three kinds of open source metrics
07 December 2017
As far as I can tell, there are three kinds of open source metrics.
Impact metrics cover how much value the software creates. Possible good ones include count of projects dependent on this one, mentions of this project in job postings, books, papers, and conference talks, and, of course sales of products that bundle this project.
Contributor reward metrics cover how the software is a positive experience for the people who contribute to it. Job postings are a contributor reward metric as well as an impact metric. Contributor retention metrics and positive results on contributor experience surveys are some other examples.
But impact metrics and contributor reward metrics tend to be harder to collect, or slower-moving, than other kinds of metrics, which I'll lump together as activity metrics. Activity metrics include most of the things you see on open source project dashboards, such as pull request counts, time to respond to bug reports, and many others. Other activity metrics can be the output of natural language processing on project discussions. An example of that is FOSS Heartbeat, which does sentiment analysis, but you could also do other kinds of metrics based on text.
IMHO, the most interesting questions in the open source metrics area are all about: how do you predict impact metrics and contributor reward metrics from activity metrics? Activity metrics are easy to automate, and make a nice-looking dashboard, but there are many activity metrics to choose from—so which ones should you look at?
Which activity metrics are correlated to any impact metrics?
Which activity metrics are correlated to any contributor reward metrics?
Those questions are key to deciding which of the activity metrics to pay attention to. I'm optimistic that we'll be seeing some interesting correlations soon.