I very much like your proposal of an open decentralized process for investigating the problem and solution spaces. IRC sounds like a reasonable place for this kind of thing to happen. I also agree with Ryan Grant's comment about in-person cut-through (ie cut through the BS and resolve misunderstandings). Perhaps every 3 IRC meetings or so, an in-person meetup can be organized in various locations to facilitate that kind of cut through. 

I would imagine the phases the group could go through is:

1. Define the phases (these phases). This list of 6 phases could be a starting point, but its probably best to open the floor to whether this feels like a reasonable approach and if more phases are needed or if some aren't. 

2. Define and prioritize the motivations (ie the various features and functionality we want out of covenants, like the ones you listed). By prioritize, I mostly mean figure out which motivations are most motivating to people and rate them by strength of motivation (rather than a ranked list). 

3. Define and prioritize the relevant constraints. These are things to avoid in any covenant implementation. Constraints that have been brought up in the past are things like preventing the possibility of infinite covenant recursion, full enumeration, preventing dynamic state, etc. By prioritize here, it might be useful to categorize them into categories like "no tolerance", "some tolerance", "no reservations". Eg it might turn out most people don't have any tolerance for infinite recursion, but don't mind non-full enumeration. 

4. Other criteria? These are other criteria we might want to evaluate proposals according to. And some kind of way to prioritize them / evaluate them against each other as trade offs.

5. Evaluate the proposals based on motivations, constraints, and other criteria. This phase shouldn't involve comparing them to each other.

6. Produce a set of conclusions/opinions on which proposals are worth pursuing further. This would be the phase where proposals are compared. 

Each phase would probably span over more than one meeting. I imagine each phase basically consisting of discussing each individual nominated item (ie motivations, constraints, other criteria, or proposals) sequentially. The consensus reached at the end of each phase would be considered of course a group consensus of those who participated, not a global consensus, not a "bitcoin community consensus". After each phase, the results of that phase would be published more widely to get broader community feedback. These results would include what the major opinions are, what level of consensus each major opinion has, what the reasons/justifications behind each opinion are, and various detailed opinions from individuals. It would be especially great to have detailed evaluations of each proposal published by various people so anyone can go back and understand their thought process (as opposed to a list of names attached to basically a thumbs up or thumbs down). Think like a supreme court decision kind of thing. 

The process doesn't need to be complete after phase 6. Any previous phase could be revisited, but after a phase is revisited, the phases after it should probably be also revisited in order - or at least until its decided a previous phase needs to be revisited again. Each iteration would solidify consensus more about each phase. I would imagine the group might loop through phases 2, 3, and 4 a couple times (since constraints might conflict with motivating features). It might be likely that in phase 5 while evaluating proposals, people realize that there are additional criteria that should be added and can propose going back to step 4 to do that. Hopefully we would get to the point where the motivations and constraints and relatively solid consensuses and iterations can loop through phases 5 and 6 until the set of proposals the group thinks is worth pursuing  is narrowed down (ideally to 1 or 2).