On Mon, 9 Jun 2003 14:25:47 +0000 (UTC), Robert Hyatt
wrote:

wrote:

When I set my chess computer to search a certain number of plies, are
odd numbers or even numbers stronger?

With today's programs it is very unlikely there is any difference.

Did it used to make a difference?

I remember when looking at simple alpha-beta search algorithms without
extensions, many years ago, it seemed the odd plies took a lot more
time to get through if there was decent move ordering.

(But my memory may be hazy or I may be confused or I may just not have
understood a dang thing.)



Deeper is stronger of course, but also slower.

In chess a ply is a one-half move. Many show the ply as depth using the
abreviation d as d=4 on the
Setting of 4 ply has the machine seach to a depth of two full moves w-b-w-b
or b-w-b-w
depending who has the move.
One ply is to look one move w or b deep.

The more plies the deeper the search, which can increase the time to search
considerably.
"Grinch" wrote in message
...
On Mon, 9 Jun 2003 14:25:47 +0000 (UTC), Robert Hyatt
wrote:

wrote:

When I set my chess computer to search a certain number of plies, are
odd numbers or even numbers stronger?

With today's programs it is very unlikely there is any difference.

Did it used to make a difference?

I remember when looking at simple alpha-beta search algorithms without
extensions, many years ago, it seemed the odd plies took a lot more
time to get through if there was decent move ordering.

(But my memory may be hazy or I may be confused or I may just not have
understood a dang thing.)



Deeper is stronger of course, but also slower.

On Fri, 1 Aug 2003 14:56:09 -0400, "Jim Roe"
wrote:

In chess a ply is a one-half move. Many show the ply as depth using the
abreviation d as d=4 on the
Setting of 4 ply has the machine seach to a depth of two full moves w-b-w-b
or b-w-b-w
depending who has the move.
One ply is to look one move w or b deep.

The more plies the deeper the search, which can increase the time to search
considerably.

Yes, obviously, but the point is that on ply one (odd) one must
generate and examine every possible move to find the best one, while
at ply two (even) one need look at only a fraction of the moves using
alpha-beta to find the best response (assuming decent move ordering),
and then so on through plies three & four, etc. So in this sense the
even plies seems more "efficient", seeing ahead a move with less
additonal effort.

Of course, that's with the simplest of algorithms. I wouldn't doubt
that with search extensions et al the difference washes out

"Grinch" wrote in message
.. .
On Mon, 9 Jun 2003 14:25:47 +0000 (UTC), Robert Hyatt
wrote:

wrote:

When I set my chess computer to search a certain number of plies, are
odd numbers or even numbers stronger?

With today's programs it is very unlikely there is any difference.

Did it used to make a difference?

I remember when looking at simple alpha-beta search algorithms without
extensions, many years ago, it seemed the odd plies took a lot more
time to get through if there was decent move ordering.

(But my memory may be hazy or I may be confused or I may just not have
understood a dang thing.)



Deeper is stronger of course, but also slower.