[computer-go] scalability with the quality of play-outs.

[Don Dailey]

Gian-Carlo Pascutto wrote:
> Don Dailey wrote:
>
>> First of all,   I am not aware of any published work on this specific
>> thing.   There may be some, but I'm not aware of it.   
>
> Thanks, this was what I was curious about.
>
> The rest of your story is rather anecdotal and I won't comment on it.
Are you trying to be politely condescending? 
 
>
> Note that I agree on the starting premise and by extension the result
> from the "light" study is the one I most expected.
I expected this result too,  but there is nothing like seeing it
verified to be sure.     I'm not sure what you mean by the "starting
premise."     What is the starting premise? 

> I was surprised by the original Mogo-Leela result but the "light"
> result seems to show it was a bit of a coincidence.
I'm sorry, I don't follow.   What is surprising about the original
Mogo-Leela result?   Is it better or worse than you expected? 

>
>> So in the early days it seemed more productive to make your program
>> faster - but now it's more productive to make them smarter.       I
>> believe that if you ask any computer chess programmer that has been
>> doing this for 20 years or more, you would get this same
>> understanding.    
>
> I think it depends on how you define "smarter". Is that like "more
> intelligent" ?
What I mean is that the evaluation function is of better quality - knows
more about chess in some sense.  

In some ways I admit that it's hard to separate the evaluation function
from the program.    A poor EF can affect other parts of the program
such as the quality of the selectivity and even the efficiency of the
cutoffs.    

- Don