In the recent FOSDEM 2016 SDR Devroom, the
Q&A session following a presentation on OpenAirInterface touched the topic of its
controversial licensing. As I happen to be involved deeply with Free
Software licensing and Free Software telecom topics, I thought I might
have some things to say about this topic. Unfortunately the Q&A session
was short, hence this blog post.
As a side note, the presentation was actually certainly the least
technical presentation in all of the FOSDEM SDR track, and that with a
deeply technical audience. And probably the only presentation at all at
FOSDEM talking a lot about "Strategic Industry Partners".
Let me also state that I actually have respect for what OAI/OSA has been
and still is doing. I just don't think it is attractive to the Free
Software community - and it might actually not be Free Software at all.
OpenAirInterface / History
Within EURECOM, a group around Prof.
Raymond Knopp has been working on a Free Software implementation of all
layers of the LTE (4G) system known as OpenAirInterface. It includes the physical layer
and goes through to the core network.
The OpenAirInterface code was for many years under GPL license (GPLv2,
other parts GPLv3). Initially the SVN repositories were not public
(despite the license), but after some friendly mails one (at least I)
could get access.
I've read through the code at several points in the past, it often
seemed much more like a (quick and dirty?) proof of concept
implementation to me, than anything more general-purpose. But then,
that might have been a wrong impression on my behalf, or it might be
that this was simply sufficient for the kind of research they wanted to
do. After all, scientific research and FOSS often have a complicated
relationship. Researchers naturally have their papers as primary output
of their work, and software implementations often are more like a
necessary evil than the actual goal. But then, I digress.
Now at some point in 2014, a new organization the OpenAirInterface
Software Association (OSA) was established. The idea apparently was to
get involved with the tier-1 telecom suppliers (like Alcatel, Huawei,
Ericsson, ...) and work together on an implementation of Free Software
for future mobile data, so-called 5G technologies.
Re-Licensing
The existing GPLv2/GPLv3 license of the OpenAirInterface code of course
would have meant that contributions from the patent-holding telecom
industry would have to come with appropriate royalty-free patent
licenses. After all, of what use is it if the software is free in terms
of copyright licensing, but then you still have the patents that make it
non-free.
Now the big industry of course wouldn't want to do that, so the OSA
decided to re-license the code-base under a new license.
As we apparently don't yet have sufficient existing Free Software
licenses, they decided to create a new license. That new license (the
OSA Public License V1.0
not only does away with copyleft, but also does away with a normal
patent grant.
This is very sad in several ways:
license proliferation is always bad. Major experts and basically all
major entities in the Free Software world (FSF, FSFE, OSI, ...) are
opposed to it and see it as a problem. Even companies like Intel and
Google have publicly raised concern about license Proliferation.
abandoning copyleft. Many people particularly from a GNU/Linux
background would agree that copyleft is a fair deal. It ensures that
everyone modifying the software will have to share such modifications
with other users in a fair way. Nobody can create proprietary
derivatives.
taking away the patent grant. Even the non-copyleft Apache 2.0
License the OSA used as template has a broad patent grant, even for
commercial applications. The OSA Public License has only a patent
grant for use in research context
In addition to this license change, the OSA also requires a copyright
assignment from all contributors.
Consequences
What kind of effect does this have in case I want to contribute?
I have to sign away my copyright. The OSA can at any given point
in time grant anyone whatever license they want to this code.
I have to agree to a permissive license without copyleft, i.e.
everyone else can create proprietary derivatives of my work
I do not even get a patent grant from the other contributors (like
the large Telecom companies).
So basically, I have to sign away my copyright, and I get nothing in
return. No copyleft that ensures other people's modifications will be
available under the same license, no patent grant, and I don't even keep
my own copyright to be able to veto any future license changes.
My personal opinion (and apparently those of other FOSDEM attendees) is
thus that the OAI / OSA invitation to contributions from the community
is not a very attractive one. It might all be well and fine for large
industry and research institutes. But I don't think the Free Software
community has much to gain in all of this.
Now OSA will claim that the above is not true, and that all contributors
(including the Telecom vendors) have agreed to license their patents
under FRAND conditions to all other contributors. It even seemed to me
that the speaker at FOSDEM believed this was something positive in any
way. I can only laugh at that ;)
FRAND
FRAND (Fair, Reasonable and Non-Discriminatory) is a frequently invoked
buzzword for patent licensing schemes. It isn't actually defined
anywhere, and is most likely just meant to sound nice to people who
don't understand what it really means. Like, let's say, political
decision makers.
In practise, it is a disaster for individuals and small/medium sized
companies. I can tell you first hand from having tried to obtain patent
licenses from FRAND schemes before. While they might have reasonable
per-unit royalties and they might offer those royalties to everyone,
they typically come with ridiculous minimum annual fees.
For example let's say they state in their FRAND license conditions you
have to pay 1 USD per device, but a minimum of USD 100,000 per year. Or
a similarly large one-time fee at the time of signing the contract.
That's of course very fair to the large corporations, but it makes it
impossible for a small company who sells maybe 10 to 100 devices per
year, as the 100,000 / 10 then equals to USD 10k per device in terms of
royalties. Does that sound fair and Non-Discriminatory to you?
Summary
OAI/OSA are trying to get a non-commercial / research-oriented foot into
the design and specification process of future mobile telecom network
standardization. That's a big and difficult challenge.
However, the decisions they have taken in terms of licensing show that
they are primarily interested in aligning with the large corporate
telecom industry, and have thus created something that isn't really Free
Software (missing non-research patent grant) and might in the end only
help the large telecom vendors to uni-directionally consume
contributions from academic research, small/medium sized companies and
individual hackers.