The amount of paid developers in project

First of all, I have to make clear that most developers don’t see any difference in general between paid developer and contributor. However, they see differences between the two if it is asked indirectly. I quess the ‘I don’t care’ answer refers to results, where it does not matter who did it as long as job gets done. Also it might refer to who is counted as member of community, both are important and members of (project) developer community. One difference has been visible in interviews that I have conducted during the last spring/summer. Some developers think that the quality of code coming from developers who are participating for ‘just to get paid’ or ‘it’s my job’, is less good. It must be noted, that this kind of thinking might be based on gutt feeling rather than cold facts.

Nevertheless, I have been curious to know what is the amount of paid developers for example in MeeGo and how to get a grasp of it. Of course one could do a survey, but my gutt feeling (yes, not a fact) is that MeeGo community is not mature enough to want much information about itself unless it’s more or less technical. In other words, information retrieved with sociological or anthropology tools are not interesting. At least not yet. That leaves out surveys then. What next? Are there any other ways to get at least a hint of the distribution of paid and contributor developers? In a way defining method to get that information would be a tool to analyze communities in early stages. This conclusion was enough to inspire me to think about other possibilities than surveys. Besides surveys are normally run once a year or in 6 month periods. More realtime data would be better.

Why such information is needed?

For developers point of view it is not. For community managers point of view it might function as a tool to define or adjust communication methods and tools to suite community status/maturity/needs. As it has been said by one of the ‘leading’ community managers in the world, Jono Bacon, fluent communication is important to any successful and ‘alive’ community.[1] It might be a minor thing in your mind, but moment counts. It does make a difference when you ask a question or release some information.[2] Another possible implementation or use case for this information is to target communication to ‘correct’ media. In my interviews (and on some other research/articles) I have noticed that different kind of developers prefer different kind of communication methods or tool. Some prefer to use forums, others mailing list.

Ok, so the tool might be needed by some, that’s enough for now. Let’s take a brief look at the idea. In brief: Open source people, who do not work for the project/goal, will most likely be more active during evenings, weekends and holidays. The paid developers will most likely be the opposite.

Use activity as measure

Determining activity in various areas such as bug reports, commits and discussion, we could get at least a glimpse of the distribution of paid and contributor developers. Using this method does not give exact information, but more like direction. How would that work? Let’s take bug reports as an example. Bug reports are filed into one database (or bugtracker if you prefer that word) and submissions have a lot of information such as date and who filed it. One might do a query to database and see when most of the bugs are reported. Are they reported during weekdays? At what hours of day? The percentage of bug reports submitted during weekdays between 9 to 5 could indicate (roughly) how many of the developers are paid. Then the same analysis would be done on code commits. Of course these two methods do not make any clear distinction between paid and volunteer developers, since one developer can be both. Nevertheless, the above might be part of the solution. If the same analysis is done for discussions and wiki updates, then we might get more accurate result.

There is one small problem with the time based analysis. Ecosystem is ‘alive’ around the clock and around the globe! The analysis tool should be able to identify local time for each operation it uses in analysis.

So far I have discussed only hour and day based analysis. Analysis could and perhaps should be broadened to month and year scale too. In other words, method or tool should be able to analyze data in year period too. In that focus would be to identify holiday months and what happens to activity (in different forms: commits, bug reports) during normal holiday seasons.

If all the above and some more indicators are combined to calculate or at least estimate the distribution of paid and contributor developers, the result might be accurate enough. The idea is still rough and might be used already (did not bother to search if such solution exists already), and it might be a bit goofy. Yet, it might be usefull.

[1] Jono Bacon, The Art of Community, p. 67. http://www.artofcommunityonline.org/downloads/jonobacon-theartofcommunity-1ed.pdf
[2] Strategic release of information on Friday: evidence from earnings announcements. http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.106.9097&rep=rep1&type=pdf

Content is available under 2011, Jarkko Moilanen. CC BY-NC-SA 3.0

Related posts:

1. Developers – refers to what?