I wonder how likely it is that council will have members, which are not participating in the current meeting. This is a critical question because according to this proposal the penalty for not being able to reach consensus would be harder on those members of a council, which participate in the current meeting, even though non-participation of council members is the thing that might make reaching consensus on proposals hard or impossible.
To reach 2/3 of 12 we need 8 council members. What if more than 4 council members are not participating in the current meeting? Reaching consenus is impossible and the penalty is actually less severe on members who caused it (they were earning less that week, because of non-participation, therefore their penalty would be smaller).
I believe that the 12 people according to this proposal are those 12 with the highest weekly averages out of those who attended the round 2 meeting.
Good thoughts though, I think the wording in this proposal could be changed to clarify that point.
Then I see another problem, the same one that I see in the old interim consensus mechanism. Leaders from any one particular consensus meeting might not be good representatives of a community. I mean leaders of one consensus meeting might make a proposal pass, while leaders of next 10 meetings would have rejected it. I guess leaders of the subsequent meetings could undo any proposals. But proposals getting approved, then rolled-back, re-approved... - This jumping between approval/rejection would be confusing to people, and I'd say we better avoid it. That's why I liked this new proposal initially because as I understood it, it would take 12 people with the highest weekly averages, weighting results of the current meeting the most, but taking into account older ones as well.
Maybe it is not a huge issue for an interim consensus process, but something to be aware of.