I started testing a new engine, and I am very enthousiastic about it: it's not very famous and really deserves to be. It's called Robbolito, and is available on http://www.chesslogik.com. I know what some of you are already thinking: Robbolito ? but this is a Rybka clone and was forbidden on most forums etc. Please read the following and think for yourselves
As mentionned in the disclaimer for new joiners, I can guaranty that I am no stranger to the computer chess community: I have written the program BibiChess 0.5 which (to my surprise) appears on your rating list. BibiChess is quite weak (2117 ELO in CCRL), but attempting this project made me understand a lot of things about computer chess programming. My main source of inspiration was Fruit 2.1 at the time, even though BibiChess was written from scratch. I later had a look at other source codes such as Glaurung/Stockfish for example.
Anyway, all that to say that I know what I am talking about if I tell you that Robbolito is certainly not a clone of anything that exists, and surely not Rybka. Here are the main reasons (against non existant proofs of the contrary!):
* Rybka is not open source and anyone who is litterate in computer programming knows that reverse engineering compiled code is impossible. At most you'd get a linear code with unnamed variables and structures like if/goto everywhere, basically it would look no better than assembly code, and you cant do anything with that!
* Robbolito is open source: I have to say the code of robbolito is quite mystical and difficult for a simple brain like mine to understand, but just 2 remarks about it. First in the coding style, it looks like an extra-terrestrial wrote it, with *.c files including themselves recursively etc. Secondly it has little chess knowledge, its skill is based on deep search, it all depends on the position but it's not rare to see Robbo at depth 18 when Rybka is stuck at 10! This isn't a point of detail and has lots and lots of implications in the search algorithm: pruning, extensions, etc.
* Another thing is multi cpu: Robbo doesn't support it and Rybka does. It makes a big difference in the code, you can have a look at StockFish's open source code to understand how a chess search algorithm can use optimally multithreading. Believe me it's not just a trivial fix and show once again that there could not be any resemblence between both source codes.
* Another thing again, the authors from Robbo originally wrote from scratch a linux program that creates EGTB 3,4,5,6 - so called Robobases and open source again. Rybka just uses the code and date from the famous proffessor Nalimov from Novosibirsk University. Most programs use Nalimov or Daniel Shawul bitbases (which are better than Nalimov because the data is compressed and faster access and better storage in memory). Stefen Mayer Kahlen wrote his own bitbases too, but closed source code (before Daniel Shawul).
* Now for those who dont know much about comp chess programming, and will want to look at the game to make an opinion. Please have a look at the attached quick tournament between Rybka 3 (w32 2cpu), StockFish 1.5.1 (w32 2cpu), and Robbolito 0085e3 (w32 1cpu). My hardware isnt very powerful but it's the same for all 3 engines competing each other on my labtop, so it doesnt matter - actually Robbo uses only one core out of 2. Besides proving that Robbo is the strongest, it shows a huge difference in playing style. In particular have a look at game 16 where Robbolito refutes Rybka's sacrificial attack like no other chess engine could dream of doing. And look how they never agree on scores and continuations! Basically it looks like Kasparov (as Rybka) playing against Deep Blue (as Robbolito). Fot that matter I personnaly prefer the playing style of Rybka, but still Robbo just scores more points at the end of the tourney, somehow.
Robbo was banned by all these commercial sites that i wont name, and who have an interest in preventing the world from knowing the truth. Obviously if everyone knew the existence of Robbo and Rybka, then no one would buy Fritz ? Those who are ready to pay and want a nice inteface with all the stuff (chess database) would go for Rybka, and those who want just a strong computer that is free would download Robbolito. I have also read on the Rybka forum that the definition of a Rybka clone is "any program that is stronger than rybka" (so much stupidity just infuriates me but well, it's best to just smile and forget about it).
At some point the truth will triumph, and it's to the honour of CCRL if it starts from here. Besides, for all the work of its genious authors who donated it to the world *free of charge*, Robbolito deserves to stand on the CCRL list for everyone to know that it exists and what it's worth.
I hope enough people will read and understand this, and try to make a non-influenced opinion of their own. And when eventually enough of you agree we can start testing Robbolito
PS: Some people owe the community public excuses for their unfounded accusations. And particularly Vasik for Strelka (which was fruit based and i know the fruit source code like my pocket to i wont even argument it), and for Robbolito of course.
