I originally designed this to answer these questions:

• How many people are active in Fedora over time?
• How many of those contributors are highly or constantly active?
• Are we gaining new contributors?
• Are we retaining the new folks who show up?
• Are we retaining older contributors?
  • I would actually like to see more on this, like the average length of contribution over time, and perhaps even patterns in the "shape" of contribution level (like, small activity at first and then growth; abrupt stops vs slowly ramping down, etc.)
• What percentage of work is done by a few people, vs. percentage where a lot of people contribute each a small amount
• What percentage of contributors work for Red Hat (or other companies that are paying people to work on Fedora)?
  • this required some manual analysis, because there is no "I work for Red Hat" metadata

Some things that aren't well-explored but could be:

• What areas are new users most active in?
• Does activity in one area (like QA or Ask Fedora) flow to activity in other areas?
• Are the most active people the most vocal? (
  • We need https://pagure.io/fedora-infrastructure/issue/9576 for this; also, it's hard because Fedora Accounts don't map to mailing list IDs easily so I just avoided that.

One thing that's an explicit non-goal: anything that is easily tied to a specific individual in the results, except maybe for very high level totals.

I'd like to work on this issue, it would be helpful if you could point me where exactly should I begin working from.

@kaushambisharma Well, this particular issue is basically to collect ideas. So, one useful thing would be to talk to people on the Fedora Council (see this forum category) and the Mindshare Committee to collect ideas. Or, if you have any other thoughts on insights we could extract using the message bus data, that'd be valuable as well.

Some of my questions and metric ideas, supported by the CHAOSS Metrics release:

1. Types of Contributions: Not sure if we collect this, but we need to be clear on what contributions we are counting, because we will never measure 100% of what happens in Fedora; it is impossible.
2. Activity Dates and Time: When do people contribute to Fedora? Beware time zones.
3. Burstiness: Do we have sudden activity spikes leading up to a release? When does the momentum begin to build?
4. Review Cycle Duration within a Change Request: How long does it take for a Fedora Change to be reviewed? How long on average does it take for a dist-git Pull Request to get reviewed and merged?
5. Time to First Response: When someone first engages in Fedora, how long are they kept waiting in suspense to hear back?
6. Time to Close: We use a lot of tickets and issues in Fedora, but how long do they take to close? Are some teams more responsive and agile than others? Are some teams/SIGs struggling to close any issues?
7. Contributors: Again not sure if we collect, but we need to be clear enough on what counts as a Fedora contribution or not (i.e. does anything on the fedora-messaging bus count?)
8. Contributor Location: Where y'all at? @bt0dotninja has had scripting for this work for a long time, and it is really important to measure.
9. Inactive Contributors: How many Contributors have gone inactive over a specific period of time? Back in 2015, we knew the average contributor dropped off after three months. How are we doing today?
10. New Contributors: How many contributors are making their first contribution to a given project and who are they?
11. New Contributors Closing Issues: How many contributors are closing issues for the first time in a given project?
12. Job Opportunities: I admit I am not clear how to measure, but there are unsung stories of folks getting a boost in their career by participating in a community with as diverse as a range of skills needed as Fedora.

@jflory7 Responses by the numbers...

1. This is based on what we have in the Fedora message bus. Adding in some other sources could be interesting but is also significant scope-creep. Additionally, I've chosen to just count activity which is tied to a Fedora account, so that means bugzilla (which uses its own accounts) isn't included. The things that are measured are: dist-git, bodhi karma, wiki edits, and irc cookies.There are also bits for pagure commits and new tickets filed in pagure trackers but those parts need work.

2. Activity date and time might be interesting, but I want to avoid drilling down too much into personally identifiable activity reports here.

3. From what I see on the metrics so far, the overall answer is "no", The one observable cyclical pattern is the end-of-year holiday shutdown every year. But that's with everything aggregated together. It might be that by breaking up the data into smaller sets ("activity on packages in Fedora Workstation") there might be more of that.

4. Change requests and dist-git PRs are ... pretty different. I think some metrics on those would be interesting but basically a separate project. dist-git PRs are maybe something we could add, and I do think that's an interesting metric. A related one would be time-to-close of "a new version of package ___ is available" bugs.

5. There are a lot of entry points for someone engaging in Fedora for the first time. I also think this is an interesting one, but it also seems like a separate project.

6. Assuming most teams use pagure for issue tracking, that might be doable. Maybe taiga as well.

7. Yeah, this project is counting activity and uses that as a proxy for contribution. If we add more data-sources (like irc or matrix or mailing lists), that might be less useful. Likewise pagure tickets and comments.

8. Contributor location is valuable but also has data-collection risks. Also, none of it (I hope!) is on the message bus.

9. YES this one I'm really interested in.

10. I'm not sure how we can measure that without external knowledge.

11. I think this can be generalized into some cross-sectional reports -- once you have users bucketed into new/intermediate/older groups, you can look at different activities (like closing bugs, getting irc karma, opening new tickets) by group.

12. Yeah I'm also not clear on this one.