Lies, Damn Lies, and Statistics

The c|net News Blog recently reported on a study performed by Palamida, an open-source risk management company, that claims GPLv3 adoption at 50% of all GPL projects.   As the blogger says: “That’s huge.”  Or rather, it would be if it were true.

According to Palamida, 382 projects have announced that they would switch to GPLv3 as of August 21, 2007.   That’s the number of projects that have announced, not necessarily the number of projects that have released code under a GPLv3 license.  For example, this number includes the Samba and SugarCRM projects, which have publicly announced that their next releases would be licensed under GPLv3 but have not yet released GPLv3-licensed code.  But even assuming that all announced projects will switch, c|net News and Palamida have resorted to some pretty tricky footwork to report that this relatively small number is a 50% conversion rate.

Basically, what they did was to count all the projects licensed under version 2 “or later” as having converted to GPLv3.  Furthermore they used unexplained estimates of “conversion rates” and “impact” to arrive at the “or later” figure in the first place.  No matter how you slice it, that is dishonest.

Palamida themselves wrote in a different post that “[b]y using the ‘or later’ phrase, the copyright holder is effectively leaving it up to the end user to decide which version of the license they want to use.”  This interpretation meshes with what the Free Software Foundation says (FSF), and the FSF also adds that even if the recipient of code licensed version 2 or later wants to switch to version 3, that user cannot bind the distributor of the code to the requirements of version 3.

What’s worse is that both Palamida and the c|net blogger know that what they reported is not true.  Just a month ago, Palamida reported that 116 projects had converted to GPLv3 and said: “Palamida estimates that the conversion represents less than 1% of GPL’d projects.” 

Frankly, having 382 converted and announced projects this soon after GPLv3 was released is more than respectable.  It should take a long time for projects to decide whether they want to switch – this is a serious decision that will impact everyone involved.  The FSF intended for GPLv3 to have a major impact on intellectual property policies, and project stake-holders should make sure they agree with the intended effects of this new license (if they can understand the effects, but that’s a different post).

What is not respectable, what is not serious, is reporting the 382 announced projects as a 50% conversion rate to GPLv3.  The free software community deserves better, and so do we.

Comments

[Ernest Park]

In Support of the Facts - Palamida's GPLv3 Statistics
Noah Clements recently posted a blog questioning Palamida's GPLv3 conversion numbers. Healthy debate is a sign of an involved and engaged community and therefore I felt it was important to respond directly to the points Noah makes to clarify key statements.

The inclusion of statistics specific to "GPL, or later" clauses on Palamida's GPLv3 site is intended to provide an indication of the current and future support for the Free Software Foundation (FSF) and the variants of the existing GPL licenses. By providing statistics on which licenses now include the "GPLv2 or later" clause and which are moving that direction, Palamida is in no way stating that these projects are to be considered actual conversions to the GPLv3 or LGPLv3 licenses.

Palamida's research team takes the responsibility of reporting accurate and detailed information very seriously and we work overtime to provide timely data regarding licensing issues, adoption of specific licenses, and their unique considerations such as the "or later" clause that leaves the license decision up to the individual user. We feel that the scope of this information is an important indicator of the OSS community's reaction to the new GPLv3 license. As Palamida continues to refine our site and enhance our data, we invite feedback and participation from our users, and in fact, require it in order to effectively improve upon how we deliver our content.

In the event that any of the information we provide lacks clarity, or as Noah feels, accuracy, we encourage users to contact us with their concerns and we will work to ensure that our reporting is made more clear in the future.

Should you have any questions or comments regarding the GPLv3 and OSS licensing related data, please feel free to contact me directly at ernest.park@palamida.com.

--Ernie Park