Editor's Note: Andy Oram reports on the possible implications of a recent study that explored the reasons behind the widespread use of, and support for, free and open source software.
A major research project under the name Free/Libre and Open Source Software: Survey and Study, or FLOSS, recently explored reasons behind the widespread use of, and support for, free and open source software. This is, to my knowledge, the first large-scale, rigorous study concerning any aspect of free software. It involves interviews with thousands of developers and hundreds of businesses, with carefully-chosen questions and a correlation of results.
Equally significant is the funder of the study: the European Union. These government representatives have displayed a growing curiosity about, and sympathy for, the free software phenomenon; see, for instance, the recent announcement concerning OpenEvidence, a project in digital certificates and signatures. The EU is an excellent body to sponsor a sympathetic but demanding inquiry into the purposes and processes of free software development. (The study was carried out for the EU by Berlecon Research GmbH and the University of Maastricht. The business survey involved only European participants, but the developer survey was worldwide.)
This article, like most of mine, will involve a modest portion of summary accompanied by an ample serving of my own impressions and analysis.
The burning question many people bring to this survey -- whether friend or foe -- is: "Why would organizations choose to use free software?" We know that the "free as in beer" aspect of free software appeals to underdeveloped countries (see recent news items from Peru and Venezuela, along with China's Red Flag Linux). But in more affluent Europe, price is not the issue. Over and over in the FLOSS study, organizations place cost savings low on their list of reasons for choosing free software. Ideological reasons will be discussed later in this article. Overwhelmingly, the highest ranking reasons for using free software center on quality:
Admittedly, "better price-to-performance ratio" turns up with a high rating. But the survey notes that true cost comparisons are hard to make with any confidence. Organizations also like the absence or low burden of license fees, but I'm not sure that this is a cost issue. Organizations might simply want to avoid the pain in the ass of predicting needs, negotiating with the vendor, dealing with malfunctioning license servers, and so on.
Still, quality issues clearly trump cost issues in the FLOSS survey. This is powerful ammunition for activists fighting the old misconception, still far too prevalent, that free software is a poor man's consolation for the lack of funds to buy really good software.
Looking deeper, I find another lesson in this confidence expressed by businesses and nonprofits. The relatively slow pace of development in free software is one of its strengths. Proprietary software companies have earned a terrible reputation over many decades for shoveling in check-off boxes as fast as marketing representatives can think them up. Bugs inevitably abound. Users complain about bloatware and features that merely get in their way, as well as trying to fix bugs by upgrading to the next feature release and getting more bugs.
Perhaps this is why MySQL is gaining market share, even though it started off quite feature-poor; MySQL AB has taken its sweet time adding such basics as transaction support and encrypted data transfers. What they offer is rock-solid reliability (along with the performance that one achieves by leaving out computationally-heavy functionality).
Security, which is now on administrators' minds more than ever, has always been understood to be a function of stability and code quality. Modern Windows systems have a number of security features -- ACLs and encrypted filesystems, for example -- that Linux and the BSDs lack or require special patches for. But security features are not what most users are looking for; they want security, plain and simple. Linux and the BSDs offer that more reliably (unless Bill Gates's recent conversion to the creed of high security is matched by growing adherents throughout Microsoft).
We must also remember that new features do not change user behavior the moment they're released; they take some time to percolate through the ground and up the root systems of the user community. In particular, programmers are people, too. They require time to learn about new features, recognize their benefits (if any), and upgrade their applications. Each delay reduces the utility of a feature upgrade.
Please understand that I believe in evolution. But the changes that make people feel an urgent need to upgrade are those that radically reform their jobs and their ways of interacting, such as graphical interfaces, the Web, and cross-platform code development. These sorts of innovation can occur in both free and proprietary software. In contrast, incremental change is not a big selling point.
I have not yet discussed ease of use, a measure where free software presumably doesn't come off so well -- at least for new and nonprofessional users. The FLOSS study addressed this in a couple criteria, especially "Cost savings regarding training and introduction of users," which predictably came out low as a reason for using free software.
In general, free software has still not achieved the widespread familiarity of Microsoft software. In a recent analysis regarding the elusive "Total Cost of Ownership" (TCO) measure, the analysts noted the familiarity of Microsoft software as one of its main advantages. More exposure to free software can close the gap.
The FLOSS study itself throws up its hands when dealing with TCO. They report that companies "were generally unable to provide even rough estimates about the monetary value derived from using open source software," even concerning "simple questions like license fee savings or hardware cost savings."
A sense of moral imperative concerning information freedom motivates a lot of free software developers. Nearly all of the programmers surveyed understood the philosophical difference between "free software" and "open source," as articulated by Richard M. Stallman and the Free Software Foundation. Rishab Aiyer Ghosh, chief writer of the FLOSS study, finds in the developer results several thought-provoking observations:
Although organizations trying to commercialize free software (such as my own company, O'Reilly & Associates) promote the term "open source," more of the developers identify with the "free software" side than the "open source" side (48 percent versus 38 percent).
Most "free software" developers were content to coexist with the "open source" label, but 38 percent of them -- 18 percent of all developers in the survey -- took a stronger position and refused to identify with the "open source" community. Furthermore, 38 percent of all developers agreed with the statement that software should not be proprietary.
Participation in free software projects tends to sharpen developers' philosophical and political concerns. Belief that "software should not be a proprietary good" and a desire to "limit the power of large software companies" were larger motivations for developers continuing to work on projects than they had been in attracting these developers in the first place.
It would be interesting to imagine what would happen to the open source movement if proprietary software projects dried up. The demographic part of the FLOSS survey showed that the majority of contributors to free software are professional programmers. Most of them work on open source projects as a hobby, often for just a couple of hours a week. In short, proprietary software provides an income for fully half of the people who work on open source software. But perhaps many of these programmers work on in-house software rather than on commercially-sold software (a point made by Eric Raymond in his essay "The Magic Cauldron," published in The Cathedral & the Bazaar).
The researchers conducting the FLOSS study could not detect much of a concern for moral issues in the organizations using free software (whether commercial or nonprofit). But my interpretation of the statistics shows more of a groundswell of concern among customers than the researchers allowed for.
The percentage of organizations that agreed with the statement "We prefer using open source software -- that's part of our company policy" was low (19 percent), but still pretty impressive, when you consider how rarely businesses take stands on questions outside of purely-practical considerations. Similarly, I consider it an achievement when 35 percent agree with the statement "By using open source software we want to support the open source community."
On the other hand, I do not feel a sense of confidence that companies take a moral position by allowing their developers to contribute to open source projects during work hours. While 36 percent say they allow it, they could be doing so simply for pragmatic reasons.
The FLOSS researchers correlated organizations' responses with many demographic characteristics, but did not correlate the responses to particular statements with responses to other statements. In any survey, it is hard to draw conclusions about beliefs and motives. So both the researchers and the readers of the FLOSS survey are left guessing about the meaning behind many responses, just as we are left guessing why free software programmers (a whopping 48 percent) prefer the Debian GNU/Linux distribution far more than any other operating system.
What about the availability of source code? Open source's defining characteristic, it's been touted as an unbeatable competitive advantage. Here the researchers find very little interest among users. But once again, I think they underestimate the importance of the interest they do find. Some 12 percent of organizations rate the openness of source code "very important" (the percentage rises to 19 percent for desktop software and Web sites) and an even greater percentage rates it lower but still "important." I think that 12 to 19 percent of a user base is quite enough to create a supportive community, and is competitively significant, as well.
It's no surprise that more companies are interested in modifying their desktop software (which is currently not too stable) and Web site software than their operating system or database, just as a typical home owner is more comfortable rehanging a window than ripping out the plumbing.
The FLOSS study spends a good deal of time looking at what kind of person writes free software, and why he (because 99 percent of these developers are male) chooses to do so. The results tend to validate another recent study made by the Boston Consulting Group, wherever the two studies overlap.
As mentioned earlier in this article, the majority of contributors are professional programmers. Many can boast an academic computer science background. They tend to be young, but (contrary to popular belief in some quarters) students do not predominate. A good number of developers are in their thirties. These are heavily represented among project leaders. But most projects have only one or two contributors.
How about motivations? Non-monetary rewards, such as respect in the community, are important. Both the Boston Consulting Group and the FLOSS surveys also found that one of the most popular reasons for joining a project was to learn new skills.
And indeed, the investment of time pays off. A slight majority -- if one includes indirect employment, such as support -- earn some money from free software. Perhaps it is not so remarkable that a majority of contributors are convinced they take out of the projects more than they give.
The FLOSS report is long but quite readable, and I recommend exploring the subtleties of its reasoning for yourself. While the results will interest anybody who thinks about the prospects for the growth and spread of free software, the FLOSS project is ultimately about much more. Among the goals stated in the survey's overview are:
"A better realization of political aims [presumably those of the EU] in the context of open source software" (p. 8)
"Identifying the impact of and recommending changes in government policy and regulatory environments with regards to OS/FS" (p. 3)
Drawing "broader" conclusions about "non-monetary and trans-monetary activity in the information society, beyond the domain of OS/FS" (p. 3)
The researchers are even mining the source code on a huge number of open source projects to trace the connections between the projects and their authors. Although these efforts are tentative, someday they may help to firm up our understanding of the unique opportunities for education, code reuse, and synergistic evolution presented when programmers publish source code to the whole world.
Andy Oram is an editor for O'Reilly Media, specializing in Linux and free software books, and a member of Computer Professionals for Social Responsibility. His web site is www.praxagora.com/andyo.
Read more Platform Independent columns.
Return to the O'Reilly Network.
Copyright © 2009 O'Reilly Media, Inc.