Tim wrote:

MK wrote:

compare one ratio with another ratio.
> There's a more direct way to investigate the
> question as I've formulated it.

I don't know how to understand the word "direct"
in this context. Would you care to explain?

> Fix some number of games, say 10, and define a
> "session" to be 10 games ..... Play a session
> ..... Record who wins (meaning just who comes
> out ahead at the end of the session; ignore
> the margin of victory) ..... play 1000 sessions,
> and record which bot wins. Then ask the question,
> does the weaker bot win a greater fraction of
> the 1000 sessions when the cube is in play?

Can you offer a reason for why this will be more
telling that playing sessions of 10,000 games?

The purpose of the cube is to expedite gambling.
Thus the "margin of victory" is essential. What
you are doing is reducing a "money game session"
to a "sort of match play", which restricts the
ability (created by the cube) to capitalize on
fluctuations of luck, in trying to win as many
points as possible.

It was "intuitively" (and falsely) believed for
so many years that the players with more of the
so-called cube skill would win more.

Now that I debunked this "skill theory based on
jackoffski formulas", it seems to me that your
suggestion is just a contortion to at least
mitigate the defeat since you can't disprove
that the cube magnifies luck instead of skill.

This discussion is not strictly about GnuBG but
I think is appropriate in this forum as it may
help the bot developers in improving the luck
related calculations, etc.

MK

PS: I don't know how you are posting here but
unlike posts from other people, I don't receive
yours by email and wouldn't know about them if
I hadn't happened to visit the forum's web site.


Reply via email to