IPON results for Houdini 4

Discussion of computer chess matches and engine tournaments.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
Laskos
Posts: 10240
Joined: Wed Jul 26, 2006 8:21 pm
Full name: Kai Laskos

Re: IPON results for Houdini 4

Post by Laskos » Thu Dec 05, 2013 9:57 am

Laskos wrote:
pohl4711 wrote:
Laskos wrote: And you probably weakened the engine by 200-500 ELO points. No wonder you will get a low similarity...
Definitly not! It is a aggressive-playing setting, which I used some time ago on playchess. It is perhaps 30 Elo weaker than Stockfish default, definitly not 200-500...

Stefan
Maybe you can post here the settings, a self-play match (not necessarily very long, just to have the order of magnitude of weakening), and the results from Sim.
I set some extreme values:
Aggressiveness=200
Cowardice=0
Mobility (Midgame)=150
Mobility (Endgame)=150

Similarity is 61%. Engine is weakened by ~180 ELO points. Would be curious how did you get 53% similarity weakening by only 30 points.

IGarcia
Posts: 543
Joined: Mon Jul 05, 2010 8:27 pm

Re: IPON results for Houdini 4

Post by IGarcia » Thu Dec 05, 2013 11:25 am

Modern Times wrote:
IGarcia wrote: The fact you are selling H2, H3, H4 and nobody is suing you, does not make your engine original. It has been marked as controversial (along with Rybka and other engines) in CCRL and by community in general.
CCRL got rid of the controversial designation a long time ago. It is too subjective, and it isn't our job to be making judgements like that. [...].

Well, at one point CCRL refused to test clones. When then they had to delete Rybka an houdini from lists the come with this thing to classify as "controversial" and keep going.

I was going to point CCRL not longer indicates controversial engines, but it does not mater. I mentioned as an example to remember it was clear for community the unclear origin of Houdini (and other engines).

The important is the TRUE, like H4 is the top engine, but its also true is controversial and its funny to see how fast people forget that, even the author.

Regards.

Modern Times
Posts: 2554
Joined: Thu Jun 07, 2012 9:02 pm

Re: IPON results for Houdini 4

Post by Modern Times » Thu Dec 05, 2013 11:35 am

IGarcia wrote: Well, at one point CCRL refused to test clones. When then they had to delete Rybka an houdini from lists the come with this thing to classify as "controversial" and keep going.
We never "had to delete Rybka". Personally I regard Rybka as an original engine, with everything that has come out after the ICGA "verdict".

We are in an impossible position with clones and derivatives - we aren't qualified to make judgements like that. So we took away all restrictions. Test everything or nothing, that is what it came down to.

User avatar
Houdini
Posts: 1471
Joined: Mon Mar 15, 2010 11:00 pm
Contact:

Re: IPON results for Houdini 4

Post by Houdini » Thu Dec 05, 2013 11:51 am

IGarcia wrote:The important is the TRUE, like H4 is the top engine, but its also true is controversial and its funny to see how fast people forget that, even the author.
Outside the narrow world of Talkchess geeks there is no controversy about Houdini.
To put it differently, the so-called "controversy" is entirely inside your head ;).

Pablo Vazquez
Posts: 154
Joined: Thu May 31, 2007 7:05 pm
Location: Madrid, Spain
Contact:

Re: IPON results for Houdini 4

Post by Pablo Vazquez » Thu Dec 05, 2013 12:52 pm

IGarcia wrote:
Modern Times wrote:
IGarcia wrote: The fact you are selling H2, H3, H4 and nobody is suing you, does not make your engine original. It has been marked as controversial (along with Rybka and other engines) in CCRL and by community in general.
CCRL got rid of the controversial designation a long time ago. It is too subjective, and it isn't our job to be making judgements like that. [...].

Well, at one point CCRL refused to test clones. When then they had to delete Rybka an houdini from lists the come with this thing to classify as "controversial" and keep going.

I was going to point CCRL not longer indicates controversial engines, but it does not mater. I mentioned as an example to remember it was clear for community the unclear origin of Houdini (and other engines).

The important is the TRUE, like H4 is the top engine, but its also true is controversial and its funny to see how fast people forget that, even the author.

Regards.
The origin of Houdini is not unclear at all, so I don't think there's actually any controversy about that engine.

User avatar
kranium
Posts: 1877
Joined: Thu May 29, 2008 8:43 am

Re: IPON results for Houdini 4

Post by kranium » Thu Dec 05, 2013 2:24 pm

Houdini wrote:
IGarcia wrote:The important is the TRUE, like H4 is the top engine, but its also true is controversial and its funny to see how fast people forget that, even the author.
Outside the narrow world of Talkchess geeks there is no controversy about Houdini.
To put it differently, the so-called "controversy" is entirely inside your head ;).
Pablo Vazquez wrote:
IGarcia wrote:
Modern Times wrote:
IGarcia wrote: The fact you are selling H2, H3, H4 and nobody is suing you, does not make your engine original. It has been marked as controversial (along with Rybka and other engines) in CCRL and by community in general.
CCRL got rid of the controversial designation a long time ago. It is too subjective, and it isn't our job to be making judgements like that. [...].

Well, at one point CCRL refused to test clones. When then they had to delete Rybka an houdini from lists the come with this thing to classify as "controversial" and keep going.

I was going to point CCRL not longer indicates controversial engines, but it does not mater. I mentioned as an example to remember it was clear for community the unclear origin of Houdini (and other engines).

The important is the TRUE, like H4 is the top engine, but its also true is controversial and its funny to see how fast people forget that, even the author.

Regards.
The origin of Houdini is not unclear at all, so I don't think there's actually any controversy about that engine.
info about "the Ippolit controversy" can be found everywhere on the Internet, even on Wikipedia,
and it's well documented and well known how Houdini started

regardless of Robert's attempts to whitewash Houdini's history,
it is not exempt from controversy

especially when the author says things like:
"Houdini does NOT contain any Ippolit code."
Last edited by kranium on Thu Dec 05, 2013 2:38 pm, edited 3 times in total.

pohl4711
Posts: 1269
Joined: Sat Sep 03, 2011 5:25 am
Location: Berlin, Germany
Contact:

Re: IPON results for Houdini 4

Post by pohl4711 » Thu Dec 05, 2013 2:32 pm

Laskos wrote:
Laskos wrote:
pohl4711 wrote:
Laskos wrote: And you probably weakened the engine by 200-500 ELO points. No wonder you will get a low similarity...
Definitly not! It is a aggressive-playing setting, which I used some time ago on playchess. It is perhaps 30 Elo weaker than Stockfish default, definitly not 200-500...

Stefan
Maybe you can post here the settings, a self-play match (not necessarily very long, just to have the order of magnitude of weakening), and the results from Sim.
I set some extreme values:
Aggressiveness=200
Cowardice=0
Mobility (Midgame)=150
Mobility (Endgame)=150

Similarity is 61%. Engine is weakened by ~180 ELO points. Would be curious how did you get 53% similarity weakening by only 30 points.
It was an older Stockfish (dont know exactly which one).
But the real point and problem is, that even if the sim-tester would work perfectly, why should 60% be the deadline? And what about an engine with 59.9% or 60.1% similarity?
All deadlines beneath 100% (real clones, only renamed engines etc.) are arbitrarily and not objectively. And because of that problem, I decided not to do so and to test all engines, which are not 100%-clones.
I dont know, if this is the best solution, but I dont see a better one.

Stefan

User avatar
michiguel
Posts: 6401
Joined: Thu Mar 09, 2006 7:30 pm
Location: Chicago, Illinois, USA
Contact:

Re: IPON results for Houdini 4

Post by michiguel » Thu Dec 05, 2013 2:57 pm

Laskos wrote:
Laskos wrote:
pohl4711 wrote:
Laskos wrote: And you probably weakened the engine by 200-500 ELO points. No wonder you will get a low similarity...
Definitly not! It is a aggressive-playing setting, which I used some time ago on playchess. It is perhaps 30 Elo weaker than Stockfish default, definitly not 200-500...

Stefan
Maybe you can post here the settings, a self-play match (not necessarily very long, just to have the order of magnitude of weakening), and the results from Sim.
I set some extreme values:
Aggressiveness=200
Cowardice=0
Mobility (Midgame)=150
Mobility (Endgame)=150

Similarity is 61%. Engine is weakened by ~180 ELO points. Would be curious how did you get 53% similarity weakening by only 30 points.
I do not believe it until I see it. Unless of course it is done renaming engines or something like it.

Miguel

User avatar
michiguel
Posts: 6401
Joined: Thu Mar 09, 2006 7:30 pm
Location: Chicago, Illinois, USA
Contact:

Re: IPON results for Houdini 4

Post by michiguel » Thu Dec 05, 2013 3:10 pm

pohl4711 wrote:
Laskos wrote:
Laskos wrote:
pohl4711 wrote:
Laskos wrote: And you probably weakened the engine by 200-500 ELO points. No wonder you will get a low similarity...
Definitly not! It is a aggressive-playing setting, which I used some time ago on playchess. It is perhaps 30 Elo weaker than Stockfish default, definitly not 200-500...

Stefan
Maybe you can post here the settings, a self-play match (not necessarily very long, just to have the order of magnitude of weakening), and the results from Sim.
I set some extreme values:
Aggressiveness=200
Cowardice=0
Mobility (Midgame)=150
Mobility (Endgame)=150

Similarity is 61%. Engine is weakened by ~180 ELO points. Would be curious how did you get 53% similarity weakening by only 30 points.
It was an older Stockfish (dont know exactly which one).
But the real point and problem is, that even if the sim-tester would work perfectly, why should 60% be the deadline? And what about an engine with 59.9% or 60.1% similarity?
All deadlines beneath 100% (real clones, only renamed engines etc.) are arbitrarily and not objectively. And because of that problem, I decided not to do so and to test all engines, which are not 100%-clones.
I dont know, if this is the best solution, but I dont see a better one.

Stefan
A real clone won't give you 100% in that test because the non-deterministic noise. The similarity test do not detect clones, but "look-alikes" to be precise. The connection between look-alikes and clones is not directly intended by the tool, but in some cases it is an obvious interpretation.

So, the blanket statement that the tool does not work is misleading. It woks perfectly for the purpose it was designed.

Miguel

ouachita
Posts: 454
Joined: Tue Jan 15, 2013 3:33 pm
Location: Ritz-Carlton, NYC
Full name: Bobby Johnson

Why does it matter?

Post by ouachita » Thu Dec 05, 2013 3:15 pm

Houdini wrote:"Houdini does NOT contain any Ippolit code."
I see this seemingly neverending stream of claims that this engine's origin is that, and this engine's origin is this, etc., but I fail to the see the importance of such allegations. Why does it matter?

If Ippolit or any developer feels that Robert or anyone else has misappropriated its code, then such developer can pursue legal action. I have seen nothing on the internet to suggest that Robert and/or his company have been or may be sued for any such reason.

If it's not a violation of some international law, in the end it doesn't really matter and is immaterial to most of us.

Are these criticisms mainly born from Robert's alleged failure to provide adequate attribution to some earlier engine developer or . . . what?

My comments are also aimed at Rybka which would probably still be #1 but-for such code disputes.
SIM, PhD, MBA, PE

Post Reply