The silence of Robert Houdart

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

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
mhull
Posts: 12263
Joined: Wed Mar 08, 2006 8:02 pm
Location: Dallas, Texas
Full name: Matthew Hull

Re: The silence of Robert Houdart

Post by mhull » Wed Mar 16, 2011 4:31 pm

benstoker wrote:Maybe I got it all wrong. So, what exactly is one doing when picking through the ippo* code trying to figure it out to see if any ideas can be discovered to implement in one's own engine? What is that called? Ditto disassembly of Houdini, Rybka, etc.? You disagree with Muller that Fabien's accomplishment and originality is relegated to little more than debugging? Why do you disagree with that, if you do? If good debugging earns one the "originality" predicate a la Muller, then why does Don dismiss Richard's opinion that Houdart has done something original? What are your percentages? Dear sir, please divide up each of the Crafty functions below into two groups: 1-1 and 1-many. Let's start there. Please don't bother, if all you got to say is "you know it when you see it". We have a long ways to go; but, you appear to have interest in defining "originality". In your opinion, what % of chess engine source code is 1:1 and what percentage is 1:many? Are there sub-parts within search() that are 1:1? What category do magic bitboards fall into? kpk bitbases go into the 1:1 bucket, right? For, instance, if I copy the following null move code into my search(), does that diminish "originality"?:


Probably every top engine has source code that is virtually identical to the above null-move routine. Would you penalize my "originality" if I literally copied the above code into my search()? Why? What if I throw in Tabibi's verified null move code? Is null-move 1:1? It is, isn't it. Everything around it may not be, but null move reduction is. Or isn't?
It's not much of a stretch to realize you're vicariously developing a defense of your own actions here -- beads of sweat having formed on your own brow for some time now.
Matthew Hull

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

Re: The silence of Robert Houdart

Post by benstoker » Wed Mar 16, 2011 4:55 pm

michiguel wrote:
benstoker wrote:
Albert Silver wrote:
benstoker wrote:
Albert Silver wrote:
benstoker wrote:
Houdini wrote:@Carlos, The two options you present are not mutually exclusive :D. But yes, I'm currently working hard on Houdini 2.0.

@Gerold, I'm not in the least influenced by the Fruit/Rybka thing, I have no dealings with the ICGA.

Cheers,
Robert
Some things we don't know. Other things we do know. And we do know, on the word of Don Dailey, that even if you, Mr. Houdart, muster another +50 ELO out of Houdini 2.0, it won't be the result of any new chess engine ideas you will have created, but just the handiwork of a skilled debugger.

You see, although Komodo is 150 ELO below Houdini, it's 100% "original" as to the 1% of the total code that matters, i.e., approx. 5% of the search() and about 3% of the eval() code. Of course, these days, around 99% of the source code is ministerial in nature, one-input-one-output crap functions, public domain junk like magic bitboards, bit counting asm garbage, etc.

It's a pity that fine debuggers like you don't get as much credit as the guys spinning their wheels creating buggy "original" code comprising just around 1% of the engine.

Robert, you and Don should partner up! Think about it. Do you smell the dollars, the euros?! You could explain to Don and Larry the Ippo* code, and then Don could figure out how to rewrite the code so that it's "original". You follow? Then, you, Robert, debug the holy shit out of Don's original rewrite. Sign a 50/50 partnership agreement, sell through Chessbase and go to the bank!
It is really odd that you insist in lambasting Don's work like this.

Why suggest that only 1% of his work is original, that he wants to rewrite IPPO code, and finally that his programming skills are so poor that they require an outsider to debug them?

What do you have against him?
Albert, take this as your opportunity to, for the very first time, define that word of yours, "original". Start with that.
Instead, why don't you answer the question(s)?
Why do you ask for a recapitulation of all the threads over the last year? Too much work. Where the heck were you? Don has expounded at length here why he thinks Houdart has no new engine ideas, but instead has only taken someone else's code and just proven himself to be a very good debugger.
I do not know where you were, but Don's recognized RH talent when Houdini 1.5 was released. For instance, he said:

"...So I guess we both agree that both Vas and Robert Houdart are talented and original thinkers!"
http://www.talkchess.com/forum/viewtopi ... ini#381753
He consistently said that improving on a top engine required talent.

Your sarcasm, accusations, and aggressiveness towards DD are completely out of line.

Miguel
Whata freakin joke.

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

Re: The silence of Robert Houdart

Post by benstoker » Wed Mar 16, 2011 4:56 pm

mhull wrote:
benstoker wrote:Maybe I got it all wrong. So, what exactly is one doing when picking through the ippo* code trying to figure it out to see if any ideas can be discovered to implement in one's own engine? What is that called? Ditto disassembly of Houdini, Rybka, etc.? You disagree with Muller that Fabien's accomplishment and originality is relegated to little more than debugging? Why do you disagree with that, if you do? If good debugging earns one the "originality" predicate a la Muller, then why does Don dismiss Richard's opinion that Houdart has done something original? What are your percentages? Dear sir, please divide up each of the Crafty functions below into two groups: 1-1 and 1-many. Let's start there. Please don't bother, if all you got to say is "you know it when you see it". We have a long ways to go; but, you appear to have interest in defining "originality". In your opinion, what % of chess engine source code is 1:1 and what percentage is 1:many? Are there sub-parts within search() that are 1:1? What category do magic bitboards fall into? kpk bitbases go into the 1:1 bucket, right? For, instance, if I copy the following null move code into my search(), does that diminish "originality"?:


Probably every top engine has source code that is virtually identical to the above null-move routine. Would you penalize my "originality" if I literally copied the above code into my search()? Why? What if I throw in Tabibi's verified null move code? Is null-move 1:1? It is, isn't it. Everything around it may not be, but null move reduction is. Or isn't?
It's not much of a stretch to realize you're vicariously developing a defense of your own actions here -- beads of sweat having formed on your own brow for some time now.
4th down. Punt

Albert Silver
Posts: 2829
Joined: Wed Mar 08, 2006 8:57 pm
Location: Rio de Janeiro, Brazil

Re: The silence of Robert Houdart

Post by Albert Silver » Wed Mar 16, 2011 5:03 pm

benstoker wrote:
michiguel wrote:
benstoker wrote:Why do you ask for a recapitulation of all the threads over the last year? Too much work. Where the heck were you? Don has expounded at length here why he thinks Houdart has no new engine ideas, but instead has only taken someone else's code and just proven himself to be a very good debugger.
I do not know where you were, but Don's recognized RH talent when Houdini 1.5 was released. For instance, he said:

"...So I guess we both agree that both Vas and Robert Houdart are talented and original thinkers!"
http://www.talkchess.com/forum/viewtopi ... ini#381753
He consistently said that improving on a top engine required talent.

Your sarcasm, accusations, and aggressiveness towards DD are completely out of line.

Miguel
Whata freakin joke.
Care to elaborate what you mean?
"Tactics are the bricks and sticks that make up a game, but positional play is the architectural blueprint."

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

Re: The silence of Robert Houdart

Post by benstoker » Wed Mar 16, 2011 5:05 pm

Albert Silver wrote:
benstoker wrote:
michiguel wrote:
benstoker wrote:Why do you ask for a recapitulation of all the threads over the last year? Too much work. Where the heck were you? Don has expounded at length here why he thinks Houdart has no new engine ideas, but instead has only taken someone else's code and just proven himself to be a very good debugger.
I do not know where you were, but Don's recognized RH talent when Houdini 1.5 was released. For instance, he said:

"...So I guess we both agree that both Vas and Robert Houdart are talented and original thinkers!"
http://www.talkchess.com/forum/viewtopi ... ini#381753
He consistently said that improving on a top engine required talent.

Your sarcasm, accusations, and aggressiveness towards DD are completely out of line.

Miguel
Whata freakin joke.
Care to elaborate what you mean?
It's your turn, chump.

User avatar
mhull
Posts: 12263
Joined: Wed Mar 08, 2006 8:02 pm
Location: Dallas, Texas
Full name: Matthew Hull

Re: The silence of Robert Houdart

Post by mhull » Wed Mar 16, 2011 5:17 pm

benstoker wrote:
Albert Silver wrote: Care to elaborate what you mean?
It's your turn, chump.
I don't know why you bother to wear a mask. You spread darkness wherever you go. ;)
Matthew Hull

bob
Posts: 20392
Joined: Mon Feb 27, 2006 6:30 pm
Location: Birmingham, AL

Re: The silence of Robert Houdart

Post by bob » Wed Mar 16, 2011 5:26 pm

mhull wrote:
benstoker wrote:Maybe I got it all wrong. So, what exactly is one doing when picking through the ippo* code trying to figure it out to see if any ideas can be discovered to implement in one's own engine? What is that called? Ditto disassembly of Houdini, Rybka, etc.? You disagree with Muller that Fabien's accomplishment and originality is relegated to little more than debugging? Why do you disagree with that, if you do? If good debugging earns one the "originality" predicate a la Muller, then why does Don dismiss Richard's opinion that Houdart has done something original? What are your percentages? Dear sir, please divide up each of the Crafty functions below into two groups: 1-1 and 1-many. Let's start there. Please don't bother, if all you got to say is "you know it when you see it". We have a long ways to go; but, you appear to have interest in defining "originality". In your opinion, what % of chess engine source code is 1:1 and what percentage is 1:many? Are there sub-parts within search() that are 1:1? What category do magic bitboards fall into? kpk bitbases go into the 1:1 bucket, right? For, instance, if I copy the following null move code into my search(), does that diminish "originality"?:


Probably every top engine has source code that is virtually identical to the above null-move routine. Would you penalize my "originality" if I literally copied the above code into my search()? Why? What if I throw in Tabibi's verified null move code? Is null-move 1:1? It is, isn't it. Everything around it may not be, but null move reduction is. Or isn't?
It's not much of a stretch to realize you're vicariously developing a defense of your own actions here -- beads of sweat having formed on your own brow for some time now.
Just to address parts. Null-move is not one-to-one. There is normal null-move, null-move with an offset window, null-move with verification, double-null-move vs disallowing two consecutive null-moves, etc. I can't think of much in search that would meet my suggested 1:1 test. SEE() is a possible example, since there is a standard SEE() algorithm that most use. One can add absolute or relative pins, but one could define those as SEEA and SEER since they are, individually, 1:1. So this one is not clear, but I would not object to anyone copying my Swap() function, which is where Crafty's SEE algorithm is done.

However, for clarity, SEE might be excluded since there are multiple potential implementations (as above) and probably others.

My suggestion was an alpha-level abstraction. It would need significant refinement to make it to beta and then to production. Whether there is agreement on that or not is unknown. Clearly we need something that is both concise and all-encompassing, so that it can be enforced without a lot of subjective bantering back and forth...

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

Re: The silence of Robert Houdart

Post by benstoker » Wed Mar 16, 2011 6:07 pm

bob wrote:
mhull wrote:
benstoker wrote:Maybe I got it all wrong. So, what exactly is one doing when picking through the ippo* code trying to figure it out to see if any ideas can be discovered to implement in one's own engine? What is that called? Ditto disassembly of Houdini, Rybka, etc.? You disagree with Muller that Fabien's accomplishment and originality is relegated to little more than debugging? Why do you disagree with that, if you do? If good debugging earns one the "originality" predicate a la Muller, then why does Don dismiss Richard's opinion that Houdart has done something original? What are your percentages? Dear sir, please divide up each of the Crafty functions below into two groups: 1-1 and 1-many. Let's start there. Please don't bother, if all you got to say is "you know it when you see it". We have a long ways to go; but, you appear to have interest in defining "originality". In your opinion, what % of chess engine source code is 1:1 and what percentage is 1:many? Are there sub-parts within search() that are 1:1? What category do magic bitboards fall into? kpk bitbases go into the 1:1 bucket, right? For, instance, if I copy the following null move code into my search(), does that diminish "originality"?:


Probably every top engine has source code that is virtually identical to the above null-move routine. Would you penalize my "originality" if I literally copied the above code into my search()? Why? What if I throw in Tabibi's verified null move code? Is null-move 1:1? It is, isn't it. Everything around it may not be, but null move reduction is. Or isn't?
It's not much of a stretch to realize you're vicariously developing a defense of your own actions here -- beads of sweat having formed on your own brow for some time now.
Just to address parts. Null-move is not one-to-one. There is normal null-move, null-move with an offset window, null-move with verification, double-null-move vs disallowing two consecutive null-moves, etc. I can't think of much in search that would meet my suggested 1:1 test. SEE() is a possible example, since there is a standard SEE() algorithm that most use. One can add absolute or relative pins, but one could define those as SEEA and SEER since they are, individually, 1:1. So this one is not clear, but I would not object to anyone copying my Swap() function, which is where Crafty's SEE algorithm is done.

However, for clarity, SEE might be excluded since there are multiple potential implementations (as above) and probably others.

My suggestion was an alpha-level abstraction. It would need significant refinement to make it to beta and then to production. Whether there is agreement on that or not is unknown. Clearly we need something that is both concise and all-encompassing, so that it can be enforced without a lot of subjective bantering back and forth...
Finally, someone advances the ball. Same as anything else: players on the field and drunk spectators in the bleachers flying their colors.

Milos
Posts: 3374
Joined: Wed Nov 25, 2009 12:47 am

Re: The silence of Robert Houdart

Post by Milos » Wed Mar 16, 2011 6:25 pm

bob wrote:Null-move is not one-to-one. There is normal null-move, null-move with an offset window, null-move with verification, double-null-move vs disallowing two consecutive null-moves, etc. I can't think of much in search that would meet my suggested 1:1 test.
This is a false logic. You can have a 1000 different implementations of null-move. They'd all be crap (offset window, verification, double null-move, just bunch of things that don't work, a graveyard of dead ideas, some "clever" ppl once thought are smart, and that today bring only misconception) and one and only one is there that is working.
However, there is a thing that is important. And that's null-move reduction implementation. It differs good from great (are you still living in the prehistoric times of Heinz, or not). However, it is far to simple to be any criterion for originality.

Albert Silver
Posts: 2829
Joined: Wed Mar 08, 2006 8:57 pm
Location: Rio de Janeiro, Brazil

Re: The silence of Robert Houdart

Post by Albert Silver » Wed Mar 16, 2011 6:50 pm

benstoker wrote:
Albert Silver wrote:
benstoker wrote:
michiguel wrote:
benstoker wrote:Why do you ask for a recapitulation of all the threads over the last year? Too much work. Where the heck were you? Don has expounded at length here why he thinks Houdart has no new engine ideas, but instead has only taken someone else's code and just proven himself to be a very good debugger.
I do not know where you were, but Don's recognized RH talent when Houdini 1.5 was released. For instance, he said:

"...So I guess we both agree that both Vas and Robert Houdart are talented and original thinkers!"
http://www.talkchess.com/forum/viewtopi ... ini#381753
He consistently said that improving on a top engine required talent.

Your sarcasm, accusations, and aggressiveness towards DD are completely out of line.

Miguel
Whata freakin joke.
Care to elaborate what you mean?
It's your turn, chump.
So be it. FWIW, I hadn't classified you as a troll, and did give you the benefit of the doubt.
"Tactics are the bricks and sticks that make up a game, but positional play is the architectural blueprint."

Post Reply