After the most recent 6-month flood of email about the state of the incubator, I want to propose a set of incremental changes that would address some problems. I recognize that these incremental changes will never satisfy those who would prefer to replace the IPMC as we know it with some other structure. However, I think that the IPMC should try to continue to do a better job even if that kind of radical change might be out over the horizon.
1. Commitment from mentors and the champion
Yes, we understand that we are all volunteers with conflicts on our time. However, the fundamental job of the IPMC is to supervise the activities of the podlings. The mentors are on the front lines of this process. Not everyone has to be paying close attention every day, or every week, but I suspect that some people sign up as mentors with only a vague sense that they will be available to answer questions. The good news, on the other hand, is that we are succeeding in reducing the duration of incubation, so we aren't asking for so much. Further, if a mentor finds the need to step down, we'd like to hope that enough time has passed that someone from the podling could join the IPMC and take up the job.
In particular, I'd propose to ask Champions feel responsible for monitoring the overall health of the podling and reporting to the IPMC in the event of problems.
At project inception, let's ask mentors to make a positive statement of commitment in the proposal, not just sign their names. In particular, ask them to sign up to:
2. The monthly reporting cycle
Our current process calls for monthly reports at first and quarterly reports later, with a template that asks some questions. Given our efforts to speed incubation, and our need to tell whether supervision is happening, I propose a change.
- *Every Month*, at about the beginning of the calendar month, *every podling's champion* must submit a *very* short report. That report should be simple heartbeat: either it says, 'the podling is alive and the mentors are on board' or it says 'there are problems here.' If the champion can't do it some month, the champion recruits a substitute from amongst the mentors. If the report is negative, the requirement expands to an explanation added to the monthly report to the board. I hope that the mechanics here will be to add a line to some metadata and check into svn (for the initial heartbeat).
- When a podling reports, it is absolutely required to provide a list of releases. The chair should not have to scrape mailing lists.
The effect of item 1 is that the IPMC will know, well before time to prepare the report to the board, if there are podlings in distress.
3. Shepherds as stable presences (not)
Jukka's shepherd system double-checks supervision, by getting an independent person to compare the podling report to reality, as represented on the mailing lists. There is a time dilemma here; for shepherding to work this way, the reports have to arrive soon enough to give shepherds time to poke around before the board report is due.
I propose to decouple the shepherd process from the board report process. Shepherds would start their look when the reports are filed. If they find something interesting, that becomes input to corrective action, and input to the subsequent month's board report.
I am not including any of the ideas here for longer term or more formal shepherds. My thought here is that the Champion role, if taken seriously, should serve the purpose of long-term engagement.