Re: Creative Commons Attribution license element

[Evan Prodromou <evan@debian.org>]

>>>>> "AS" == Andrew Suffield <asuffield@debian.org> writes:

    Me> One thing that bothers me, though, is how this becomes 'barely
    Me> free'.

    AS> Freedom is a binary test; a work is either free, or it is
    AS> not. There is no "partially free" or "semi-free". So "barely
    AS> free" is "free, but very close to the line; make this any
    AS> stronger and it will be non-free".

I understood that part. The part I don't understand is, what line does
the URL requirement get close to? What is the line between acceptable
modification restrictions and unacceptable ones?

I'll try to present one dimension of modification restrictions, being
what type of content those restrictions require to remain in the
modified version:

  1. No restrictions on modifications whatsoever.

  2. Restrictions to make sure that downstream users get the following
     4 things: copyright notices, license notification, changelogs,
     warranty disclaimers.

  3. Restrictions to make sure that downstream users get general
     information about the work (metadata), including the above 4
     things, but maybe perhaps some others. Examples: original author
     name, original work title, original metadata URL, bug-reporting
     site or address, main download site URL.

  4. Restrictions that "protect" certain parts of the work itself
     (e.g., invariant sections).

  5. Restrictions that prevent any modification of the work
     whatsoever.

I'd posit that our line for free-vs.-non-free is between 3 and 4 here.

    Me> Would it be non-free because it's not possible for the licensee
    Me> to comply because the license is vague?

    AS> Licenses which are vague are particularly nasty, because you
    AS> can go with the "obvious" interpretation, and then get sued by
    AS> the copyright holder who turns out to have a different
    AS> one. Certainly we've had some copyright holders applying
    AS> strange interpretations to apparently free licenses before
    AS> now. To provide reasonable assurance to our users that
    AS> everything in main is free, we have to take the most
    AS> pessimistic interpretation, and see if that is free.

Cool. So, if I can redirect back to the Attribution license, we have a
worst-case interpretation with the Attribution license, where the
author's credits have to be listed in _every_ place other credits are
given, even if those places are possibly inconvenient (file name, work
title). Even though it's vague (could mean "some" places, could mean
"all" places), our New Math set theory teaches us that if we give
credit in all places, we will have also have given credit in some
places. So, by giving credit in _all_ places, we can comply.

Now, given the "all places" idea: is that non-free? It may suck, but
is it non-free?

~ESP

-- 
Evan Prodromou
Email: evan@debian.org
Jabber: evan@jabber.debian.net