Eclipse Ecosystem

A blog devoted to promoting the Eclipse ecosystem

Monday, September 18, 2006

Factors Contributing to the Success of Open Source (Part 1 of 3)

Rizwan Ur Rehman's thesis at Carleton University last year was titled "Factors that Contribute to Open Source Software Project Success". The results were interesting, and I've had a good look at the study and data. Over the next few days I'll blog about the results and some of the theories for the results. If you'd like a copy of the work, drop me an email. My email is my first name dot my last name and I work for

In this part I would like to describe the background and basis for the research.

The first challenge Rizwan faced was defining "success". To me, success would be a healthy ecosystem with lots of commercial and non-commercial activity. This would be pretty difficult to measure, not only to find publicly available data, but because it's not clear how to measure the "health" of an ecosystem. There is hopefully an opportunity for research in this area moving forward.

So, to define "success", Rizwan did what any good OS citizen would - he asked the OS developer community - what does it mean to be a successful open source project? The results were very clear - Downloads and Releases. Therefore, the measure of success for his research was the number of downloads and also the number of releases.

The next challenge was to determine what factors to study. I can think of dozens of factors that may or may not contribute to the success of an open source project, but Rizwan narrowed it down to six:

1 - Number of developers working on the project.

2 - Experience of the developers (in years) working on the project.

3 - If the OS Project is for end users (application software), for sys admins or for developers (developer tools/frameworks, etc). Basically - who is the target audience for the OS Software being created.

4 - If a "commonly used" programming language (sic) is used, where "common" is defined as C, C++, Java or PHP.

5 - Is the OS Project an Application, development or deployment tools, or frameworks and infrastructure.

6 - Type of OSS License used ("Very Restrictive" versus "Mildly Restrictive" versus "Non-restrictive"). Only OSI approved licenses were considered. (rest of this paragraph - sic) - GPL was considered "Very Restrictive". LGPL, EPL, MPL and otheres were considered "Mildly Restrictive" and Apache, BSD, MIT (etc) were consisdered "Non-restrictive".

The study focused on a completely random slection of 350 projects from SourceForge. At the time, there were 100,341 possible projects on SourceForge that could have been selected.

In Part 2 I will list the results from the first part of the research. Of the Six factors, there were four that showed a correlation to a successful project and two showed no correlation. Can you guess what they were? And bonus - can you put the four in order of strength?

Find out in part 2.
- Don


Post a Comment

<< Home