# of Games in Databases: Chess vs Go

[tk2000]

In Chess one can purchase a games database with ~5 million games (e.g. ChessBase Mega Database 2011), whereas in Go
a database contains around ~70 thousand.

What are the main factors that account for this orders of magnitude difference?
Just curious ... does anyone know what the number of games in a typical Shogi database is?

[p2501]

I wondered about the same thing when I listened to:
http://www.radiolab.org/2011/aug/23/

I figured, that the chess database includes much more player levels for chess than the ones for go - the ones I know for go 'only' have 20-25k games in it, but they are all pro games. Also Chess has usually much less moves/game than go.

[Suji]

I think one reason is that there are significantly more chess players than go players. In chess, it seems that no matter what, one tournament is always going somewhere in the world. Whereas, go tournaments seem to be few and far between.

[moonrabbit]

There are a bunch of reasons.

The typical chess database includes master-level games, not just GM games. If you count only pro-level chess games (GM and strong IM) the number of games drops dramatically. For example, http://chesstempo.com has a free database with 2 million games, but if you restrict to games where the players have 2500+ ratings, there are a couple of hundred thousand games.

All tournament chess games, even amateur games, are recorded by the players. On the other hand, my understanding is that many professional go games in the past weren't officially recorded.

Professional go tournaments are usually elimination, while high level chess tournaments are usually round-robin (and therefore have more games per player.)

The chess database enterprise has a lot more manpower behind it than go databases (the best go database, GoGoD, is the work of two -- TMark and John F.)

[hyperpape]

I wonder what search and space requirements would be like for a million go games. A Gb?

It probably helps database analysis that chess openings have a relatively simple tree structure compared to go openings.

[tchan001]

Take a look at Bigo Assistant Full for the biggest go database for purchase
http://bigo.baduk.org/

[John Fairbairn]

Several reasons exist and many are mentioned above, but one not mentioned so far has been stressed to me by a chess master. It is that knowledge of the latest openings is a vital tool for chess players, which leads to a culture where they expect to share their games with each other. Go pros don't seem at all interested in databases, on the whole, and even less interested in sharing their games. This may be influenced by the fact that, in Japan at least, sponsors, being most often newspapers, expect to have first publication rights.

[Suji]

Several reasons exist and many are mentioned above, but one not mentioned so far has been stressed to me by a chess master. It is that knowledge of the latest openings is a vital tool for chess players, which leads to a culture where they expect to share their games with each other. Go pros don't seem at all interested in databases, on the whole, and even less interested in sharing their games. This may be influenced by the fact that, in Japan at least, sponsors, being most often newspapers, expect to have first publication rights.

So this is why Go openings are so freaking hard to find and study....