[OSM-legal-talk] Circumnavigating Share-Alike through software / now and future

Joseph Gentle josephg at gmail.com
Tue Oct 28 12:43:26 GMT 2008


On Tue, Oct 28, 2008 at 9:45 PM, Rob Myers <rob at robmyers.org> wrote:
> The scenario that you describe looks similar to the "distributing
> proprietary object files for end users to link with GPL software"
> exploit that I think was explicitly tackled in GPL 3. This scenario
> should have been known to CC when they were drafting their licences,
> and I'd hope they addressed it.

How did they tackle it?

It seems to be one of those class 1 vs class 2 errors. Your set of
rules will either allow things it shouldn't, or disallow things it
should. Its very hard to avoid both cases.

I think that happens because we're moving away from writing what we
actually want the license to say. Instead we're trying to write a
legalese approximation to it. It will never say the same thing.

I think we want to say: "We want people to share back improvements to
the map when practically possible. Bigmegaevilcorp can't use our maps
at all unless they stop being so darn evil first."

Instead our license reads like: "when making a derivative work, you
have to share it back in this case, but not this case. Research is ok.
DRM is not ok. Certain other open source projects can use our stuff.
Other mapping providers can't combine our stuff with theirs unless
they share their work too or sacrifice a goat in a full moon on their
database cluster...."

I also worry that some of these use cases could be written multiple
ways: "a poor research student wants to make maps like this" vs "big
evilmegacorp wants to make maps like this" and people will interpret
them as positive or negative examples based on the student vs megacorp
rather than on how they're using the maps.

If we invent an arbitrary set of rules which match our attempted
positive / negative examples rather than based on the principles on
which the license is theoretically based, we will end up with a bad
license. Certainly in software, designing like this always leads to
brittle code that isn't able to handle any of the edge conditions you
forgot to test for. Robust code is made by writing what you mean, not
just writing what you want the computer to do.

The classic example of this is the GPL. Stallman originally said "I
want any software which builds on my software to maintain these basic
freedoms". Then they wrote the GPL which instead reads: "anything
derivative works must use this exact license" - a statement that
sounds similar but is perversely different in complex and confounding
ways.

-J

> - Rob.
>
> _______________________________________________
> legal-talk mailing list
> legal-talk at openstreetmap.org
> http://lists.openstreetmap.org/listinfo/legal-talk
>




More information about the legal-talk mailing list