Re: [computer-go] future KGS Computer Go Tournaments - two sections?

[Martin Mueller <mmueller@xxxxxxxxxxxxxx>]

David Doshay wrote:

> As mentioned, Explorer and Go Intellect are derivative programs, and 
> these are considered "original enough" or "divergent enough" from the 
> original to be considered unique. It is worth noting that these 
> programs are derived from a software base explicitly designed to be a 
> "workbench" for the development of game programs.
There are three programs in this group: Ken's Go Intellect, our 
Explorer and Anders' current SmartGo. All of these programs have 
completely separate and independently developed playing engines. Their 
common point is that they have evolved from the same base-level code 
such as Go Board, legal moves, liberties, etc. Probably the highest 
level shared code is the ladder reading code (I'm not sure if Ken still 
uses the same one).

So from the point of playing logic, these are not "derivative 
programs", they are completely independent efforts. From the point of 
sharing code and effort, they are not independent.

I appreciate the effort in this discussion to clarify these issues. For 
example, all programs that use standard C or C++ libraries share some 
code. Are they "derivative" of the libraries? Clearly not. So what if a 
program uses a publicly available or private Go-specific library? Does 
it depend on what exactly is in that library? Where to draw the line? 
Not easy questions to me.

As a more concrete example, what if one of my students develops his or 
her own Go engine which is branched off from Explorer. Will it always 
be "derivative work"? Will it count as a separate program if it is 
"sufficiently different" from Explorer? Who decides? If I say it is 
different enough, does that count?

Here is yet another scenario: assume we see a second, third,... strong 
open source program, with different strengths and weaknesses. Now 
people find ways to "mix and match" parts of different programs and 
create a stronger one. Let's say they play by the rules and make it 
open source. Now who will be allowed to participate? Does it need the 
consent of all programs it is built on? Or only those programs that 
contain Go-specific code? What if the main contribution is some new 
game-independent search method, which is taken from a (say) Othello or 
Hex program?

A final comment: let's say there is an Atari-Go tournament, and someone 
enters a modified GNU Go. Should that be allowed? Why, or why not? What 
if different people enter different modified GNU Go's?

Lots of room for thought.

	Martin

_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/