Subject: Re: An explanation of the difficulty of solving licenseproliferation in one sentence
From: Alex Bligh <alex@alex.org.uk>
Date: Wed, 09 Mar 2005 20:09:09 +0000



--On 09 March 2005 14:56 -0500 Ian Lance Taylor <ian@airs.com> wrote:

> As far as I can tell, this more or less assumes that good licenses
> will somehow convert code under bad licenses to be under good
> licenses.  What seems more likely to me is that code under bad
> licenses will be lost forever, and/or code will survive in various
> balkanized regimes under incompatible licenses.
>
> Your selection argument is tying together two things that I don't
> think should be tied together.  Natural selection of code is good.
> Natural selection of licenses is good.  Code being driven extinct
> because it is bad is good.  Code being driven extinct because it has a
> bad license is bad.

I am arguing that it's the author's own choice to contribute work
under a license that risks being balkanized. Note (because of the
asymmetry of license compatibility) that's far greater a threat
if they contribute under the GPL than the BSD license, but I don't
see many open source advocates arguing this makes the GPL inherently
bad (as opposed to "not my choice"). Equally, when a contributor
contributes to a project that already has a fixed license, that's
they take the risk of license incompatibility.

I don't see why this is any different to the risks an author / contributor
takes to code which is
a) written in an obscure language
b) ill documented
c) suffers from poor programming technique (global variables, poor
   error handling etc.)
d) otherwise incomprehensible

Having been in the position of deciding "which open source project
are we going to spend an awful number of man-hours modifying and
contributing to" before, I can tell you our choice was based on
(a)-(d) above. And when we'd decided, we checked the license was
not a problem for us.

What is a fair point is that I've used the word "choice" a lot above.
It the choice is not an informed choice, that's a bad thing, and the
OSI can help.

> Mind you, if all licenses were compatible, or permitted appropriate
> relicensing, I would not be concerned.

I can't see a situation (at least where reciprocal licenses exist)
where all licenses will ever be compatible. But I agree, features
to encourage in new licenses are (a) compatibility, (b) making
relicensing as painless as possible. Version rolling (for instance)
helps with (b).

Alex