# Computer Chess Club Archives

## Messages

### Subject: Re: Branching factor, make me confuse more that ever.

Author: Robert Hyatt

Date: 19:32:30 04/03/00

Go up one level in this thread

```On April 03, 2000 at 17:56:20, Tom Kerrigan wrote:

>On April 03, 2000 at 15:33:48, Robert Hyatt wrote:
>
>>On April 03, 2000 at 00:06:03, Tom Kerrigan wrote:
>>
>>>On April 01, 2000 at 13:38:00, leonid wrote:
>>>
>>>>Hello!
>>>>
>>>>Maybe you could take me out of my endless confusion about "branching factor".
>>>>Confusion come from the way that you can compare two different games. Would like
>>>
>>>You are taking a totally different approach to computer chess than everybody
>>>else in the world.
>>>
>>>You are driving a boat when everybody else is driving a car.
>>>
>>>This is fine, but the problem is that you are insisting on comparing your boat
>>>to everybody's car. You're trying to equate sail size to wheel diameter. It's
>>>possible, but it couldn't be more useless.
>>>
>>>Your program does not do quiescence searches, it does not do extensions, it
>>>probably doesn't do iterative deepening, etc. Comparing your program to other
>>>programs which DO have these features is not productive.
>>>
>>>Until you decide to add these features, you should simply concentrate on
>>>improving your program and not worry about what other people are doing.
>>>
>>>If you have to know, here's how you can compute your branching factor: count how
>>>many moves you search at each node. Divide by the number of nodes.
>>>
>>>-Tom
>>
>>
>>That's not 'branching factor'.  that is "effective branching factor".  Because
>>at many nodes you search 1 branch, but there are obviously many more moves there
>>that _could_ be searched...
>>
>>this has been a source of confusion almost forever...
>
>So it's totally correct for me to say that Crafty's branching factor is 38.
>
>-Tom

NO

Crafty doesn't have a "branching factor" of 38.  The game of chess has a
branching factor of roughly 38.  Crafty has an "effective branching factor"
of 2.5-3.0, roughly.

As I said, branching factor is the average number of legal moves at any node
in the tree, which is why checkers is often given as 38, while go is on the
order of 120 or so.  Alpha/beta can 'effectively' lower these numbers, but
the proper term is then "effective branching factor".

```