COUNTING.RST
Open Source projects and communities, like most things, are full of humans. Humans have egos.
- WE FOCUS ON the CARROT, and not the STICK
- COUNT things in a way to PRAISE many, without CONDEMNING some.
- Count all the things
- Don't report on just coding
- Report on bugs filed
- Report on reviews
- Report on meetups hosted
- Report on email threads contributed to
- Report on speeches given
- Report on deployments achieved
- Count all the repos (http://git.openstack.org/cgit)
- Don't count LOC except as a last resort
- Count number of modules
- Count number of tests
- Count change to cyclomatic complexity
- Count within a module, not across modules
- Count within a language, not across languages
- Count company absolutes, if you have to, but favor counting deltas
- Increase in contributors (by percentage)
- Increase in rate of contributions (by percentage)
- Count WITH, not VERSUS
- Favor INCLUDING more things in the counting
Projects that count with these principles in mind could be referred to as "Open Count Software" or something.
I like the specifics and the framework of this. Thinking about the cultural aspects, I think you may be tipped more towards tracking & reporting (which are "softer") than counting. I like the tracking deltas > tracking absolutes.
It's true, what's measured gets results. I'd add what results you want to get in these counters because I think that would add valuable perspective.
(note: Victor Lowther is reading over my shoulder and thinks that you need to provide a link for cyclomatic complexity)