🚀 go-pugleaf

Concerning the initial poster's idea of describing the complexity of a
game according to how difficult it is for one human player to describe
to another,

At present this is not an accurate measure of the computational cost of
the game (more or less, how long it would take a computer to decide on a
move).

Consider the way computers and humans play chess:

A computer does it by evaluating all the possible moves for the next few
turns.

A human cannot possibly beat a computer at this, since the computer can
obviously calculate far faster than the human. Instead, a good human
chess player will form a high-level view of the game, using "chunks" of
information about the strategic position. (To demonstrate this, it has
been shown that a good human chess player, given a brief look at chess
board and asked to reconstruct the position, will reconstruct a
strategically equivalent but not identical position.)

Therefore, the human player has to evaluate far fewer positions than the
computer; this is why the world chess champion is still a human.

So, while a game might be reasonably simple from the point of view of a
good human player, because the human player can grasp the position in
terms of high-level strategic concepts, the game is probably much more
complex from the point of view of the computer which does not understand
such concepts.

For example, war games are usually simpler from the point of view of a
human player because they are strategically simpler. However, because
there are more possible moves, it is much more difficult for a computer
to play one well by looking at the possible moves.

Of course, hopefully at some stage a computer will be able to think in
terms of high-level strategic concepts about a game; but for the moment,
the complexity of a game from a human's viewpoint is not a good guide to
its complexity from a computer's viewpoint.

--
"To summarize the summary of the summary: people are a problem"
Russell Wallace, Trinity College, Dublin
rwallace@unix1.tcd.ie