[computer-go] Collaboration project. Once again

[Mark Boon]

On 2-aug-2006, at 19:16, House, Jason J. wrote:

> If by "how to start" you mean coming up with a high level design for
> the eventual growth/development of a bot, I think you should.  Even if
> the discussion goes nowhere (e.g. no general consensus), it will still
> have a number of benefits.  Most notably, those who resonate with your
> ideas are more likely to join your project.  I firmly believe that by
> discussing things in groups, at least a few new/good ideas come out.
> By defending your design, you'll create an even better one.
>

I'm afraid that by doing that we'll end up bickering about lots of  
trivial things without actually really getting anything done. But  
let's say I'll give a serious thought whether that would be a  
worthwhile exercise. I agree that presenting and defending ideas and  
designs strengthens the original thought. I will need to find a  
balance between having enough discussion and exchange of ideas and  
not letting endless discussion bog down the project into compelte  
paralysis.

I also want to stress that I currently don't have a high-level design  
ready. What I have is a general idea which way I want to go, a lot of  
past experience of what kind of designs I need in practice and some  
rough parts that I think every Go playing program needs. It's  
impossible to design the whole thing in advance, so I prefer to start  
with one of the rough parts and start fleshing it out. We'll learn as  
we go. One of these rough parts is an evaluation module. I have a  
pretty good idea how to make the basic necessities for one. And I  
think I have a pretty good idea of what I need to make an extensible  
one and a supporting framework. So I think that's a good place to start.

Mark