Extraordinary performance of Strelka 5.

[MM]

http://www.talkchess.com/forum/viewtopi ... w=&start=0

Look down there on the second graph for Strelka 5. As the evaluation goes, every monotonic function is invertible.

Kai

One reason for the difference in evaluations is that Jury (purposely?) forgot to divide the output by 2 to obtain centipawns - Houdini uses 1/200 pawn units internally.

Seeing the results published on CEGT, I actually feel some sympathy for Norman's complaints about rating lists. If Strelka is readily adopted by the CEGT people, why continue to exclude Ivanhoe or Fire?

With Strelka we have an engine that is, by its author's admission, the direct result of reverse engineering Houdini 1.5. Unsurprisingly it produces identical results in move choices (see the similarity diagram) and in Elo strength (see the CEGT results).

Personally I don't mind Houdini (in whatever form) occupying all the upper ranks of the rating lists, but I'm not sure it adds to the rating list's relevance .

Robert

Hi,
yes, i agree with '' If Strelka is readily adopted by the CEGT people, why continue to exclude Ivanhoe or Fire?''.

[Frank Quisinsky]

Hi Maurizio,

yes, I know!
But to see each time the same playing style, same strengths can be very boring. Speaking from myself in SWCR.

On the other hand ...
I can see that the style of Rybka, Fire, and IvanHoe is different. OK, same strengths in late middlegame, early endgame but the style is an other. Easy to see in game analyzes!

Now, if we look in all that what is today available ...

Examples:
Deep Saros, Leopard, Vitruvius and so one ... perhaps 5% own work? If a rating list add now all this engines the list is more and more sick. The question is how sick are the rating lists today are available, means how many of the programmers used Fruit ideas, other ideas for create all the improvements?

For around 6-8 years new engines comes with perhaps 30-50 ELO more if a new version is available. Today often to see that an engine is 100 ELO stronger. A brainstorm for so many programmers to the same time?

Optimal situation for people which search strong engines is:
To have perhaps 8 with a complete other style and own faces. That would be great for analyzes of the own games.

And if we look a bit deeper we can see that one of the best available engine in middlegames are Spark. 250 ELO weaker as the number 1. Means to have engines with own strengths is much more important as to have engines which do the same.

Best
Frank

[peter]

Personally I don't mind Houdini (in whatever form) occupying all the upper ranks of the rating lists, but I'm not sure it adds to the rating list's relevance .
Robert

Hi!
+1!
I guess, that's just what Vas might have thought the one and another year ago too about Rybka, maybe even Fabien could have had such ideas once in a while.

[Frank Quisinsky]

reverse engineering

Not important is that reverse engineering is done with Rybka or Houdini. Much more important is that this one can be done with each other engine too.

Collect all the ideas and create a new one ...

The question is how many secrets computer chess have today. I have here the same opinion Marco Costalba wrote in our interview. The secrets of computer chess are open and well known.

With the result:
Programmers which are able to develops a new idea, new code and are very pride of such a work will be the victim of the swindler.

If the next Komodo version will be on 1 the Komodo team have the same problem.

Best
Frank

[MM]

Hi Maurizio,

yes, I know!
But to see each time the same playing style, same strengths can be very boring. Speaking from myself in SWCR.

On the other hand ...
I can see that the style of Rybka, Fire, and IvanHoe is different. OK, same strengths in late middlegame, early endgame but the style is an other. Easy to see in game analyzes!

Now, if we look in all that what is today available ...

Examples:
Deep Saros, Leopard, Vitruvius and so one ... perhaps 5% own work? If a rating list add now all this engines the list is more and more sick. The question is how sick are the rating lists today are available, means how many of the programmers used Fruit ideas, other ideas for create all the improvements?

For around 6-8 years new engines comes with perhaps 30-50 ELO more if a new version is available. Today often to see that an engine is 100 ELO stronger. A brainstorm for so many programmers to the same time?

Optimal situation for people which search strong engines is:
To have perhaps 8 with a complete other style and own faces. That would be great for analyzes of the own games.

And if we look a bit deeper we can see that one of the best available engine in middlegames are Spark. 250 ELO weaker as the number 1. Means to have engines with own strengths is much more important as to have engines which do the same.

Best
Frank

Hi Frank,

again i agree. The last part of your post is the most interesting: what is more important? It depends by the user, if i study the opening without book or fischerandom i could choose Shredder, if i analyze middlegame i choose Houdini, if i analyse endgames i could choose Rybka 4, but they are examples.
But i could choose an engine just to have the fun to let it play against another and watch wonderful games.
Yes, if the style of play is too similar, games become boring and honestly, lately, many, too many engines games are boring

[Frank Quisinsky]

Hi Maurizio,

your latest sentence:
If I speaking from myself (like more attacks and good middlegames) ...
90% from the engine-engine matches are boring. Each one the same endgame slides. Means each one the same boring endgames. And about this endgames we discuss here. We discuss about endgame strength in computer chess because a lot of engines with perhaps 2.400 ELO playing a very interesting middle game.

I am thinking last year that perhaps the time control can be the reason. In SWCR champions league I used 40 in 150, 40 in 150 again = 15x more time as in SWCR rating list.

Honest:
95% from the engine-engine matches in champions-league are boring. 5% more ... more and more boring remis games.

Perhaps a grandmaster will like such things.
Not chess player with perhaps 2.000 ELO.

Only 14 of 44 engines I tested so far in SWCR I can say ... yes I like it. But this is my personal opinion, not more not less.

Best
Frank

[MM]

Hi Maurizio,

your latest sentence:
If I speaking from myself (like more attacks and good middlegames) ...
90% from the engine-engine matches are boring. Each one the same endgame slides. Means each one the same boring endgames. And about this endgames we discuss here. We discuss about endgame strength in computer chess because a lot of engines with perhaps 2.400 ELO playing a very interesting middle game.

I am thinking last year that perhaps the time control can be the reason. In SWCR champions league I used 40 in 150, 40 in 150 again = 15x more time as in SWCR rating list.

Honest:
95% from the engine-engine matches in champions-league are boring. 5% more ... more and more boring remis games.

Perhaps a grandmaster will like such things.
Not chess player with perhaps 2.000 ELO.

Only 14 of 44 engines I tested so far in SWCR I can say ... yes I like it. But this is my personal opinion, not more not less.

Best
Frank

Hi Frank,

I see computer chess as a science, i totally appreciate the Botvinnik idea of creating an engine playing with human thought, because machines can't plan from their own, humans can, so adding machines to human concepts it should be the best mix. I think Komodo's team has more or less this idea.

Anyway, for me, considering computer chess as a science (for humans it's different, it can be fight, art, fear ect.) i just search the engine that can give me the best chances to find the best moves. Style doesnt attract me.

I recently made a match between Strelka and Houdini 1.5a and games were very boring, but i think i consider them boring not because the style but because i have the suspect that their play was uncorrect!

Best regards

[Frank Quisinsky]

Hi Maurizio,

I think that today the programmers should try to give here engines an own face. This is much more important as playing strength alone. Easy, we have enough strong engines which do the same.

Example:
Tatwaffe Springer (german book).
ISBN: 978-3-940417-06-0

Knowledge with knights!
In self programming work.
Give me only one engine where you can say ... Frank this one have it. Sorry, only Arthur have it.

That what I meaning!
I think programmres forget to gave his programs "chess kownledge". This one can be much more interesting as only "speed, speed and search".

Yes, the Komodo team goes an interesting way. Komodo have an own face and plays other chess, like Karpow style in middlegame

Not the style I like OK, but an own style with success. On the other hand the latest Komodo version are stronger in tactics. I saw here much very good games too.

Not important ...
I think not only the name of engine should be different.

Best
Frank

[gerold]

Hi Maurizio,

yes, I know!
But to see each time the same playing style, same strengths can be very boring. Speaking from myself in SWCR.

On the other hand ...
I can see that the style of Rybka, Fire, and IvanHoe is different. OK, same strengths in late middlegame, early endgame but the style is an other. Easy to see in game analyzes!

Now, if we look in all that what is today available ...

Examples:
Deep Saros, Leopard, Vitruvius and so one ... perhaps 5% own work? If a rating list add now all this engines the list is more and more sick. The question is how sick are the rating lists today are available, means how many of the programmers used Fruit ideas, other ideas for create all the improvements?

For around 6-8 years new engines comes with perhaps 30-50 ELO more if a new version is available. Today often to see that an engine is 100 ELO stronger. A brainstorm for so many programmers to the same time?

Optimal situation for people which search strong engines is:
To have perhaps 8 with a complete other style and own faces. That would be great for analyzes of the own games.

And if we look a bit deeper we can see that one of the best available engine in middlegames are Spark. 250 ELO weaker as the number 1. Means to have engines with own strengths is much more important as to have engines which do the same.

Best
Frank

Hi Frank,

again i agree. The last part of your post is the most interesting: what is more important? It depends by the user, if i study the opening without book or fischerandom i could choose Shredder, if i analyze middlegame i choose Houdini, if i analyse endgames i could choose Rybka 4, but they are examples.
But i could choose an engine just to have the fun to let it play against another and watch wonderful games.
Yes, if the style of play is too similar, games become boring and honestly, lately, many, too many engines games are boring

Critter vs. Strelka. 5/3 TC. After 180 games many boring draw games. However some really nice exciting wild games also. At this point
both are equal. Never one ahead of the other by more than 14 elo.

Best,
Gerold.

[Rebel]

Hi, i watched many games against Houdini. Well, their play is very similar but evaluation is very different, i think some important changes have been made in Strelka 5.

Score is multiplied by 4 or something like that.

Put S5 in analysis mode from the start position, you get something like 0.68