OpenChess

If it is better than Rybka 4, then why is it not being entered into the WCCC

1) I don't claim that Houdini is better than Rybka 4.

2) You've asked exactly the same question two months ago at the Talkchess forum (on June 3, 2010, see topic "Houdini And The WCCC"), let me copy my answer for the benefit of other readers:

Robert Houdart wrote:Good question, with a simple answer: time, money, relevance.

Time & Money
A week in Japan would be a very expensive undertaking. Apart from the voyage and the time spent, it would also mean investing in competitive hardware, an opening book, etc. If somebody can find a sponsor for this activity, I'd certainly consider participating.

Limited Relevance of this kind of event
An 11 round event is hardly an appropriate way to decide which computer engine is the strongest. It's a remnant from the pre-Internet era in which people actually had to meet in person to play computer matches. For engine matches 11 games simply isn't enough... remember that to validate a 20 Elo difference it is necessary to play about 1000 games.

Have a nice day.

And of course there is the ip*/robo*/etc issue, which would make it extremely unlikely the ICGA would allow the program due to originality issues...

Robert Houdart wrote:1) I don't claim that Houdini is better than Rybka 4.

2) You've asked exactly the same question two months ago at the Talkchess forum (on June 3, 2010, see topic "Houdini And The WCCC"), let me copy my answer for the benefit of other readers:

Robert Houdart wrote: Good question, with a simple answer: time, money, relevance.

Time & Money
A week in Japan would be a very expensive undertaking. Apart from the voyage and the time spent, it would also mean investing in competitive hardware, an opening book, etc. If somebody can find a sponsor for this activity, I'd certainly consider participating.

Rubbish, have someone else run the program that lives at the location!

Robert Houdart wrote:Limited Relevance of this kind of event
An 11 round event is hardly an appropriate way to decide which computer engine is the strongest. It's a remnant from the pre-Internet era in which people actually had to meet in person to play computer matches. For engine matches 11 games simply isn't enough... remember that to validate a 20 Elo difference it is necessary to play about 1000 games.

Rubbish, this event title is sought by many of the top chess programs.

Robert Houdart wrote:Have a nice day.

Perhaps, your day would not be so nice if you entered Houdini and it was determined to be a clone of a clone! In other words, you are reaping the benefits of someone else's labour and you give them zero credit for it.

Here is a direct and easy question for you: Did you write Houdini from scratch, i.e. is it all your own code?

Cordially,

Sean

And of course there is the ip*/robo*/etc issue, which would make it extremely unlikely the ICGA would allow the program due to originality issues...

There's also the detail that ChessBase is a sponsor of the event (which it was not in 2009, though prior to that it was in various years).

Rubbish, have someone else run the program that lives at the location!

The ICGA charges a lot for an "outside operator" (500 euros or something I think -- I can't even find mention of this option this time around, so maybe I'm thinking of a different event?), though it seems that almost anyone can be considered a "team member" in some cases. IMO the whole idea of "operators" is outdated. Another note would be that there would be a good chance of a travel subsidy (1000 euros).

Rubbish, this event title is sought by many of the top chess programs.

For commercial products, this is more true. But I don't expect Stockfish to compete, for instance. The dire need for an opening book is also an issue. As an example, Zach has inherited Zappa's book author (Günes -- who is also listed as a book author for Sjeng), or else he'd be paddling uphill here. VR made the astute decision to hire Noomen (and later others) at an early stage. I think the fact that these events are held annually now tends to reduce their prestige also (why compete this year in Japan, when the event is likely to be a short European jaunt the next time around?).

hyatt wrote:And of course there is the ip*/robo*/etc issue, which would make it extremely unlikely the ICGA would allow the program due to originality issues...

When the time and the conditions are right (maybe next year, who knows), I'll duly present the source code to whoever is in charge of this decision.

Robert

Sean Evans wrote: Perhaps, your day would not be so nice if you entered Houdini and it was determined to be a clone of a clone! In other words, you are reaping the benefits of someone else's labour and you give them zero credit for it.

Here is a direct and easy question for you: Did you write Houdini from scratch, i.e. is it all your own code?

Sean, exactly like two months ago, it appears that you're not really interested in my answers. You just seek a stage for expressing your own preconceived opinions.

To answer your direct and easy question with a direct and easy answer, Houdini is all my development, and I've always given due credit in the readme file, on the web site, and in the various forums.
As I replied to Bob, when the time and the conditions are right (maybe next year, who knows), I'll duly present the source code to whoever is in charge of the WCCC.

Robert

BB+ wrote:The dire need for an opening book is also an issue.

This is crucial. Most people underestimate the importance of Jeroen Noomen's contribution to Rybka's WCCC successes.
Another factor is hardware, it would be ridiculous to go with an ordinary high-end computer and get slaughtered by Rybka and Deep Sjeng clusters. One would probably need at least 10k (€ or $) to construct some more or less competitive hardware.

Then again, all this wasn't the point of Sean's question. He just wanted a stage for using one more time the phrase "a clone of a clone"...

Robert

Sean Evans wrote:If it is better than Rybka 4, then why is it not being entered into the WCCC

HI Sean.
Houdini is playing now for the world title. .
Martin is doing a very fine job holding these games.
All games are on the same hardware so its more even .
Check it out.

Best,
Gerold.

P.S. WHo needs the xyzs .

Robert Houdart wrote:

Sean Evans wrote: Perhaps, your day would not be so nice if you entered Houdini and it was determined to be a clone of a clone! In other words, you are reaping the benefits of someone else's labour and you give them zero credit for it.

Here is a direct and easy question for you: Did you write Houdini from scratch, i.e. is it all your own code?

Sean, exactly like two months ago, it appears that you're not really interested in my answers. You just seek a stage for expressing your own preconceived opinions.

To answer your direct and easy question with a direct and easy answer, Houdini is all my development, and I've always given due credit in the readme file, on the web site, and in the various forums.
As I replied to Bob, when the time and the conditions are right (maybe next year, who knows), I'll duly present the source code to whoever is in charge of the WCCC.

Robert

Robert, you can stop the charade, the evidence was already posted that Houdini is an IPPO family clone, case closed.

Cordially,

Sean