The only public information on the process has been the Process Definition document released in January 2006, and whatever can be gleaned indirectly from the rest of the GPLv3 site. Now, as reactions to the second draft are pouring in and the process of writing the third draft begins, a more complete answer is coming clear. NewsForge talked to several of those involved in the process. In particular, Richard Fontana, a lawyer at the Software Freedom Law Center and one of the main writers of the revised license, offered his perspective. Although Fontana considers himself bound by his professional ethics as a lawyer to respect the confidentiality of Richard Stallman and the Free Software Foundation as his clients, his comments offer the first clear indications of the process. From Fontana's description, it is a process that involves unprecedented consultation for a legal document, followed by intense work by a handful of experts to produce a text that they hope will find a consensus in the communities.
Fontana describes the methods for receiving input on the revisions as a mixture of "formality and informality." On the Web page for GPLv3, viewers can see comments made about the current draft and register to add their own. These comments, Fontana says, are analyzed "very carefully and systematically."
At other times, input may be received more informally. For instance, in April at the GPLv3 conference in Brazil, Fontana heard a presentation and approached the speaker afterwards for more details.
However, much of the input comes from the four discussion committees organized at the start of the process. Consisting of more than 130 members of the free and open source software communities, the four committees are organized loosely by constituency:
Committee A: Representatives of major free software projects, such as Perl, Apache, and GNOME.
Committee B: Representatives of major distributors of GPL software, such as MySQL, Cisco Systems, Novell, IBM, and Red Hat. Many of these distributors are major patent holders, and have expertise on patent law.
Committee C: Representatives of major organizational users of free software, including corporations, universities, and governments, mostly from the United States. Committee C also includes experts on legal and policy issues surrounding software.
Committee D: Members of what Fontana calls the "hacker intelligentsia, including Don Armstrong and Benjamin Mako Hill from Debian, and Louis Suarez-Potts of OpenOffice.org.
"None of the four committees has a monolithic makeup," Fontana says. The committees were organized to represent a cross-section of the community and corporate interests in the GPL. Despite the potentially divisive differences in viewpoints, he says that, so far, the committees have "worked fairly well. These groups are not necessarily as far apart as one might think. A committee with a lot of large corporations sometimes expresses similar concerns as a committee made up of free software hackers."
Each committee has its own organization, forming its own sub-committees as members think is necessary. "This has been done particularly by Committee C," Fontana says, adding, "We are currently looking to ways of facilitating cross-committee dialogue on particularly difficult issues such as DRM and patents."
Other committees and individual members have taken on the tasks of recruiting new members and acting as ombudsmen for others. Benjamin Mako Hill, for example, says, "I've been approached by probably half a dozen people. Many had issues they wanted resolved, and I helped them file comments. A few others wanted to be on a committee, and that was possible as well."
Both as a team and as individuals, committee members raise issues and consult with the drafters. For the second draft, Fontana describes the process as "leisurely" for some weeks, but, as the time to publish the draft approached, the drafters asked the committees "to wrap up their work during that initial phase and provide us with the issues they had identified so that we could consider them."
Writing the draft
In contrast to the consultation process, the actual writing of a draft is private. "The drafting itself is done internally," Fontana says, "and it involves privileged discussions between Mr. Stallman and his lawyers. So we do not involve the committees in that."
However, Fontana did reveal that the writing of GPLv3 is being done primarily by Richard Stallman and Eben Moglen -- both veterans of previous versions of the GPL -- and himself.
In addition, David Turner, the former GPL compliance engineer at the Free Software Foundation, has also contributed "a fair amount" to discussions of the process. After five years of experience of working with the existing license, Turner says half-seriously, "I used to have most of version 2 memorized. Then version 3 came along, and I can't keep track of which is which."
Because the members of the drafting committee travel frequently, especially Stallman and Moglen, some of the work is done by telephone or email. However, according to Fontana, it is in face-to-face discussion that "we get a lot of the important work done." For himself, Fontana describes the work as "pretty much a full-time job" since he started at the Software Freedom Law Center in December. "I do some other work," he says, "but the bulk of my time is spent working for the FSF on GPLv3."
The drafters feel themselves bound by the Process Definition document, and writing is generally "a collaboration" between members. However, once the drafting begins, the drafters do not send their work back to the committees for input. In addition, Fontana says, "Richard Stallman is the final arbiter. Eben and I, being lawyers, are supplying him with legal advice during the process."
As work on the third draft begins, Fontana sees work on GPLv3 as entering a more intense phase. "We hope that the second discussion draft resolved a lot of the less thorny issues involved in the license. [Now] we want to focus on the more difficult issues that couldn't be more fully resolved." Judging from feedback in the media, these issues are likely to include the language on patents and digital rights management technologies.
Asked what would happen if no agreements could be reached with major stakeholders on outstanding issues, Fontana says, "We do not plan to delay the process. We plan to hold to the scheduled that we outlined in the GPLv3 process definition document. What we will do is work very hard to build consensus around the difficult issues."
The process developed for GPLv3 may become the template for revisions of other licenses, including the GNU Free Documentation license, which Stallman indicated last year is due for revision after the GPL. However, Fontana sees the process has having even larger implications.
"We're doing something that has not really been done before," he says. "No license has been drafted with the benefit of this level of public scrutiny over this extended period of time. Traditionally, someone writes a license, perhaps with the aid of a lawyer, then makes it public. We're showing a very different way of producing this type of document. And I think this may have relevance beyond license drafting. It may have something to show the world about how similar documents can be drafted, like treaties and standards and other types of legal instruments that affect large constituencies of people with varying viewpoints on many different issues."
Whether the process will have that degree of influence will undoubtedly depend on the degree of consensus it achieves. Meanwhile, as the process passes the halfway point, Fontana says, "I expect that there will be future changes. Discussion is not over."
Bruce Byfield is a course designer and instructor, and a computer journalist who writes regularly for NewsForge, Linux.com and IT Manager's Journal.