Maddeningly, I have come out the loser in my dispute with the policy researcher. Community building emerges as the most popular reason, with 1,076 votes (51.0 on the adjusted scale), whereas personal growth earned only 1,070 votes (50.6 on the adjusted scale).
Speaking seriously, the documentation survey has proven its value by demonstrating strongly that software users care about the communities that they feel they are a part of. Not only was community building the top reason, but the closely related reason of mutual aid was third, and gratitude fourth. [Exact results in tabular form]
Another chart (Figure 3) comparing the adjusted results for each of the eight reasons shows how powerful the community motivation is. The legend in the chart shows the reasons, ranked in order of importance. Uniquely, not only does community building come out at the top, but it's the only reason that more people rated at the highest level (4, extremely important) than the next higher level (3, important). Such results uncontestably affirm the contributors' attachment to their community, in whatever way they define community (we'll look a bit at that at the end of the article).
Figure 3. Reasons ranked by importance
The personal benefits of learning through teaching emerge as a top-ranking motivation, too. But reputation--that intricate mesh of information and relationships that lies at the center of so much research--falls to the sixth or seventh rank.
What does this say about suggestions by researchers that participation on forums could be bolstered by ranking contributors and somehow honoring those who make the best contributions? When the MSDN wiki began, for instance, the wiki's managers explained to me how they were planning to highlight experts they had come to know in the user community and within Microsoft, to let participants on the wiki know that contributions by these people were worth special attention. The Slashdot-style "karma" rating system offers another model for rewarding contributors.
The documentation survey shows that reputation is not one of the top motivations for contributing. But I wouldn't abandon efforts to provide technical and social support for reputation building. There are many reasons it still might be worthwhile:
Reputation building is less important than other factors, but it's not unimportant. It got 833 votes, a comfortable 75% of the top-ranking reason.
Rating systems help participants spend their time more wisely, by choosing what to read and how seriously to take claims. In other words, they have a benefit to the whole community, not just the person earning the reputation.
Reputation is not the top vote-getter across all participants, but for all we know, it might have been the top vote-getter across the most knowledgeable and articulate participants--the participants we most want to encourage. We won't know whether this is true unless we run a similar survey among a carefully selected group of contributors we consider knowledgeable and articulate. But it makes senses to me that someone knowledgeable and articulate has more demands on his or her time than other participants on a forum, and would also have opportunities to exploit a good reputation, making it more attractive.
Were simple and effective reputation-building mechanisms implemented, many people might start to take advantage of them. This premise lies behind the myriad systems for endorsements in social networks.
In regard to the third point made in this list, it would have been interesting to correlate responses with the quality of contributions. But I couldn't think of any effective way to ask respondents, "Are your contributions any good?" Nor was it ethical or feasible to seek out documentation from each respondent and judge its quality.
Nevertheless, the survey shows that participation is driven by many factors that don't bear immediate rewards. Responses to the "other factors" question, described later, demonstrate this.
As mentioned earlier, some people believe that participants on free software projects stress different motivations from participants on proprietary projects. I have found that people are people everywhere, and that generosity and empathy abound in all demographics. So I decided I would debunk the idea that something was special about free software projects by comparing the reasons given by free software contributors and proprietary software contributors.
My greatest disappointment in this survey is that only 13 people claimed to contribute documentation on proprietary products, and I had to eliminate 4 from my statistical results because they contribute to both proprietary and free projects. (You can't compare two samples if someone belongs to both.)
I couldn't in good conscience claim that nine samples were representative of all people who contribute to documentation about proprietary projects. But just for fun, I decided to create some charts (Figures 4 and 5). And I was a bit surprised by the results.
Figure 4. Free versus proprietary, raw results
Figure 5. Free versus proprietary, adjusted results
The free software contributors ranked mutual aid and community noticeably higher than the nine proprietary contributors--precisely the motivations where a difference would be predicted by the hypothesis that free software is different. Given the crudity of the four-point scale in my questions, and the tiny size of the proprietary sample, I can't make any claims about differences. I merely point it out as a suggestive coincidence.