[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Bug-gnubg] Fuel for the fire

From: Joseph Heled
Subject: Re: [Bug-gnubg] Fuel for the fire
Date: Wed, 17 Sep 2003 09:41:21 +1200
User-agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4) Gecko/20030624

Might be a silly idea, but perhaps what we need is an equivalent to "confidence interval". I am not saying I know how to compute one, but if it said (say) "Casual to Advanced" we would hear less complaints from users.


Jim Segrave wrote:
On Tue 16 Sep 2003 (17:26 -0300), Albert Silver wrote:

I realize this is just fuel for the fire, but here's an example of what
I consider to be exaggerated strictness on the cube grading. I played a
5-point match that lasted 2 games it is true. I missed exactly ONE Cash
Point double for losing 0.041 equity and was judged a Casual Player (in
cube decisions). I think this is not right personally:

I think people have unreasonable expectations of gnubg's rating of
your play over a single match or a short session.
There are good arguments for only counting close or actual cube
decisions, the same as not including forced moves when calculating
chequer errors. In matches where your actual opportunities to make a
real decision are rare - for example when the dice are against you and
your *only* decision is a take/drop, then there is no way to make a
reasonable estimate of your skill or lack thereof with the cube.
It's on the same level as having the final examination in a course
consist of one question with a yes/no answer. You simply don't have
enough information to go on to make any realistic judgements at
all. The only thing that will work is to watch how your play goes over
a large number of matches. It's not nice having gnubg tell you your
cube play was Awful! because you made one very bad take in a seven
point match. It's a true comment on what your play was in that match,
it says nothing whatsover about your abilities except that you
sometimes make mistakes. If you are dropping 0.030 or so over 50 or
one hundred near or close decisions, then you need to study. If you
drop an average of .030 in one match, you don't really can't draw any
conclusions at all.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]