On 2/19/07, Winter <[EMAIL PROTECTED]> wrote:
> With version 0.94 (yes, I know 0.95 is out), winning percentage is
> shown alongside valuation, which raises a question.
>
> When simming a game to see how I did, what should I do when the play
> with the highest win % does not have the highest valuation?

Years ago I used to have rules for myself for stuff like this. I
rarely fully analyze my games anymore, and I definitely don't keep
records of my error rates. There's no official way to analyze a game,
and I'm reluctant to endorse this kind of thing. Even if there were a
very perfect way to reduce how well or poorly a person played in a
game to a number, doing so would accomplish very little. It can be
interesting, but so is figuring out the highest scoring 4x4x3x3 in
Super Scrabble. And like that, it's ultimately useless for becoming a
top player. There are many ways to improve without knowing exactly how
well you're playing. I'd rather see the interesting positions than the
final stats.

I don't want to be too snobby about this, and people are welcome to
discuss this if they like. I'll just abstain. Jason's view isn't far
from mine, so it is unlikely that an answer to your question will ever
appear in Quackle documentation.

I like to think of a Scrabble game being like a hand of poker. If a
poker player wants to know if he's a winning player, he probably
wouldn't do it by analyzing his decisions on every hand. Some hands
and some games are more interesting than others, and it's worthwhile
to discuss and sim those, but it's an inexact science. If you really
want to know how well you're playing, I'd look at the results after a
statistically significant sample of games. If you play at the same
Scrabble club every week and you have a better record and better
average margin now than you did last year, that probably means
something, and the great thing is that it accounts for all of the
things that sims can't.

If Quackle's report function becomes more sophisticated, I'd rather
have it classify the types of errors a player makes. Graph the lengths
and playabilities of the words they miss, etc. More like the teacher's
comments on a report card than the letter grade itself.

John O'Laughlin

Reply via email to