Houdini 4 is still something - only 77 ELO behind Komodo 9.1

[reflectionofpower]

On the current CCRL list. Only 77 ELO behind the top one - Komodo 9.1 and it's been a little over 19 months since it's release!

http://computerchess.org.uk/ccrl/4040/

I still like its style very much for my human games.

[Sean Evans]

77 ELO is a large gap in computer chess!

[reflectionofpower]

77 ELO is a large gap in computer chess!

But it was released 19 months ago. That's the point I am making. It has been 19 months and Komodo 9.1 is only 77 points stronger and Stockfish 6 is only 45 points stronger. Where would Houdini be if it was still being actively developed? I would venture to say it would be #1 or close to it.

19 months is a large gap too.

[supersharp77]

Agreed.......my tests show Houdini IDeA & Houdini 4 not far at all behind the top programs.....still competitive.......still investigating the Houdini cfg exe file and what effect that has on overall strength......

[Eelco de Groot]

On the current CCRL list. Only 77 ELO behind the top one - Komodo 9.1 and it's been a little over 19 months since it's release!

http://computerchess.org.uk/ccrl/4040/

I still like its style very much for my human games.

There was a time when successive versions of Shredder did not make more than say 7 Elo per year, and it was enough to stay at the top. Before Fruit and Toga showed what was hidden in terms of possible progress and the mathematical optimization of a numerical approximation function to chess, so to say. And Vas capitalized on that, adding bitbases from Crafty and some ideas of himself but mainly the industrial testing and tuning the approximation function, against Shredder for his first, -and Free!-, Rybka 1.0 Beta.

Fruit might have kept up, (it had the same basis), going commercial before Vas planted his bomb, but the best version was released only years later and Fabien had left again. Vas stayed and added singular extensions and more tuning. Larry brought in his grandmaster perspective.

Nobody could keep up with Rybka, until it was decoded. Everyone 'profited' and the industrial testing from Rybka and the open source, including that 'stolen' from Rybka, made progress faster. But from the time of Shredder's dominance seen, Houdini is now 11 years behind...

[rcmaddox]

On the current CCRL list. Only 77 ELO behind the top one - Komodo 9.1 and it's been a little over 19 months since it's release!

http://computerchess.org.uk/ccrl/4040/

I still like its style very much for my human games.

I agree with you but I'm convinced it's an illusion. Style, I mean. The only reason I know which engine I'm using is because I selected that engine. If I were to play against (or analyze with) any one of the top engines, with its identity hidden from me, I would be unable to identify the engine based upon my perception of its "style." Maybe some players can do that, but I sure as hell can't.

Dedicated units are another story. I can tell those apart (to a degree) based upon how they play.

That said, I'll still no doubt purchase updated versions of Shredder, HIARCS and Houdini should these ever become available. I really should have my head examined!

[Werewolf]

On the current CCRL list. Only 77 ELO behind the top one - Komodo 9.1 and it's been a little over 19 months since it's release!

http://computerchess.org.uk/ccrl/4040/

I still like its style very much for my human games.

I agree with you but I'm convinced it's an illusion. Style, I mean. The only reason I know which engine I'm using is because I selected that engine. If I were to play against (or analyze with) any one of the top engines, with its identity hidden from me, I would be unable to identify the engine based upon my perception of its "style." Maybe some players can do that, but I sure as hell can't.


Dedicated units are another story. I can tell those apart (to a degree) based upon how they play.

The stronger an entity gets the better it plays all positions and style evaporates.

Personally I think it's a load of marketing BS

[acase]

I like Houdini 4 also, especially it's style, but If you look at the head to head, that is, Komodo 9.1 vs Houdini 4 and Stockfish 6 vs Houdini 4. Houdini 4 got destroyed by those 2 programs by huge margins.


Stockfish scored +43−5=66 against Houdini 4
Komodo 9.1 scored +25−3=31 against Houdini 4


This is also a good indication that improvements in software are in fact more important than hardware improvements.

[reflectionofpower]

I like Houdini 4 also, especially it's style, but If you look at the head to head, that is, Komodo 9.1 vs Houdini 4 and Stockfish 6 vs Houdini 4. Houdini 4 got destroyed by those 2 programs by huge margins.


Stockfish scored +43−5=66 against Houdini 4
Komodo 9.1 scored +25−3=31 against Houdini 4


This is also a good indication that improvements in software are in fact more important than hardware improvements.

Yes, it did not do well but what do you expect from a program that has not been improved in 19 months? In those 19 months Komodo & SF have been worked on consistently and being continually refined so this should be expected.

[acase]

I like Houdini 4 also, especially it's style, but If you look at the head to head, that is, Komodo 9.1 vs Houdini 4 and Stockfish 6 vs Houdini 4. Houdini 4 got destroyed by those 2 programs by huge margins.


Stockfish scored +43−5=66 against Houdini 4
Komodo 9.1 scored +25−3=31 against Houdini 4


This is also a good indication that improvements in software are in fact more important than hardware improvements.

Yes, it did not do well but what do you expect from a program that has not been improved in 19 months? In those 19 months Komodo & SF have been worked on consistently and being continually refined so this should be expected.

True, but I wouldn't think that Houdini 4 would have been beat by such a wide margin from the top 2. Certainly 77 elo points doesn't translate into that much of a margin of victory for Stockfish and Komodo.