Re: New 'Public Domain' Licence
On Fri, Jun 03, 2005 at 12:53:34PM -0700, Don Armstrong wrote:
> First of, please use your real name when discussing things upon this
> list. Anonymity makes it rather difficult for others to follow your
> arguments, and interferes with the primary mission of debian-legal.
I usually just ignore anonymous/pseudonymous messages on technical lists.
> > The above copyright notice and this permission notice shall be
> > included in all copies or substantial portions of the Software].
> The part above is almost a no-op, and a good idea regardless, because
> it informs recipients of the work what their rights are, and enables
> them to sanely to exercise the granted rights upon the work.
It doesn't necessarily allow people to do much of anything. If I'm
distributing a closed work that uses a few pages of MIT-licensed code,
there's no practical value in showing the MIT license (or other permissive
licenses) to users. Why, as a user or a programmer, should I care that
the portions of a black box that come from Lua are under a permissive
license? Saying "this program uses Lua" has value--giving credit--but
telling me that I can freely distribute the part that is Lua has no value,
since I can't actually do so (it's tucked away inside a binary; if I want
Lua, I'll go download the source).
In fact, it's potentially confusing; you have to be careful to be clear that
only certain embedded portions are under the license, not the work as a whole.
Also, due to license proliferation, different MIT-ish projects are actually
under a collection of slightly varying permissive licenses, which prevents
simply listing the one license and merging the copyright holder names--it's
not hard to end up having to list half a dozen variants. For a project
whose documentation is a simple "plug it in, turn it on, don't stick your
fingers in the fan" pamplet, this isn't a trivial problem--the licenses
can be bigger than the documentation.
I use the MIT license myself, but I can say from experience that there in
some scenarios, it has costs without any benefit. I do think this is a
minor bug in the license: many people (such as myself) who use permissive
licenses do so specifically to make it easy for anyone to use their code
for any purpose, even proprietarily, without licensing getting in the way.
I don't consider this significant enough to offset the cost of proliferating
a new license, but I think it's worth acknowledging.
If anyone really wants a license that doesn't have this problem, there's
libpng's, which only requires the license statement in source distributions.
(Unfortunately, the name of the software, "The PNG Reference Library", is
used in the main body of the license, so changes beyond the copyright notice
are required--unlike the MIT license, it's not a simple drop-in license. It
also has an obnoxiouly wordy "Contributing Authors" section, instead of
simply using a (c) notice.)