In its first news release, Dunc-Tank describes itself as an "experiment" and its members as "an independent group of developers, users, and supporters of Debian." In fact, however, Dunc-Tank members are among the most active and prominent Debian developers. The news release announcing the group lists both Anthony Towns, the Debian Leader, and Steve McIntyre, assistant to the Debian Leader, among its supporters, as well as Raphaël Hertzog, Joey Hess, and Ted Ts'o -- all of whose names should be familiar to anyone who follows the Debian project. Ts'o is also a long-time kernel developer.
Although the cynical might say that Dunc-Tank represents the continuation of official Debian business by unofficial means, the description of the group is apparently intended to acknowledge that its members are acting on their own initiative, and that no consensus on their ideas has been reached within the project itself.
Dunc-Tank's first goal is to fund Debian's two release managers. The plan is to employ Steve Langasek full-time during October and Andreas Barth full-time during November. Enough pledges have already been made to employ Langasek as planned, according to Ts'o.
The first goal was chosen largely for practical reasons. Hertzog says that some developer favored a project to resolve the problem of non-free firmware in Debian's main repository, but that Dunc-Tank's members saw meeting the release date as more important to a majority of Debian developers. The upcoming release still has more than 200 critical bugs, according to Hertzog, and he suggests that, without a major innovation such as Dunc-Tank, Debian cannot hope to meet its release deadline. "A large majority of the project wants a timely release," Hertzog asserts.
At the same time, Dunc-Tank's proposal is also intended to address widespread concerns about the notorious slowness of the Debian release cycle. Although this slowness may be part of the inevitable consequence of Debian's support of multiple hardware architectures, how to mitigate it has been a subject of numerous debates within Debian almost since its beginning. Recently, Debian has come up with rules that each hardware port has to follow, and expanded the team of release managers with a number of assistants, but the concern -- to say nothing of the perception -- remains.
For now, Dunc-Tank is not actually accepting funds. "Pending setting up formal mechanisms to receive and disburse funds, we have not asked any of the donors to send us money just yet," Ts'o says. "That will hopefully be coming very soon."
A discussion of Dunc-Tank is on the agenda for the October meeting of Software in the Public Interest, the non-profit organization whose task includes collecting donations to Debian.
Extrinsic vs. intrinsic motivations
Public criticism of Dunc-Tank's rationale began the same day that the group was announced. In a blog entry that was widely distributed to a number of discussion groups, Debian developer Lucas Nussbaum cites research by Luis Villa of GNOME that suggests that "sometimes, paying volunteers decreases the overall participation" in projects.
In his blog, Villa summarizes the results of his online research about the potential problems with introducing payments into a group of volunteers who have previously been motivated by "their own intrinsic goodness (i.e., for reasons other than payment)." As example, he mentions the classic case of an Israeli day care center that found that, when parents were fined for picking up their children late, more late pick-ups occurred than before. Moreover, when the fines were canceled, the number of late pick-ups did not decrease to its original level. The literature suggests that the reason for this phenomenon is that introducing payment means reducing volunteers' "sense of self-determination and self-esteem" as well as other motivations such as volunteering "because they want to help others, or because it is fun."
Villa tells NewsForge that he has "no personal experience in the GNOME context that I could draw concrete conclusions from," suggesting that his blog entry is academic and not necessarily advocating any particular action. Nonetheless, Nussbaum uses Villa's research to question not only Dunc-Tank, but also Google's Summer of Code and the GNOME's Women's Summer Outreach Program, two other programs to provide payment for free software developers.
T'so suggests Nussbaum's comments may be an over-simplification, noting that many charities include both volunteer and paid workers, and that their paid workers generally receive less than they would if they did the same work for a commercial company. "If money were among anybody's primary motivators," he says, then "they probably wouldn't be accepting a grant from Dunc-Tank; they could probably make more money by applying for a job with Google -- or Microsoft."
"Furthermore," T'so says, drawing on his 15 years of kernel development, "I would note that most of the core kernel developers have made the transition from volunteer to paid work, and I can't say that I've seen any negative efforts within the Linux kernel community as a result of that change."
Hertzog says, "We're quite aware of potential drawbacks," but suggests that providing funding can also increase the internal motivations of Debian developers. "Many Debian developers are motivated to work when things evolve," he says, addressing the frustration that many have expressed about the project's long release cycle, "and releasing [on-time] in December is certainly the best reward for all of us. And we want to prove that we can release on time and still have the same quality release that people are used to having."
Hertzog describes the payment of the release manager as a "first experiment." Once this experiment is concluded, T'so suggests, Dunc-Tank's board and supporters will be in a better position to evaluate the concept of paying developers.
Hertzog is already in the process of developing suggestions of how Dunc-Tank might operate on a permanent basis. "This method will most probably involve all the Debian developers or at least give them the possibility of being involved," he says. While stressing that his ideas have not yet been discussed, much less accepted by Dunc-Tank or the greater Debian community, Hertzog favors a wiki on which any Debian developer could propose a project for funding. Donors could then choose which projects they wanted to fund, or, alternatively, all Debian developers would have the chance to vote on which projects received funding.
In addition, Hertzog would like to develop specifications for proposals to ensure that they are realistic, and possibly a board elected by Debian developers to handle the inevitable problems, such as money pledged to a project that is withdrawn or developed for free. T'so suggests that a "formal set of bylaws" might also be in order to replace Dunc-Tanks' current policy of seeking consensus among board members.
These are ambitious plans, but Hertzog admits that, for now, they will have to wait. For now, Dunc-Tank's goal is to succeed in its first experiment. After that, he says, "The discussion will continue within Debian, and we hope that this first experiment will bring some real experience into the discussion and that we'll be able to have a saner discussion. Otherwise, you can only rely on academic studies, like the one mentioned in Lucas Nussbaum's blog, and you will never reach a conclusion only by studying these documents."
Bruce Byfield is a course designer and instructor, and a computer journalist who writes regularly for NewsForge, Linux.com and IT Manager's Journal.