Measuring the metrics [LWN.net]

Measuring the metrics

Posted Sep 4, 2019 6:16 UTC (Wed) by shemminger (subscriber, #5739)
Parent article: CHAOSS project bringing order to open-source metrics

It would be really useful to evaluate these metrics on known good vs known unhealthy projects.
For example, the Linux kernel versus some of the critical infrastructure projects that are on life support.

It doesn't take fine tuned metrics to find the dead ones.

to post comments

Measuring the metrics

Posted Sep 13, 2019 21:09 UTC (Fri) by GeorgLink (guest, #134403) [Link]

Hi shemminger,

Yes, that would be interesting. When I was at the University of Nebraska at Omaha, working with Matt Germonprez, we worked with a student who was doing that kind of analysis. If I recall correctly, it is more complicated and we could not determine any pattern. A possible reason might have been that open source projects work very differently and although they use similar tools, the way they use them is different. Another explanation for why we couldn't find a pattern may have been that we were not looking at all of the relevant metrics. Also, we discussed comparing the data across months vs. releases vs. other ways to slice the data and try to determine ebbs and flow of contributions.

I'm sharing this experience not to discourage anyone from trying it again, but to let you know what I know has been tried so far. That said, if anyone is interested in pursuing this kind of analysis, I would love to have a conversation in the CHAOSS project and help with it as best we can.

Best,
Georg