And the Ippolit author is...

[Alexander Schmidt]

First contact from my side was during Patriot 1 Time. We all thought it is clean. Patriot 2 was a copy of Toga with some formulas to change score and depth. I think for depth 3 was added, and the score had a factor of 1.5. The rest was identical.

After the Patriot 2 desaster I took a closer look on Patriot 1, and there where similaries to Crafty. It was based on Crafty but with many changes. But the frame was Crafty.

After Patriot 2 he released a new version of Kaissa, and it looked like Patriot 1, and I don't think he removed the Crafty stuff. Later it supported UCI, and a the traces of Crafty vanished, but still there where a few of them visible, like move generation.

[Alexander Schmidt]

Running a match between Ippolit and Rybka. Compared to other results I have seen with really long timecontrol: 2min/game + 1sec/move. The match is running under Fritz 8, Rybka with Rybka book, Ippolit with Fritz 8 book.

Rybka contempt set to 0 as advised by some here.

Results so far:

20 games, Ippolit won 6, Rybka won only 1, 13 draws. That means according to Fritz +88 ELO.

So it looks like that Ippolit wins also vs. Rybka 3 with contempt 0

One interesting observation:

The absolut score is mostly similar between both. If there is a change in the score, Ippolit can see it 2-3 moves before Rybka. Similar evaluation and better search in Ippolit?

I had also some time losses under Arena. No hang, just loses on time. Maybe, and now I indeed start speculating, maybe Yelin implemented a special Arena feature because of the Patriot 2 desaster (Patriot 2 should have been sold via Arena, but turned out to be a Fruit clone). There is no reason why an engine should stop playing under Arena and not under Fritz. After the game has started the protocol is handled in the same way by both GUI's.

Vas confirmed the engine is a clone of Rybka

The Rybka guys don't like the idea of an engine stronger than Rybka 3, which is NOT based on Rybka. This statement is done without any proof.

It is much more likely that Ippolit is a patchwork of several engines, based on Kaissa. It is even possible only ideas where taken from other engines, as most other authors do it.

[zamar]

With latest Robbolito version it says:

id author Yakov Petrovich Golyadkin, Igor Igorovoich Igoronov, Roberto Pescatore

Are these true names or is this only joke?

[Alexander Schmidt]

I guess it is all done by one, but I can only look at the engine and it's roots.

[Graham Banks]

Vas confirmed the engine is a clone of Rybka

The Rybka guys don't like the idea of an engine stronger than Rybka 3, which is NOT based on Rybka. This statement is done without any proof.

It is much more likely that Ippolit is a patchwork of several engines, based on Kaissa. It is even possible only ideas where taken from other engines, as most other authors do it.

Vas posted the following:

Reverse engineering isn't the issue of course - I am sure Rybka 3 has been decompiled and disassembled many times, as has for example ChessBase's .ctg format and so on. The issue is distributing stolen code - if this were allowed, nobody would ever have to pay for any software.

Vas

[zamar]

The Rybka guys don't like the idea of an engine stronger than Rybka 3, which is NOT based on Rybka.

It is much more likely that Ippolit is a patchwork of several engines, it is even possible only ideas where taken from other engines. It is probably based on Kaissa.

The FAQ says it directly:

Q. IPPOLIT pursues previous ships?
A. True, with KAISSA, Crafty, Fruit/Toga/Rybka/Strelka, etc. Yet beyond solely saponification from predescendants.

Now the very interesting question is: how much you can take code from other programs, modify it and then call it yours? This more or less what every chess programmer do in one from or another (or did all of you invent AlphaBeta yourself?). Even Vasik Rajlich did this.

For sure they have decompiled Rybka, looked how it works, then wrote very similar eval/search and even improved it! But the fact that these guys are afraid to publish their names makes me think they are _likely_ doing sth ilegal...

[Laskos]

With latest Robbolito version it says:

id author Yakov Petrovich Golyadkin, Igor Igorovoich Igoronov, Roberto Pescatore

Are these true names or is this only joke?

Y. P. Golyadkin is a personage in Fyodor Dostoevsky's "The Double: A Petersburg Poem". R. Pescatore means R. Fisherman. I. I. Igoronov seem variations on Igor (name, father's name, family name).

Probably all are jokes.

Kai

[Laskos]

Never mind, someone posted this information earlier.

Kai

[Michel]

Reverse engineering isn't the issue of course - I am sure Rybka 3 has been decompiled and disassembled many times, as has for example ChessBase's .ctg format and so on. The issue is distributing stolen code - if this were allowed, nobody would ever have to pay for any software.

Vas

Wow. Does he actually mean that the people behind ippolit somehow got hold of the original source code of Rybka 3 (or 4 even)?

It would be nice if Vas were less ambiguous in his statements.

[Leto]

Running a match between Ippolit and Rybka. Compared to other results I have seen with really long timecontrol: 2min/game + 1sec/move. The match is running under Fritz 8, Rybka with Rybka book, Ippolit with Fritz 8 book.

Rybka contempt set to 0 as advised by some here.

Results so far:

20 games, Ippolit won 6, Rybka won only 1, 13 draws. That means according to Fritz +88 ELO.

So it looks like that Ippolit wins also vs. Rybka 3 with contempt 0

One interesting observation:

The absolut score is mostly similar between both. If there is a change in the score, Ippolit can see it 2-3 moves before Rybka. Similar evaluation and better search in Ippolit?

I had also some time losses under Arena. No hang, just loses on time. Maybe, and now I indeed start speculating, maybe Yelin implemented a special Arena feature because of the Patriot 2 desaster (Patriot 2 should have been sold via Arena, but turned out to be a Fruit clone). There is no reason why an engine should stop playing under Arena and not under Fritz. After the game has started the protocol is handled in the same way by both GUI's.

Vas confirmed the engine is a clone of Rybka

The Rybka guys don't like the idea of an engine stronger than Rybka 3, which is NOT based on Rybka. This statement is done without any proof.

It is much more likely that Ippolit is a patchwork of several engines, based on Kaissa. It is even possible only ideas where taken from other engines, as most other authors do it.

I've run a 40 game match between Rybka 3 (with contempt set to 0) and Ippolit at 5 minute blitz and Ippolit only won by a two game margin. I also previously run a 100 game match with Rybka 3 set to default settings and Ippolit also only won by a margin of two games.

I don't think we yet have evidence that Ippolit is stronger. Or perhaps this is due to the different versions and compiles of Ippolit...