professional choice - game theory comment
Charles Matthews Just an observation to try to clarify at a general level some of the statistical analyses of shapes that have gone on here, giving rise to justified scepticism about some of the proverbs or conventional wisdom about, for example, the cross-cut, the staircase.
Pro-level play very clearly differs from what amateurs do in the way that if there's a standard, known way to play in a situation and it works in a given context, a pro is hardly likely to miss it. That is, if it is common knowledge for the players that Plan A can be answered by Plan B, we aren't going to see Plan A in high-level play except in cases where Plan B really isn't damaging as an answer.
It is certainly one of the interesting features of database search that we can start to infer what those are: they are (phenomenological) rules which have a different status from proverbs as heuristics (don't make an empty triangle, except of course when it is correct to do so).
Joseki choice might be a case in point. Since some joseki branch rapidly, it may seem to be effectively impossible to know what will happen. The way pro opening theory as a whole seems to operate is more like 'don't give your opponent simple, ideal plays'. There are always simple-looking plays as answers, but they should either be less than ideal, or complex in a strategic sense.
There is a parallel in chess. The objective in chess is to mate the opponent's king. There are hundreds of books that teach how to mate. There thousands of books of mating problems for practice. And yet mates or mating attacks are virtually never seen in the tournament games of grandmasters. They are anticipated and prevented long before they even reach the level of a tactical point for an annotator to comment upon.