Barnes, Craig

aus Wiki, der freien Schachcomputer-Wissensdatenbank
Zur Navigation springen Zur Suche springen

Craig Barnes ist US und FIDE Meister (Elo 2280) im Schach und neben Kaplan und Levy einer der wenigen Schachprogrammierer, die das Spiel selber beherrschen. Anfang der 70er kam er mit der Schachprogrammierung an der University of California, Berkeley in Kontakt und entwickelte dort mit anderen das Programm "Ches", welches '73 an der Amerikanischen Computerschach Meisterschaft (ACM) teilnahm.

1980 wurde er von Julio Kaplan ins Team von Teletape Productions berufen und entwickelte dort das Programm für den Mattel Computer Chess.

Von 1982 bis 1995 arbeitet er als Senior Programmer bei Julio Kaplans Heuristic Software. War zunächst noch Kaplan federführend für die Entwicklung verantwortlich (Sensor Chess bis Superstar), erreichte Barnes ab dem Programm Turbostar den Status des gleichwertigen Co-Autors und wurde bei der WMCCC 1985 in Amsterdam für drei Variationen von Turbostar als Autor genannt. Es kann davon ausgegangen werden, dass Barnes bei allen weiteren Programmen für Saitek mindestens als Co-Autor an Bord war, einfache Geräte wurden i. A. von ihm allein entwickelt. Er selbst gibt an wohl die meisten Schachcomputer am Markt entwickelt zu haben. Er selber bedauerte es, dass er und Kaplan leider keine Aufträge für wirklich starke Geräte bekommen haben, aber mit der Mittelklasse ließ sich wohl am meisten Geld verdienen.

Nach '95 war er bis 2001 weiter als selbstständiger Entwickler für Saitek tätig, die Geräte dieser Epoche können ihm größtenteils zugeordnet werden.

Nachfolgend ein Auszug aus einer längeren Konversation im Hiarcs-Forum, der das Schaffen von Craig Barnes näher beleuchtet:

"I'm Craig Barnes. I worked with Julio Kaplan for many years developing chess computers. He recruited me in 1980 as part of a small company called Teletape Productions to develop a chess game for Mattel. After that project I ran off to a brief juggling career but then joined Julio in 1982 as he was starting his own company, Heuristic Software Corporation. We had a few other clients initially but ultimately wound up doing all our work for Saitek (formerly SciSys). I did the bulk of the programming although Julio also did a lot and from time to time we added other people. In 1995 the Saitek work dried up and HSC closed down. However I shortly wound up continuing to do computer chess projects for Saitek into 2001. In the later years there was a bit of melding -- Saitek acquired Mephisto (and perhaps others computer chess lines?) and in at least one instance bought code from others (Franz Morsch) and incorporated it into a Saitek product. I'm given to understand that more models and units were primarily written by me than be any other chess programmer.

For those who don't know, Julio Kaplan was an International Master and the World Junior Chess Champion in 1967. As for myself, I was a US and FIDE master, and was US National High School Champion (1972). We must have been the two strongest chessplayers involved in commercial chess games, except for British IM David Levy in the early days. Although good chess ability isn't a prerequisite to be a chess programmer (most programs even back then often became stronger than their creators), it certainly doesn't hurt when it comes to incorporating strategical conepts or recognizing bad moves. However, Julio and I rarely worked on "high-end" games, since it was much more lucrative to make the best models in the lower- and mid-range niches.

Saitek's games were sold worldwide under their own name (sometimes with Kasparov's name and endorsement) and also under Tandy (Radio Shack), although not all of Radio Shack's games were ours. I also frequently saw our games in the Sharper Image catalog and elsewhere. Over the years we made programs of all shapes and sizes, from small travel sets, to nice wooden boards with communications and plug-in modules, and with all sorts of features (voice, lessons, master chessgame databases, etc). And challenges -- we made one product that ran in only 2k ROM and 128 NIBBLES of RAM. Aaah, memories, memories...


I did so many games over the years that I haven't kept track of them all, but 20 years times a couple per year = a lot. Generally we only dealt with prototypes with exposed wires and connectors, and rarely (it seems) actually got a final product or even knew what the packaged name was (versus our internal project name). For a couple of projects I wrote software that completely emulated the target microprocessor as well as showing me the program's chess logic. ... We also did a PC game: "Kasparaov's Gambit" for Electronic Arts (1993). This actually had someone's else's chess engine, but we ported the graphics and various features.


Hmmm, ...The Mephisto shachakademie (also sold as Kasparov(Saitek) Chess Academy) had Franz Morsch's 16k or 32k chess engine, which Saitek bought the assembler source code for. I then took the code (which was commented very lightly and in Dutch) and modified the I/O and other features. We then added an extra ROM (2mb?) which contained 100 tutorials, each roughly equivalent to a long magazine article, ranging from basic how-the-pieces-move to analysis of openings and master games. The tutorials were spoken and included quizzes and other interactive features. The lessons were written by a French chess journalist and then translated into German and English; I'm not sure a French product was ever released. Combine that with engineers in Hong Kong and marketing in the US and Europe, and we had a truly international product! Speaking of internalization, some of our other games (with smaller vocabulary) spoke in multiple different languages: English, Dutch, German, French, Italian, Spanish, and Swedish, I believe. In another product, we had an 8mb ROM which contained a database of all the master chess games ever played. We used this as a "book" -- we'd play the move from the game if that side eventually won, and try something else if the player ultimately lost. But generally most products ranged from 2 to 16 or 32k ROM , 128 nibbles to 4k RAM, and speed from 100khz up to about 10 or 20Mhz. Needless to say, I got pretty resourceful at fitting complex code into small environments...


My last computer chess project was in 2001, and I haven't been following computer chess developments since then. I would expect that by now, since just about everyone has a powerful PC or iPhone or some other device which might run chess software, that any new dedicated machine would quickly become out-of-date technologically. Plus there's so much more one can do with a PC -- saving and printing games, networking, etc. Of course, since few humans play chess seriously, even the old chess machines could beat most humans at the lowest normal levels (we often implemented "handicap" levels to give novice humans a chance). So machines are still probably "good enough" opponents for all but the most serious players.


Yep, I remember the Blitz dials, although I didn't recall that it had that auto-sensor technology. By the way, at one time we worked on a technology using coils in each piecetype so we could always tell exactly what piece was where -- you could set up positions/newgames just by placing the pieces. Unfortunately I think there was problems with radio interference."


mit Infos für den deutschen Wiki-Eintrag: