draw scores in hash table

[Don]

Is there a consensus on how draw scores should be handled?

Currently my program just treats them the same. But it used to be considered bad to score a score of zero (or a repetition score that is) in the hash table.

Another system that I used to use is to make the repeition score unique by using all other score to be even numbers and using an odd value for the draw score (and not storing them in the hash table.)

Of course it's understand that none of these ideas prevent GHI problems - the only issue is whether it helps some.

[bob]

Is there a consensus on how draw scores should be handled?

Currently my program just treats them the same. But it used to be considered bad to score a score of zero (or a repetition score that is) in the hash table.

Another system that I used to use is to make the repeition score unique by using all other score to be even numbers and using an odd value for the draw score (and not storing them in the hash table.)

Of course it's understand that none of these ideas prevent GHI problems - the only issue is whether it helps some.

I avoided storing them and it slighly hurt cluster testing results. I store them normally, but for 50 move issues, when I get close to the 50 move counter, I do go thru and invalidate the scores at something like move 40 and 45, to let me search around those kinds of draws when possible.