue.
I was starting to answer to that idiot but then I read your post and I will follow your suggestion to avoid garbaging this interesting thread with main forum crap-style material. I understand that you said:"A lack of substance should not be allowed to exist anywhere on this forum." but is a fact that reality is different and some subforums are more prone to trolling than others ;-)

It would be a pity if not-developers people do start to pollute this technical threads for no reason at all BTW.

Richard made a series of _very_ interesting posts, for instance I didn't know it exsist an horizont effect exstension (as Don says it was already known to him) and I would like to ask someone to point me to some literature on the subject.
Every idea he pointed out is well known (to the extend we can tell from his lack of implementation details), with the possible exception of not reducing null threats - that is an idea I know about but I am not aware of anyone else using it. I don't view this idea as something very interesting, it's an obvious thing to try and it does not work for Komodo. For Komodo it only slows the program down for little ELO gain but it might help other programs.

Extending moves on the last ply is a very old idea but since Richard was not very specific I have no way of knowing if this extension has anything to do with the horizon thing he talked about.

Many years ago I had a conversation about measuring the horizon affect and enabling the program to react to it. I will tell you what I remember - we did not produce a specific proposal but just some ideas.

The primary characteristic of horizon moves we talked about is that they are high compulsion moves, captures and checks and threats. If the only move that "works" in a position is a capture or check that apparently accomplishes nothing special, it's probably a horizon move. It's probably not too difficult to come up with specific proposals on how this might be implemented.

Another idea of my own is very simple to implement but I have not tried it, as I don't expect it to work :-) At PV nodes, if the searched score of the first move drops by some margin (as compared to the hash table score) then you probably have a horizon issue. This could be tested by doing a zero window search on the first move instead of immediately going into the full window PV search.

User avatar
Don
Posts: 5106
Joined: Tue Apr 29, 2008 2:27 pm

Re: Why are the Ippo derivative stronger than Stockfish?

Post by Don » Mon Feb 28, 2011 1:47 pm

michiguel wrote:
benstoker wrote:
mcostalba wrote:
Houdini wrote: Again, I didn't blame, I suggested giving credit.
Robert, I think you've done a good job with Houdini, actually you are the only one that was able to obtain such a big boost out of Ippo. So I have already written and I have no problem to write again that I think you deserve sure credit for the improving you got with Houdini.

But I don't think would be correct to align all the developers on a single line: "let everybody give credit to Ippo becuase everybody took ideas from that !".

I think is not correct doing so because we blur the line between taking ideas and taking more than that.

If Richard says that Houdini was easy to disassemble because he already got 90% of source code, this for me is an important fact, also because I trust a lot Richard and I know he is not the guy that gives random numbers....

Now I stop because I think you already got the message. Anyhow let's make a funny deal: you write on your site that you started out of Ippo sources and we write on our that we have took ideas from Ippo :-) What do you think ?
Until such time as YOU or the ICGA panel fashions a sufficient definition of an impermissible derivative based on a set of specific engineering criteria, this casting of petty apsersions is just so much bullshit. You know nothing. You don't know what it means to say "got 90% of source code ...". It means absolutely nothing. If you cannot provide a definition and set of criteria for determining an impermissible derivative, then shut up. If you can provide such, then offer it to the ICGA to be vetted.
In the previous post, you taunted a GM saying the he is jealous. Here, you tell one of the authors of strongest GPL engine of the world that he knows nothing and that he should shut up. In both cases, it is a conversation between programmers and technical experts in the technical subforum. This is as subforum in which you previously asked questions and the people graciously answered you. Your lack of manners is unbearable, but that is not critical. The main problem is this is how a cesspool of trolling battles start. I am going to ask you a favor. Could you please redirect all this lack of substance to another subforum?

Miguel
Indeed. I personally feel that this is a forum for experts and that if you are not an expert you need to exercise a little restraint, respect and good manners.

This is partly our fault (as experts and program authors) because we (and I have been especially guilty) have continued to entertain idiots and by doing so we validate them. I have stopped doing this recently and I hope the others do also. There is no reason for us not to be selective about who we respond to as it's really OUR forum, not theirs. They have the general topics forum where they can flame each other.

benstoker
Posts: 342
Joined: Tue Jan 19, 2010 1:05 am

Re: Why are the Ippo derivative stronger than Stockfish?

Post by benstoker » Mon Feb 28, 2011 2:21 pm

That is right. This is a technical forum. That is the point. So please waste no more time denouncing all who question your judgment and in this forum once and for all, offer some TECHNICAL criteria for determining a derivative as opposed to an "original". Proffer will you not TECHNICAL criteria for adding the adjective "impermissible" to the word "derivative". Provide a definition of your word "innovation" as distinguished from eliminating bugs. You say Houdart has not innovated by using the techniques the Critter author described. You say Houdart has not innovated by creating the strongest engine in the world.

But you and LK pick through the code of stockfish and ippo looking for something, something to add to your OWN engine. If you take from ippo, is that "innovation"? Again, give us a set of TECHNICAL criteria to identify impermissible code copying from permissible Idea borrowing. Hyatt has. Why haven't you? It is blatantly hypocritical and atrocious to at once pick through stockfish and ippo on the one hand, and then indict Houdart as a copier, and nothing more than a fine engineer.

If you are the innovator, do turn a blind eye to open source code please.
Don wrote:
michiguel wrote:
benstoker wrote:
mcostalba wrote:
Houdini wrote: Again, I didn't blame, I suggested giving credit.
Robert, I think you've done a good job with Houdini, actually you are the only one that was able to obtain such a big boost out of Ippo. So I have already written and I have no problem to write again that I think you deserve sure credit for the improving you got with Houdini.

But I don't think would be correct to align all the developers on a single line: "let everybody give credit to Ippo becuase everybody took ideas from that !".

I think is not correct doing so because we blur the line between taking ideas and taking more than that.

If Richard says that Houdini was easy to disassemble because he already got 90% of source code, this for me is an important fact, also because I trust a lot Richard and I know he is not the guy that gives random numbers....

Now I stop because I think you already got the message. Anyhow let's make a funny deal: you write on your site that you started out of Ippo sources and we write on our that we have took ideas from Ippo :-) What do you think ?
Until such time as YOU or the ICGA panel fashions a sufficient definition of an impermissible derivative based on a set of specific engineering criteria, this casting of petty apsersions is just so much bullshit. You know nothing. You don't know what it means to say "got 90% of source code ...". It means absolutely nothing. If you cannot provide a definition and set of criteria for determining an impermissible derivative, then shut up. If you can provide such, then offer it to the ICGA to be vetted.
In the previous post, you taunted a GM saying the he is jealous. Here, you tell one of the authors of strongest GPL engine of the world that he knows nothing and that he should shut up. In both cases, it is a conversation between programmers and technical experts in the technical subforum. This is as subforum in which you previously asked questions and the people graciously answered you. Your lack of manners is unbearable, but that is not critical. The main problem is this is how a cesspool of trolling battles start. I am going to ask you a favor. Could you please redirect all this lack of substance to another subforum?

Miguel
Indeed. I personally feel that this is a forum for experts and that if you are not an expert you need to exercise a little restraint, respect and good manners.

This is partly our fault (as experts and program authors) because we (and I have been especially guilty) have continued to entertain idiots and by doing so we validate them. I have stopped doing this recently and I hope the others do also. There is no reason for us not to be selective about who we respond to as it's really OUR forum, not theirs. They have the general topics forum where they can flame each other.

bhlangonijr
Posts: 482
Joined: Thu Oct 16, 2008 2:23 am
Location: Milky Way
Contact:

Re: Why are the Ippo derivative stronger than Stockfish?

Post by bhlangonijr » Mon Feb 28, 2011 2:57 pm

benstoker wrote:That is right. This is a technical forum. That is the point. So please waste no more time denouncing all who question your judgment and in this forum once and for all, offer some TECHNICAL criteria for determining a derivative as opposed to an "original". Proffer will you not TECHNICAL criteria for adding the adjective "impermissible" to the word "derivative". Provide a definition of your word "innovation" as distinguished from eliminating bugs. You say Houdart has not innovated by using the techniques the Critter author described. You say Houdart has not innovated by creating the strongest engine in the world.

But you and LK pick through the code of stockfish and ippo looking for something, something to add to your OWN engine. If you take from ippo, is that "innovation"? Again, give us a set of TECHNICAL criteria to identify impermissible code copying from permissible Idea borrowing. Hyatt has. Why haven't you? It is blatantly hypocritical and atrocious to at once pick through stockfish and ippo on the one hand, and then indict Houdart as a copier, and nothing more than a fine engineer.

If you are the innovator, do turn a blind eye to open source code please.
What you fail to understand is that the "criteria" you are claiming for is only needed when you have a thin line which separates a legal and original program from a non-legal one. If you pick Stockfish or Komodo they are original programs created from scratch, which happens to use some Ippo ideas that worked for them. In the other hand, there is a expert consensus that Houdini was created out of Ippo sources and closed afterwards, a fact that is constantly denied by the "author". That lack of honesty is reason enough to void any positive adjectives one can make about Robert Houdart IMHO.

User avatar
Don
Posts: 5106
Joined: Tue Apr 29, 2008 2:27 pm

Re: Why are the Ippo derivative stronger than Stockfish?

Post by Don » Mon Feb 28, 2011 4:44 pm

Don wrote:
mcostalba wrote:
Roger Brown wrote: There will be no lack of substance redirected anywhere.

A lack of substance should not be allowed to exist anywhere on this forum.

I will ask that there be no responses to trigger words and that the technical discussions continue.
I was starting to answer to that idiot but then I read your post and I will follow your suggestion to avoid garbaging this interesting thread with main forum crap-style material. I understand that you said:"A lack of substance should not be allowed to exist anywhere on this forum." but is a fact that reality is different and some subforums are more prone to trolling than others ;-)

It would be a pity if not-developers people do start to pollute this technical threads for no reason at all BTW.

Richard made a series of _very_ interesting posts, for instance I didn't know it exsist an horizont effect exstension (as Don says it was already known to him) and I would like to ask someone to point me to some literature on the subject.
Every idea he pointed out is well known (to the extend we can tell from his lack of implementation details),
addendum: I was informed by someone that ideas of horizon effect extensions are decades old and comes in many forms.

For example from Phalanx in 1997:

Code: Select all

New extension trick that helps in some horizon-effect type positions, it's based on measuring difference between current score and null move result; if current score is much better, the side to move must be under threat and the line is extended, similar trick is also in gnuchess.
Also there is a paper from 1983 called Searching to Variable Depth (Kaindl) which is about horizon issues and references to a paper in 1965 by Knuth-Moore concerning variable depth searching to deal with horizon issues. I have not looked at the paper myself.

... with the possible exception of not reducing null threats - that is an idea I know about but I am not aware of anyone else using it. I don't view this idea as something very interesting, it's an obvious thing to try and it does not work for Komodo. For Komodo it only slows the program down for little ELO gain but it might help other programs.

Extending moves on the last ply is a very old idea but since Richard was not very specific I have no way of knowing if this extension has anything to do with the horizon thing he talked about.

Many years ago I had a conversation about measuring the horizon affect and enabling the program to react to it. I will tell you what I remember - we did not produce a specific proposal but just some ideas.

The primary characteristic of horizon moves we talked about is that they are high compulsion moves, captures and checks and threats. If the only move that "works" in a position is a capture or check that apparently accomplishes nothing special, it's probably a horizon move. It's probably not too difficult to come up with specific proposals on how this might be implemented.

Another idea of my own is very simple to implement but I have not tried it, as I don't expect it to work :-) At PV nodes, if the searched score of the first move drops by some margin (as compared to the hash table score) then you probably have a horizon issue. This could be tested by doing a zero window search on the first move instead of immediately going into the full window PV search.

mcostalba
Posts: 2684
Joined: Sat Jun 14, 2008 7:17 pm

Re: Why are the Ippo derivative stronger than Stockfish?

Post by mcostalba » Mon Feb 28, 2011 5:10 pm

I have also received some pm with very interesting material regarding horizont effect stuff.

I have asked him permission to re-post and as soon as I get I'll re-post here....

Post Reply