Benson's algorithm (Go)

This section does not cite any sources. Please help improve this section by adding citations to reliable sources. Unsourced material may be challenged and removed. (January 2023) (Learn how and when to remove this message)

Most strong Computer Go programs since 2008 do not actually use Benson's algorithm. "Knowledge-based" approaches to Go that attempt to simulate human strategy proved to not be very effective, and later approaches generally used tools such as Monte Carlo random playouts to "score" positions.^[3] Go positions frequently require scoring stones and territory in a more probabilistic, gradual manner where stones might be probably dead unless the opponent allows the player to make uncontested plays to save the stones, contested, alive as long as the opponent is not allowed one uncontested play, alive as long as the opponent is not allowed repeated uncontested play in the area, and so on. A system that only perceives unconditionally alive will not be very strong, as high-level play routinely involves leaving groups not entirely "finished" after their state becomes safe if protected by further plays (e.g. they will only be captured if the player lets them be captured, in which case they are presumably trading them for an even higher value objective). A system that can handle more complicated gradients of possibility will already understand unconditionally alive stones "for free" as part of whatever system is used to score and understand positions.

• Go and mathematics
• Computer Go
• Go strategy and tactics
• God's algorithm

• David B. Benson (1976). "Life in the game of Go" (PDF). Information Sciences. 10 (2). Elsevier: 17–29. doi:10.1016/s0020-0255(76)90554-5. Retrieved March 21, 2012.