Taisha Database Search
Below are the results from searching the 17,000+ games in the July 2002 version of GoGoD CD for the Taisha. Note that this search is based on an otherwise empty corner extending one line beyond the side hoshi. The taisha has other uses as a reply to various pincers against a small knight's approach to the 3-4 stone. These are eliminated from this search.
Search results
- The narabi at a
- The tobitsuke jump to b.
- The extension along the right side (c).
- The kosumitsuke at d.
The four continuations shown here account for 487 of the 499 cases of the taisha returned in the search (see below). In all four major lines White has a winning record. Ishida considers White a, c, and d all to be compromise moves "...so slight disadvantages cannot be avoided"(Ishida's Joseki Dictionary, vol. 2). However, I would say that the actual record does not bear out this assessment. It is worth noting that Ishida is a fan of the taisha as Black. He has written at least one book exclusively on the taisha in Japanese (which I, as another fan, have read :-) So his opinion may not be entirely unbiased!
Statistics:
499 matches (185/314), B: 38.3%, W: 56.9% freq B wins - W wins Wb: 323 (321), B38.4% - W57.3% Wd: 77 (77), B32.5% - W61.0% Wc: 59 (57), B44.1% - W52.5% Wa: 28 (28), B39.3% - W53.6%
The earliest example of the taisha in GoGoD CD is from an 1822 game between Inoue Ansetsu and Shinomiya Yonezo. In the 19th century it appeared in about 10% of all games listed in this collection. Its popularity waned in the 20th century, especially from the 1930s onward. This coincides with the rise in the use of hoshi as the first move in the corner which naturally reduces the opportunity for joseki such as the taisha. At the same time the overall poor results for the player initiating the taisha must be another factor in its decline. As interesting as it may be to Go theorists, the taisha does not have anywhere near a winning record. Its limited use by professionals should come as no surprise.
Hits per database
0196-1699: 0 games (of 433) 1700-99: 0 games (of 287) 1800-49: 32 games (of 344) 1850-99: 92 games (of 1007) 1900-09: 18 games (of 252) 1910-19: 15 games (of 234) 1920-29: 31 games (of 408) 1930-39: 46 games (of 999) 1940-49: 19 games (of 477) 1950-59: 19 games (of 684) 1960-69: 35 games (of 962) 1970-75: 16 games (of 727) 1976-79: 17 games (of 783) 1980: 13 games (of 321) 1981: 15 games (of 358) 1982: 13 games (of 400) 1983: 11 games (of 333) 1984: 3 games (of 177) 1985: 9 games (of 214) 1986: 4 games (of 232) 1987: 4 games (of 237) 1988: 15 games (of 324) 1989: 15 games (of 451) 1990: 9 games (of 499) 1991: 8 games (of 541) 1992: 4 games (of 550) 1993: 3 games (of 317) 1994: 1 games (of 308) 1995: 0 games (of 281) 1996: 1 games (of 385) 1997: 2 games (of 575) 1998: 3 games (of 346) 1999: 2 games (of 411) 2000: 9 games (of 795) 2001: 5 games (of 837) 2002: 4 games (of 526)
mgoetze: I'm sure the names I've chosen for each of the continuations is a horrible slaughter of the Japanese language and I invite anyone to correct them.
For some of the reasoning behind the main variations see Extension of Crosscut Tactics