It's amusing that the term "fish," which is frequently used to refer to weak chessplayers at least in American English ("bunny" is also good in the UK), has become a popular naming theme for top chess programs. Rybka means "little fish" in Czech and Polish, but the engine by Vasik Rajlich has been the big dog in computer chess for a long time now. (Five years being a very long time for dominance in just about any tech field, barring monopolistic practices.) For a decade, since the obsolescence of DOS programs, the programs Fritz, Junior, and Shredder mostly took turns as primus inter pares. Since 2005, with a few blips, Rybka as been more like Optimus Prime.
Lately, however, the computer chess message boards where a few dozen hardcore multicore ultra-geeks generate enough posts daily to utilize the bandwidth of a midsize European country, have been abuzz with fresh names and new competition. And, of course, flying accusations. (Okay, maybe there are more than a few dozen of them, but sometimes it does feel that way. I'm not sure of Permanent Brain is still holding court in the Ninja boards, but his posts on these things are always informative.) Ever heard of IPPOLIT? Robbolito? Firebird? This is a series of engines, all very strong and apparently with a common source. The controversy is over whether or not they are clones of Rybka based on reversed engineering Rajlich's code, which is what he seems to be saying. This is not rare, but it can be hard to prove. More to the point, proving it can force you to reveal things about your code you're rather not reveal. I see ChessVibes covered this a few weeks ago.
How about Stockfish? This one seems to be free of the accusations because it's an open source program that freely admits its origins in the strong and popular open source engine Glaurung. Whatever the Stockfish folks did, it appears to have put the program into the strength category of Rybka, at least in head-to-head play.
(Which, not to get too deep into the weeds, can be close to irrelevant for the purposes you may have for your chess engine, such as analyzing GM games or your own. That is, if hyper-tactical program A scores 53% against the more positionally adept B in blitz, it's still likely a human is getting a lot more useful information from B. But if you just want to quickly blunder-check some games or your analysis before posting it your blog, the quick tactician might be better for you. They are all so strong it's ridiculous to worry about which one beats the other unless you are specifically interested in which one beats the other.)
Most importantly, Stockfish is free. It's a UCI engine that requires an interface. I believe there are licensing issues that prevent open source programs from using the Nalimov tablebases, which is a shame for those of us who mostly use engines for GM game analysis. There's a long and informative interview with the Stockfish team here. This is just the latest episode in free programs pushing the top commercial engines on the rating lists. Fruit and Toga did it a few years ago before Rybka came in and shoved everything else out of the limelight.
This is why -- along the the more-true-every-year parenthetical above -- I have always said interface and features are what matter, and why ChessBase doesn't lose any sleep about the newest programs to climb the rating lists. Ever since engines got to the super-GM level my interest has been about how they could be made more useful, not just stronger. For the engine this means playing and evaluating in a way that can help a human improve, and Rybka was a step in this direction. For the interface it means features that do that, as well as providing entertainment value and a better overall user experience. ChessBase was teased in the aforementioned ultra-geek community when they basically stated this and stopped worrying about their flagship engine Fritz winning computer chess events, or even playing in them. Maybe there were a few sour grapes in there, and ChessBase still caters to different audiences with Rybka and others. But when there are free engines that can demolish any human on a laptop it's time to stop working so hard on the motor and focus on the stereo system and comfier seats. Few care if you can drive 200 mph instead of 170 and even fewer should care.
Chess professionals might, but they and other people like me who rely on chess tools constantly are a tiny part of the market. And I admit that even I haven't bothered to upgrade my own software in quite a while because I stopped getting free stuff and I didn't see any must-have features that would make my work any easier. (I feel much the same about about Adobe CS4.) That isn't to say they haven't added a lot of things that are great for people with other needs, of course.
Btw, has anyone implemented by brilliant idea for a visualization training mode yet? I suggested this probably five years ago, to have a mode in which the position on the board lags behind the actual game position by a variable number of moves. It's basically a handicap blindfold game. You have to visualize x number of moves from the notation all the time. You could keep increasing the gap until you were playing the entire game blindfold. You could even have different handicaps for different players online as a variant. A 2000 could play online with a six move visualization handicap against a 1600 with a three move handicap. It would be easy to implement and it would be fantastic exercise. Move entry could be on a blank separate or overlay board.