Tuesday, July 10, 2007

Open Source Community Interaction: Is More Less?

I have a friend who needs our help. In fact, there are potentially disturbing ramifications in the trend mentioned below, so *we* may need our help.

Here's the deal: my friend is a research assistant at McGill University in Quebec. He is working with his professor with regard to the following topic:
"Open source software (OSS) developers who interact MORE with leaders (administrators) apparently contribute LESS (e.g., write less codes). We need more insights into this finding. Basically, we want to know why this is happening. We have speculated that open source software (OSS) communities are public, voluntary-based and self-organized communities. Therefore, if OSS leaders are too controlling, then developers don't contribute as much. This was one of our speculations. I am wondering if any OSS developers agree with this."
Sadly, very few of the developers their research team has contacted have given feedback on this matter. Below is the short list of questions they are asking developers. Note that though the example of contribution was code in the above statement, these questions assume contribution to take any of the following forms:
  • source code
  • documentation
  • answering questions and providing assistance to other users via mail lists, IRC channels, forums, etc.
  • grass-roots marketing
  • other forms of community involvement that contribute positively to an open source project
Okay, so here are the questions:
  1. Does the open source project's leader/lead developer have a huge impact on your contribution?
  2. What reasons would make developers who interact more with core project contributors apparently contribute less to the open source project?
  3. How would you feel if the leaders of the open source projects to which you contributed became too controlling?
I responded when Meral sent me his email with the two examples of Twisted and Python communities, but these guys need many more points of view than a few... and they aren't getting much traction. I feel that this research is important, and when published, could affect how open source projects and contributions to said projects are viewed by a wider population.

If this is also important to you, I encourage you to send an email to meral dot hussein at mail dot mcgill dot ca. Give it the subject of "McGill University Survey" and let 'er rip. The sooner you can do this, the better -- the deadline for the first round data is July 13th.

Technorati Tags: , , , , , ,


  1. I'm not an open-source developer, but I wanted to take a shot at answering anyway. But first, I have a question about their problem statement: are they implying that the "leaders (administrators)" are by definition the "core project contributors", i.e., they are programmers who are contributing the most to the development of the project?

    The reason I ask is because if the "leaders" are NOT writing the most code, and are focusing more on organizing the team and setting goals, then I could easily see some programmers having their feathers ruffled by the feeling of being given orders - and that would, in turn, directly make them contribute less code. I would guess that most open-source developers very much prefer the feeling of equality between all team members, rather than having a single "boss" who hands out orders.

    I don't know, maybe I'm totally off here.

  2. There are two ways I can see this where it is no mystery at all:

    * Within a single project - the people just below the leaders in the project hierarchy are also leaders, just at a lower tier, managing subsystems, debating and evaluating directions, vetting patches. This is similar to any organization, the people at the top are responsible for managing even if it means they do less doing.

    * Across projects - a project would be doing more interaction because its project is less separable, thus project wide code would be produced slower both because of coordination delays and because each participant must spend some of the time they allot to the project on coordination tasks.

    I'd love to see how this study corrects or accounts for these effects, because at it's face it seems like no mystery at all; they contribute less *because* they are busy talking.