At parties, people no longer give me a blank stare when I tell them I write free software. "Oh, yes, open source—like Linux?" they say. I nod eagerly in agreement. "Yes, exactly! That's what I do." It's nice not to be completely fringe anymore. In the past, the next question was usually fairly predictable: "How do you make money doing that?" To answer, I'd summarize the economics of open source: that there are organizations in whose interest it is to have certain software exist, but that they don't need to sell copies, they just want to make sure the software is available and maintained, as a tool instead of a commodity. Lately, however, the next question has not always been about money. The business case for open source softwareThe terms "open source" and "free" are essentially synonymous in this context; they are discussed more in in . is no longer so mysterious, and many non-programmers already understand—or at least are not surprised—that there are people employed at it full time. Instead, the question I have been hearing more and more often is "Oh, how does that work?" I didn't have a satisfactory answer ready, and the harder I tried to come up with one, the more I realized how complex a topic it really is. Running a free software project is not exactly like running a business (imagine having to constantly negotiate the nature of your product with a group of volunteers, most of whom you've never met!). Nor, for various reasons, is it exactly like running a traditional non-profit organization, nor a government. It has similarities to all these things, but I have slowly come to the conclusion that free software is sui generis. There are many things with which it can be usefully compared, but none with which it can be equated. Indeed, even the assumption that free software projects can be "run" is a stretch. A free software project can be started, and it can be influenced by interested parties, often quite strongly. But its assets cannot be made the property of any single owner, and as long as there are people somewhere—anywhere—interested in continuing it, it cannot be unilaterally shut down. Everyone has infinite power; everyone has no power. It makes for an interesting dynamic. That is why I wanted to write this book. Free software projects have evolved a distinct culture, an ethos in which the liberty to make the software do anything one wants is a central tenet, and yet the result of this liberty is not a scattering of individuals each going their own separate way with the code, but enthusiastic collaboration. Indeed, competence at cooperation itself is one of the most highly valued skills in free software. To manage these projects is to engage in a kind of hypertrophied cooperation, where one's ability not only to work with others but to come up with new ways of working together can result in tangible benefits to the software. This book attempts to describe the techniques by which this may be done. It is by no means complete, but it is at least a beginning. Good free software is a worthy goal in itself, and I hope that readers who come looking for ways to achieve it will be satisfied with what they find here. But beyond that I also hope to convey something of the sheer pleasure to be had from working with a motivated team of open source developers, and from interacting with users in the wonderfully direct way that open source encourages. Participating in a successful free software project is fun, and ultimately that's what keeps the whole system going. This book is meant for software developers and managers who are considering starting an open source project, or who have started one and are wondering what to do now. It should also be helpful for people who just want to participate in an open source project but have never done so before. The reader need not be a programmer, but should know basic software engineering concepts such as source code, compilers, and patches. Prior experience with open source software, as either a user or a developer, is not necessary. Those who have worked in free software projects before will probably find at least some parts of the book a bit obvious, and may want to skip those sections. Because there's such a potentially wide range of audience experience, I've made an effort to label sections clearly, and to say when something can be skipped by those already familiar with the material. Much of the raw material for this book came from five years of working with the Subversion project (). Subversion is an open source version control system, written from scratch, and intended to replace CVS as the de facto version control system of choice in the open source community. The project was started by my employer, CollabNet (), in early 2000, and thank goodness CollabNet understood right from the start how to run it as a truly collaborative, distributed effort. We got a lot of volunteer developer buy-in early on; today there are 50-some developers on the project, of whom only a few are CollabNet employees. Subversion is in many ways a classic example of an open source project, and I ended up drawing on it more heavily than I originally expected. This was partly a matter of convenience: whenever I needed an example of a particular phenomenon, I could usually call one up from Subversion right off the top of my head. But it was also a matter of verification. Although I am involved in other free software projects to varying degrees, and talk to friends and acquaintances involved in many more, one quickly realizes when writing for print that all assertions need to be fact-checked. I didn't want to make statements about events in other projects based only on what I could read in their public mailing list archives. If someone were to try that with Subversion, I knew, she'd be right about half the time and wrong the other half. So when drawing inspiration or examples from a project with which I didn't have direct experience, I tried to first talk to an informant there, someone I could trust to explain what was really going on. Subversion has been my job for the last 5 years, but I've been involved in free software for 12. Other projects that influenced this book include: The GNU Emacs text editor project at the Free Software Foundation, in which I maintain a few small packages. Concurrent Versions System (CVS), which I worked on intensely in 1994–1995 with Jim Blandy, but have been involved with only intermittently since. The collection of open source projects known as the Apache Software Foundation, especially the Apache Portable Runtime (APR) and Apache HTTP Server. OpenOffice.org, the Berkeley Database from Sleepycat, and MySQL Database; I have not been involved with these projects personally, but have observed them and, in some cases, talked to people there. GNU Debugger (GDB) (likewise). The Debian Project (likewise). This is not a complete list, of course. Like most open source programmers, I keep loose tabs on many different projects, just to have a sense of the general state of things. I won't name all of them here, but they are mentioned in the text where appropriate. This book took four times longer to write than I thought it would, and for much of that time felt rather like a grand piano suspended above my head wherever I went. Without help from many people, I would not have been able to complete it while staying sane. Andy Oram, my editor at O'Reilly, was a writer's dream. Aside from knowing the field intimately (he suggested many of the topics), he has the rare gift of knowing what one meant to say and helping one find the right way to say it. It has been an honor to work with him. Thanks also to Chuck Toporek for steering this proposal to Andy right away. Brian Fitzpatrick reviewed almost all of the material as I wrote it, which not only made the book better, but kept me writing when I wanted to be anywhere in the world but in front of the computer. Ben Collins-Sussman and Mike Pilato also checked up on progress, and were always happy to discuss—sometimes at length—whatever topic I was trying to cover that week. They also noticed when I slowed down, and gently nagged when necessary. Thanks, guys. Biella Coleman was writing her dissertation at the same time I was writing this book. She knows what it means to sit down and write every day, and provided an inspiring example as well as a sympathetic ear. She also has a fascinating anthropologist's-eye view of the free software movement, giving both ideas and references that I was able use in the book. Alex Golub—another anthropologist with one foot in the free software world, and also finishing his dissertation at the same time—was exceptionally supportive early on, which helped a great deal. Micah Anderson somehow never seemed too oppressed by his own writing gig, which was inspiring in a sick, envy-generating sort of way, but he was ever ready with friendship, conversation, and (on at least one occasion) technical support. Thanks, Micah! Jon Trowbridge and Sander Striker gave both encouragement and concrete help—their broad experience in free software provided material I couldn't have gotten any other way. Thanks to Greg Stein not only for friendship and well-timed encouragement, but for showing the Subversion project how important regular code review is in building a programming community. Thanks also to Brian Behlendorf, who tactfully drummed into our heads the importance of having discussions publicly; I hope that principle is reflected throughout this book. Thanks to Benjamin "Mako" Hill and Seth Schoen, for various conversations about free software and its politics; to Zack Urlocker and Louis Suarez-Potts for taking time out of their busy schedules to be interviewed; to Shane on the Slashcode list for allowing his post to be quoted; and to Haggen So for his enormously helpful comparison of canned hosting sites. Thanks to Alla Dekhtyar, Polina, and Sonya for their unflagging and patient encouragement. I'm very glad that I will no longer have to end (or rather, try unsuccessfully to end) our evenings early to go home and work on "The Book." Thanks to Jack Repenning for friendship, conversation, and a stubborn refusal to ever accept an easy wrong analysis when a harder right one is available. I hope that some of his long experience with both software development and the software industry rubbed off on this book. CollabNet was exceptionally generous in allowing me a flexible schedule to write, and didn't complain when it went on far longer than originally planned. I don't know all the intricacies of how management arrives at such decisions, but I suspect Sandhya Klute, and later Mahesh Murthy, had something to do with it—my thanks to them both. The entire Subversion development team has been an inspiration for the past five years, and much of what is in this book I learned from working with them. I won't thank them all by name here, because there are too many, but I implore any reader who runs into a Subversion committer to immediately buy that committer the drink of his choice—I certainly plan to. Many times I ranted to Rachel Scollon about the state of the book; she was always willing to listen, and somehow managed to make the problems seem smaller than before we talked. That helped a lot—thanks. Thanks (again) to Noel Taylor, who must surely have wondered why I wanted to write another book given how much I complained the last time, but whose friendship and leadership of Golosá helped keep music and good fellowship in my life even in the busiest times. Thanks also to Matthew Dean and Dorothea Samtleben, friends and long-suffering musical partners, who were very understanding as my excuses for not practicing piled up. Megan Jennings was constantly supportive, and genuinely interested in the topic even though it was unfamiliar to her—a great tonic for an insecure writer. Thanks, pal! I had four knowledgeable and diligent reviewers for this book: Yoav Shapira, Andrew Stellman, Davanum Srinivas, and Ben Hyde. If I had been able to incorporate all of their excellent suggestions, this would be a better book. As it was, time constraints forced me to pick and choose, but the improvements were still significant. Any errors that remain are entirely my own. My parents, Frances and Henry, were wonderfully supportive as always, and as this book is less technical than the previous one, I hope they'll find it somewhat more readable. Finally, I would like to thank the dedicatees, Karen Underhill and Jim Blandy. Karen's friendship and understanding have meant everything to me, not only during the writing of this book but for the last seven years. I simply would not have finished without her help. Likewise for Jim, a true friend and a hacker's hacker, who first taught me about free software, much as a bird might teach an airplane about flying. The thoughts and opinions expressed in this book are my own. They do not necessarily represent the views of CollabNet or of the Subversion project.