Re: Plagiary problem

[David Fotland <fotland@xxxxxxxxxxxxxxx>]

At 03:35 PM 4/26/99 -0400, Elmer Elevator wrote:
>  During the last round of the plagiarism controversy, one wag
>thoughtlessly asked a question something like this: How can a filched
>program play better Go than the program from which it is filched? ... and I
>don't recall anyone satisfactorily answering that. If a plagiarist just
>repackages stolen code, s/he is competing against programs s/he can't
>possibly defeat, but at best only tie ... 

Strange question.  Handtalk and most other programs have some randomness
built in, and all programs are weak enough that their play is inconsistent.
With Ing rules or Japanese with Komi there are no ties.

So even if two identical copies of the same program are entered, one will
win and one will lose.  If you are competing against two copies of
your own program you will only have a 1/3 chance of winning.

>and in the six months or longer
>since the theft, the programmer from whom s/he stole the code will
>certainly have improved his/her code sufficiently to defeat the stolen
>stale version. 

Prof Chen stated publicly several times that he was rewriting his program.
He even gave it a new name (Goemate).  In his own tests, Goemate didn't
win 50% against last year's Handtalk.  So it is not surprising that the
two plagiarized programs beat Goemate.

>If a plagiarist steals effective code and changes it
>sufficiently so that it's a consistent winner in the next competitions --
>well, then, shame on him/her for the initial theft, but it's no longer the
>stolen Fiat ... s/he's enhanced it by his/her own creativity to a
>supercharged new BMW, essentially a product which would have to be viewed
>as achieving its new victories by original work. Maybe not a lot of new,
>novel, original, creative work ... 

Hundreds of people have spent many years each trying to build go programs.
Only a few become strong, and Handtalk is one of those few.  It's
not unreasonable to be upset when that work is stolen.  If BMW stole
Fiat secrets they would be sued and the people who stole the plans
would go to jail.  Why should computer go be any different.  If
people want to share code of course that is fine, and there have
been many open source go programs in the past.  Until now, none
have been very strong.

but the museums are filled with lots of
>things that were almost telephones or almost airplanes. We call each other
>on the one Bell finally improved on, and fly around the world on the one
>the Wright Brothers invented. If there is a reply to this, I hope it won't
>begin with "It's not as simple as that." But if it does, I hope the
>complicated replies are, at least, logical and rigorous. I'd be happy
>(well, maybe not happy, but content and pleased) for anyone who can
>rigorously illuminate the fatal flaws in this guy's argument. >It seems to
>me that part of the thing about coding go has less to do w/ becoming
>millionaires and >that it is an interesting problem.. 

If you spend years of your life working on this interesting problem,
and are acknowledged as the best through the world championship, then
someone steals your code, and gives you no acknowledgement, this is
upsetting.  The people who stole the code did it solely for the
money.  They licensed their code for distribution in Japan, and
probably made over $100,000 between them.

In the examples you cite, and in any scientific endeavor, people
always acknowledge the work that they build upon.  If a scientist
claims derivative work as being entirely hi/her own, that scientist
will no longer be able to work as a scientist, since he/she has shown
a lack of integrity.

You have an unusual attitude toward intellectual property :)

David

PS I only use 3 PC's regularly (two for testing, and one for development).
3 are used by my wife and kids :)

(I'm guessing that a
>generic chess proggie makes more money than 
>>your typical go program).         -avi In the global picture, there may be
>more money for chess proggies than for Go proggies, but one well-known Go
>programmer reports that there's lots of abundant corporate money available
>for good Go programs in Asia, and his seven personal computers (probably
>more by now) are living proof. So whatever the truth about this
>controversy, there is quite a bit at stake in hard cash. Bob Merkin
>http://www.javanet.com/~bobmer/ 
>