[Computer-go] Definition of single-point eye

Robert Jasiek jasiek at snafu.de
Fri Oct 24 14:16:05 PDT 2014

On 24.10.2014 22:21, Peter Drake wrote:
> The notion of one eye is useful. A group with one eye has an
> advantage in a fight against a group with none.

For this purpose, one can use a related term, which I call semeai-eye. 
It is not necessarily a life-and-death kind of eye as derived from 
two-eye-formation, independent life or uncapturable. As defined in 
Capturing Races 1, p. 26,

"For semeais, an eye is a region so that
- it consists of connected intersections,
- it is surrounded by one player's stones of essential strings,
- it does not contain any stone of an essential string,
- if the player and his opponent play only within the region and do not 
remove any surrounding stone, the player moving first can nessarily fill 
all but one of the intersections and
- if the player and his opponent play only within the region, the player 
moving second cannot necessarily permanently partition it into at least 
two regions."

Yet other terms with names related to "eye" are LD-eyes applied during 
the middle game, Landman CGT values and nakade shapes. Each of the five 
"eye" concepts describes something different. Just calling all those 
different things "eye" creates confusion. Greater naming precision is 

robert jasiek

More information about the Computer-go mailing list