Alex Bligh alex at
Thu Apr 14 10:15:20 UTC 2005

--On 13 April 2005 18:11 -0400 "Michael R.Bernstein" <webmaven at> 

> To me, there is a qualitative difference between an Initial Developer
> (ID) requiring copyright assignment (for which, if the ID is a business,
> they might even be willing to pay for) in order to roll a modification
> into the main release of a project and a license that pre-determines that
> such an assignment to the ID has already been made if the modification
> has been distributed.

As someone promulgating a blatantly asymmetric license (OVPL), I agree,
these two are qualitatively different because in the latter case the
modifier has no choice but to give the ID additional rights if the
modifier distributes.

However, I'd add a word of caution.

1. Whilst it's clear that asymmetric licenses are not suitable for most
   new projects starting as open-source, I don't think that necessarily
   means they are not "open source".

2. Similarly, I am a bit perplexed by this "recommended" license tiering.
   Clearly, I would not recommend OVPL or any other similar license for
   a project started in open source. Similarly, I would not recommend
   GPL for a project started in a closed source environment which for
   various reasons (I won't repeat here) it would not be practical to
   license (even dual license) under GPL. Surely the principle of
   recommendation should be to recommend the best-of-breed licenses FOR
   A PARTICULAR JOB. Occasionally, an asymmetric license is going to be
   REQUIRED by the ID, or there will be no open source license at all.
   I have a feeling we should concentrate on deprecating duplicative
   and poorly framed licenses, or licenses which are so specific they
   are unlikely to have wide utility, rather than look to qualities like
   asymmetry as reasons not to recommend.

3. The above said, those making license decisions should be making
   an informed choice. If a license is asymmetric, it should quite
   clearly say so. Hopefully in a preamble as well. It should not be
   "snuck in". Equally, there are licenses that appear asymmetric
   (for instance the MPL - that's Mozilla - and the CDDL because of
   their separate grant clauses), but in practice are not (except
   arguably because the steward can change the license but that can
   be resolved by the ID electing to use a fixed license version).

4. Note there is more than one sort of asymmetry. Some demand a conditional
   license back to the original ID rather than a "no strings" assignment.
   Some demand additional privileges for the ID in terms of maintaining
   credit, or distributing only in the form of original source plus

> Speaking only for myself, this [asymmetric license]
> isn't the sort of deal I would find
> attractive unless the modifications I intend to make are mostly trivial.

Exactly. I would venture to put this more generally. A license which
is asymmetric is only going to be acceptable to people to the extent
the contributions of code are similarly asymmetric.


More information about the License-discuss mailing list