A Hug is Symmetric
An embrace is warm when two are pulled together. A one arm hug is a patronizing squeeze that makes for a (bad) photo op.
This little blog post is my > 140 response to my new friend @dberkholz's post The Story Of Data: Whither the GPL? Why we don’t need it anymore. I met Donnie at FOSDEM this year just after he joined RedMonk -- the analyst firm that is essential for anyone in software development to follow. (Full Disclosure: RedMonk and Informatique, Inc. do not have any business affiliation).
While I acknowledge that permissive licensing has become fashionable I think it is a grave disservice to suggest that restrictive licensing in FLOSS is withering, unneeded or for the uneducated.
This recent dust up is a result of an ongoing meme of "the Decline of the GPL" started last year by Matt Aslett. To which fuel was added by a recent BlackDuck analysis also asserting the the decline of the GPL. In precious few seconds of research I was unable to to find the BlackDuck report itself, but only mention of it. Ultimately the approach of the BlackDuck study is one of the problems. The data and methodology have not been made available for peer review -- the basis of the scientific method which defines progress in every academic discipline.
During our first Legal Issues DevRoom at FOSDEM we had several talks touching on the impact of software (and other artifact) licensing on FLOSS. Of special note: John Sullivan, Executive Director of the Free Software Foundation gave a talk "Is copyleft being framed?" and Richard Fontana, Red Hat's Open Source Licensing and Patent Counsel gave a talk, "The (possible) decline of the GPL, and what to do about it". Slides for these and other talks are available . Sullivan's data and methodology are available for review and suggest that the use of GPL is vibrant.
I do hope that my friend and former Sun colleague Rich Sands -- who is now at BlackDuck -- can help shed some light on their analysis.
But I'm not here to quibble about the data. I want to talk directly to the assertions made from the data.
1. Compliance is complicated
In this era of continuous development and continuous deployment powered by tools like the uber awesome Jenkins you can't really say with a straight face that making a tarball and publishing it somewhere is hard. Even in the embedded space there are tools like Yocto make delivering "Complete and Corresponding Source" just one of the build products.
For the massive, commercial enterprise which is Java™ Oracle manages to publish the source code for OpenJDK. Under the GPL. And Oracle publishes it from a tightly intermingled source base comprising open as well as closed, proprietary components.
2. The collaborative development model is really all you need
Bruce Perens was right: collaboration is better. Yet collaboration is necessary, but not sufficient to build a community. As we have become more familiar with FLOSS models it has become increasingly clearly that copyright assignment or licensing agreements that put a corporation in asymmetric control of a codebase does not foster the healthiest communities.
When inbound == outbound licensing and everyone is symmetric footing collaboration and contribution thrive.
3. Commercial products == proprietary products
"Not to mention that copyleft licenses make it much harder to build proprietary products". Well maybe we should start with understanding there might be a difference between building products and making a biz model around them vs. the licensing of said products.
With Red Hat hitting the milestone of $1 billion in revenue I think we can put to rest the question, "can you make money with open source?" Certainly Red Hat has some proprietary licensed products, but the crux or their business model is based on restrictively licensed, copyleft software. Red Hat invests an enormous amount of developer time to give back to the community... And apparently they are not suffering for it. Apparently this isn't too complicated for them. And apparently the bottom line is doing just fine, thank you.
4. Restrictive licensing doesn't matter in Cloud
If anything the rise of "Cloud Computing" drives the need for an updated approach to restrictive licensing. This was the real motivation behind creating the AGPL. Why is this? It's because traditionally restrictive licensing kicks in when the software is delivered. In web services you get data, but not software (in any form).
(Secret: data is more valuable than code )
5. You can't build a business on restrictive software
Jeremy Allison has clearly articulated why the GPLv3 is essential for the commercial Samba marketplace to thrive. He talks about symmetry providing a necessary level playing field (and he spoke about this recently).
Evan Prodromou has built the StatusNet business on AGPL'd software and is selling the Decentralized Social Web into Enterprises.
Work with Me
So whether or not the GPL is in decline (or not) only scratches the surface of the how the FLOSS revolution has transformed information technology in the past 20 years. I bet the that Story of Data in the next 20 years will tell us that symmetric collaboration is the big win.
Ultimately the key thing is to remember that permissively licensed software is also Free Software.