|
|
> From time to time in my reading, I come across the phrase "technical
> play", but I have never seen an explanation of what it means. How does
> technical play differ from non-technical play? Are there any expository
> works that treat the subject? Have the bots learned technical plays?
A reasonable working definition of a technical play is a play in a position
of which the general theme (game plan) is known, so finding the best move
is more heavily dependent on calculation of shots, clearing points most
safely etc., which best follows that known theme. These 'technical' plays
are distinct from their counterparts - 'positional plays', where the
fundamental aspect of finding the best play is recognising what theme (game
plan) should be followed.
Ultimately, the two basic types of play (technical & positional) are
intertwined with each other at various levels, but it is convenient for bg
literature to seperate the two for clarity/simplicity.
As far as bots are concerned, the neural net method used nowadays doesn't
really disciminate between the two - they just gradually increase there
technical & positional 'skill' co-dependently.
Bill Robertie has written some books treating these two subjects
seperately, 'Technical Play' and 'Positional Play'.
Adam
|
|
|
Douglas Zare writes:
That's a good definition. Relatively pure examples of technical plays occur
when there is no contact, or when you are bearing off with contact.
I wouldn't say that calculation is so important, though, for example, it is
very useful to take an exact pip count for most racing cube decisions.
However, if you are bearing off, with your opponent on the bar, the quality
of your opponent's shots is important and very hard to calculate, and the
gammons and backgammons you win are important and hard to calculate. Still,
technical plays have a lot to do with tactics.
If technical plays tend to be about small amounts of equity, who cares?
The problem is that they are very common. If you make a small error every
game, that can be worse than making a large error every time a very unusual
class of positions arises. Also, the errors are not as small as one might
think.
Douglas Zare
|
|
|
|