Re: Playing Advice to Michael Petch

On May 12, 10:24 pm, bob <bob_k...@[EMAIL PROTECTED]
> wrote:
> On May 12, 10:22 pm, EthicalBackgammon <psimb...@[EMAIL PROTECTED]
> wrote:
>
>
>
> > If you really want to determine which of two players is a better
> > backgammon player, the answer is to give them both an extremely
> > difficult test with checker and cube problems and see that they both
> > get the same test and see who answers better.  I know of no other way
> > that would be a conclusive competition based purely on skill.  Of
> > course, it can be argued that this test removes the ability to read
> > your opponent and adjust to your opponent over the board, and that is
> > a valid argument.  But the winner of this competition would prove to
> > me who was a better backgammon player assuming the player was playing
> > a bot or a top world-class player.  Again, looking at comparative
> > error rates from matches is somewhat of an indicator, but in the short
> > run, one of the players may have had a lot more difficult decisions
> > thrown at him than the other.
>
> > Phil Simborg- Hide quoted text -
>
>   Lots of problems here.
>
>   1) How many questions to ask in order to conclude one player is
> better suffers the same difficulty as asking how many games to play in
> order to conclude that the lower error rate is an accurate enough
> indicator. You say that the error rates might not be relaible since
> one player may have had the misfortune of just having had tougher
> plays. But realize that scores on your quiz would suffer from
> randomness also. On a close play the player might have no idea at all
> and just guess lucky. The questions might just have been closer to one
> player's area of expertise than the other player's.
>
>   2) Playing the game of backgammon is quite a bit different from the
> test you propose. You
> only want to use difficult questions but what percentage of actual
> backgammon plays are
> what you would call difficult? Not blundering on easy problems is just
> as im****tant as getting the tough plays correct and you don't test
> that at all.
>
>   3) What weight should be put on those problems? If one is a very
> unlikely situation to occur does it matter much if someone gets it
> wrong?
>
>   4) Quiz factor is a problem. When a difficult position comes up over
> the board it does not announce itself as such.
>
>   5) For extremely difficult problems do you want to put 100% trust in
> the bot?
>
> Bob Koca

All of your points are valid.  I don't believe there is any definitive
way to absolutely determine who is better.  I only suggest that a quiz
is a fairer test of skills and knowledge than playing matches.  You
are absolutely right that the person with the better ability in a quiz
may not be the best over the board, and trying to come up with a quiz
that reduces many of the problems you raise would be quite a
challenge.

Having said that, however, while I know there are skills over the
board that cannot be tested on paper, I believe, particularly at the
higher levels of play, that the player who can best analyze
complicated positions in a test is more likely to be better over the
board as well.

It has been my experience, for example, that when I show a complicated
position to a world class player, he is able to come up with
explanations and discuss nuances of the position that go far beyond
what a lesser player is able to do.  These same players who are able
to answer these tough ones correctly and discuss them in greater
detail just happen to win more tournaments, have higher rankings in
the Giants of Backgammon, and have lower Snowie ratings when they play
over the board.  They are clearly connected and consistent.

Now, there are a couple of exceptions, and I won't embarrass people by
naming names, but I do know a couple of people who are exceptional and
extremely accurate when they look at a tough position on paper, but
they do not do well in tournaments, and, as you pointed out, often
make silly errors on simpler plays.

In summary, I agree with you completely.