Ian Skerrett last week suggested that there is a growing trend in favour of permissive non-copyleft licenses at the expense of reciprocal copyleft licenses. Ian asked “name one popular community open source project created in the last 5 years that uses the AGPL or GPL?”
The responses didn’t exactly come thick and fast. I certainly couldn’t think of one. But the question did prompt me to look for some evidence for the trend away from copyleft licenses.
The first port of call for evidence of trends related to open source license use is Black Duck’s Open Source Resource Center. The lastest figures show that GPLv2 is used for 45.33% of projects in Black Duck’s KnowledgeBase, while the GPL family accounts for roughly 61% of all projects.
While the GPL family is dominant, comparing the latest figures with those provided in June 2008, June 2009, and some previous CAOS research from March 2010 indicates a steady decline in the use of the GPL family and the GPLv2 in particular.
According to Black Duck’s figures the proportion of open source projects using the GPL family of licenses has fallen to 61% today from 70% in June 2008, while the GPLv2 has fallen to 45% from 58% three years ago.
It is worth noting that the number of projects using the GPL licenses has increased in real terms over the past few years. According to our calculations based on Black Duck’s figures, the number of GPLv2 projects rose 5.5% between June 2009 and June 2011, while the total number of open source projects grew over 16%.
We should expect to see slower growth for the GPLv2 given it has been superseded but even though the number of AGPLv3 and GPLv3 projects grew 90% and 85% respectively over the past two years, that only resulted in 29% growth for the GPL family overall (while A/L/GPLv3 adoption appears to be slowing).
In comparison the number of Apache licensed projects grew 46% over the past two years, while the number of MIT licensed projects grew 152%. Indeed Black Duck’s figures indicate that the MIT License has been the biggest gainer in the last two years, jumping from 3.8% of all projects in June 2009 to 8.23% today, leapfrogging Apache, BSD, GPLv3 and LGPLv2.1 in the process.
While the level of adoption of copyleft licenses remains dominant, and continues to rise in terms of the number of projects, there is no escaping the continuing overall decline in terms of ‘license share’.
UPDATE – Since some people dod not trust Black Duck’s data I also took a look at data collected by FLOSSmole. The results are remarkably similar. – UPDATE
Black Duck’s data is not the only indication that the importance of copyleft licenses has decreased in recent years. The research we conducted as part of of our Control and Community report also indicated a decline in the number of vendors engaging with strong copyleft licensed software.
Specifically, we evaluated the open source-related strategies of 300 software vendors and subsidiaries, including the license choice, development model, copyright strategy and revenue generator.
By plotting the results of this analysis against the year in which the companies were founded (for open source specialists) or began to engage with open source (for complementary vendors) we are able to gain a perspective on the changing popularity of the individual strategies*.
Having updated the results to the end of 2010, our analysis now covers 321 vendors and shows that 2010 was the first year in which there were more companies formed around projects with non-copyleft licences than with strong copyleft licences.
The formation of vendors around open source software with strong copyleft licenses peaked in 2006, having risen steadily between 1997 and 2006 – although there have been gains since 2007. By comparison, the formation of vendors around open source software with non-copyleft licences has been steadily increasing since 2002.
The results get even more interesting in terms of Ian’s question if we filter them by development model. Looking at community-led development projects, we see that there have been significantly more companies formed around community-led projects with non-copyleft licenses than with strong copyleft licenses since 2007.
In fact, strong copyleft licenses have been much more popular for vendor-led development projects, but even here there was an increase in the use of non-copyleft licenses in 2010.
This last chart illustrates something significant about the previous dominance of strong copyleft licenses: that it was achieved and maintained to a significant degree due to the vendor-led open source projects, rather than community-led projects.
One of the main findings of our Control and Community report was the ongoing shift away from projects controlled by a single vendor and back toward community and collaboration. While some might expect that to mean increased adoption of strong copyleft licenses – given that they are associated with collaborative development projects such as GNU and the Linux kernel – the charts above indicate a shift towards non copyleft.
As previously noted, while free software projects utilize strong copyleft to ensure that the software in question remains open (or as Bradley M Kuhn recently put it, to keep developers “honest”), vendors using the open core licensing strategy use strong copyleft licenses, along with copyright ownership, to ensure that only they have the opportunity to take it closed.
Either way, strong copyleft is used as a means of control on the code and the project, and our analysis backs up Ian’s contention that there is a trend away from control and towards more permissive non-copyleft licenses.
This is part of what we called the fourth stage of commercial open source business strategies and is being driven by the increased engagement of previously closed-source vendors with open source projects.
The fourth stage is about balancing the ability to create closed source derivatives with collaborative development through multi-vendor open source projects and permissive licensing, and as such it not only avoids the need to control a project through licensing, it actively discourages control through licensing.
That is why, in my opinion, the decline of the copyleft licenses has only just begun.
*The method is not perfect, since it plots the license being used today against the year of formation, and as such does not reflect licensing changes in the interim. It does provide us with an overview of general historical trends, however.