When you think of GitHub, you think of open source software. Of course, just putting your code on GitHub doesn’t make it open source; you still need to explicitly choose a license for your code that allows others to use it. A new look at the number of projects on GitHub made available under open source licenses reveals that a significant number of developers aren’t doing that. However, recent efforts by GitHub to encourage project maintainers to license their code and to simplify the process appear to be bearing fruit.
In July, 2013, GitHub released its license picker tool, choosealicense.com, to make choosing a license simpler for developers when they create a new project on the site and to help them adhere to best practices. To examine the effects of this initiative, last week GitHub published data on the use of open source licenses by project maintainers. The percentage of public non-forked repositories created each month with a license explicitly specified through a LICENSE file, which had been slowly declining over time, spiked to over 20% after the release of the license picker in mid-2013. Since then, the percentage has declined somewhat again, but remains close to 20%.
For more details behind the numbers, I reached out to Ben Balter, GitHub’s Government Evangelist, who leads the company’s efforts to encourage government to use and support open source software. He told me that, currently, the percentage of all public non-forked GitHub repositories that specify a license through a LICENSE file is about 16%. However, while that may seem like a low percentage, it’s a bit deceiving since it doesn’t include repositories which do specify a license in other ways, such as in a README file, code comments, or a machine-readable package management file. Balter told me that when taking these other methods into account, the number of repositories using an open source license, “may very well be the majority of public projects,” he wrote.
When asked for a theory as to why there’s been a general decline over time in projects specifying an open source license (again, through a LICENSE file), Balter told me, “The community talks less about licensing than it used to. Open source has won," he wrote in an email, "so many new developers may either take open source for granted, or may misunderstand copyright law and that by making their code available online, they are simply publishing it, not sharing it with others.”
Hence the reason that GitHub launched its license picker, the results of which are encouraging. Balter told me that the purpose of their analysis last week was, “to see how much, if at all, our efforts affected open source licensing, and the results exceeded even my highest open source expectations.” In addition, he’s heard from developers that the tool, “makes publishing your first open source project significantly less scary for both individuals and corporations.”
Balter says that GitHub plans to expand its efforts to encourage project maintainers to explicitly choose open source licenses. Look for more analysis from them of how developers on GitHub are using open source licenses and more tools to make the process easier. Ultimately, Balter said, that’s what the developer community wants. “The biggest feedback we got from the open source community was appreciation, and the idea that we need more companies like GitHub supporting open source; to show that it's not a bunch of hobbyists.”