Most FOSS leaders came into prominence during the 1980s and 90s and are now approaching, or have passed, the age when most people retire. Are free software organizations ready for the change that appears to be just around the corner?
Richard Stallman’s recent resignation raises issues that nobody likes to talk about: what happens when the current leaders of free software need to be replaced? What mechanisms are in place to find a new leader? Who, if anyone, should succeed? What provisions should be made? Answers to these questions are especially needed in projects where the idea of the Benevolent Dictator for Life prevail. Moreover, in coming years the urgency can only increase.
Stallman’s resignation was abrupt, but sooner or later these questions were going to be raised. Stallman himself is 66, one year past traditional retirement age, and has had an injured tendon in his leg since 1998. True, until the scandal proceeding his resignation, he had no reason to retire and probably no wish to. Yet sometime in the next decade he might well have stepped down for one reason or the other, perhaps for ill-health.
The same is true for many other early free software leaders. Eben Moglen is 60. Eric S. Raymond is 62. Tim O’Reilly is 65, Lawrence Lessig, Michael Widenius, and Jeremy Allison 57. At 50, Linus Torvalds is a comparative youngster. Yet all these leaders began their work thirty years ago or more, and boredom or accidents can happen to anyone. A few years from now, the problem is likely to be immediate and no longer theoretical. If an outbreak of retirements happens, how will free software cope?
Smooth Transitions, and Rough
In some cases, the question has already been answered. From the start, Eben Moglen described the Software Freedom Law Center as his plan to produce his successors, in which it appears to have succeeded. Similarly, a number of projects, including Debian and Fedora, have regular leadership elections. Debian’s elections are particularly well-defined, with detailed written guidelines that include using the Condorcet method of tallying votes, in which each candidate is compared against every other. Should either Debian or Fedora lose a leader unexpectedly, both projects would likely continue with a minimum of chaos, at least in the short term.
In fact, Debian faced a related crisis in March 2019, when no one responded to the first call for nominations. Although the situation had never happened before, there was never any doubt of how to respond. Another call for nominations went out, and the elections were eventually held as they were supposed to. However, had no candidates come forward, the speculation was that the outgoing Debian Project Leader would not have continued in office because there was no provision for him to do so. In the short term, Debian has enough officials that it could continue without a leader, although without one, duties such as budgeting could not be carried out. Until a new leader could be elected — perhaps by a method determined by a policy resolution among project members — decisions would have been made by the consensus of the chair of the technical committee and the project secretary. There would have been a period of confusion, but it would have ended with a democratically elected leader.
However, few free software projects are as organized as Debian, which takes its political practices as seriously as its licenses. Many projects have been set up with a BDFL, with little thought for succession. Given that many projects were originally set up by developers in their 20s or 30s, perhaps that is unsurprising. After all, how many people that age think of what will happen when they are no longer around? Yet as leaders age, such lack of foresight begins to loom large.
A case in point is the Linux kernel. Should Torvalds no longer be leader, in the short term one of his lieutenants (the developers to whom high-level responsibilities have been assigned) might take over the leadership. Greg Kroah-Hartman, who maintains the stable branch and speaks publicly almost as often as Torvalds, might be the most likely replacement, assuming he wants the position and could hand off his current responsibilities.
The problem is, there is no established mechanism for replacing the leader of the kernel project, perhaps because there has never been a need for one. Would kernel developers insist on a voice in the process? Would the Linux Foundation? After all, it sponsors kernel development and has Torvalds and other prominent contributors on salary. And if the Linux Foundation did intervene, would the rest of the free software community object? Like Debian, the kernel would likely continue in the short term, but would it do so at the cost of long range development? At least a short period of confusion seems certain.
And what about the Free Software Foundation (FSF) itself now that Stallman is no longer involved? Like Torvalds, Stallman had at least the appearance of a BDFL. Yet three months after Stallman’s resignation, the FSF seems in no hurry to replace him, although perhaps a search is going on behind the scenes. If, as I have often suggested, the daily operations of the FSF have been going on for several years under the direction of the executive director, replacement of the president may seem unnecessary, or at least not urgent. Even more so than Debian, the FSF seems able to weather the loss of someone in its top position. It might actually prefer to leave the presidency unfilled, or have a new president begin with a set of reforms in place.
After Me, the Flood
These sorts of difficulties seem partly the result of free software’s long standing traditions. Free software has often cultivated a certain informality, as exemplified by the prevalance of the BDFL. When the emphasis is on the code, details like who is in charge can be ignored as unimportant. Also, no matter what official positions exist, in many projects the true power tends to lie in the consensus of active contributors.
Jokes about BDFL can hide a serious problem. Dictatorships, benevolent or not, tend not to look ahead — “après moi le déluge,” as Louis XV of France is supposed to have said (“after me, the flood”). But although such an attitude can work in the short term, it can also mean that a change of leadership means chaos. Even the existence of a charitible foundation may not be enough to mitigate the problem. It may be no accident that a politically radical organization like Debian, one that works hard to prevent a dictatorship, benevolent or not, emerges as one of the projects best-equipped for a change of leadership.
What about your project? Does it have an accepted method for changing leaders at regular intervals? Are there any notes to help orient new leaders? If not, then perhaps it is time to plan for the inevitable before it happens.