In order to try to give more structure, Richard proposed to try to organize and select talks by topic more than putting all talks into one big pool and select them that way. He called this "session chairs". The linked message describes the first attempt at describing this.
From various mailing list threads and the Feb 3rd meeting, it was decided to try this. Martin (madduck) was interested in implementing this, and helped to develop the plan based on his experiences at LCA.
Given my (madduck) experience from LCA, I suggest the following, which supplements Richard's proposal of "session chairs". This is very much how LCA works and might not be fully suitable, but it *is* a great format, which partly accounts for why LCA is such a successful conference.
For ease of discussion, let's keep with the name "session". I think we might want to later rename that to "block", "track", or "miniconf", depending on what we come up with.
1. A session corresponds to a consecutive block of schedule slots in one room. A half-day session would be ~3hrs, which is enough time for 3×50 minutes or 6×25 minutes, or a combination of the two. It might make sense to think about a rigid skeleton and 5 minute breaks, so that hearers have time to switch sessions without causing too much disturbance.
1b. Based on the LCA schedule, my suggestion would be as follows:
8:00 – 9:00 morning activity (swimming, yoga, running, maybe breakfast) 9:00 – 10:00 morning event 10:00 – 12:00 2 hour slot 12:00 – 13:30 lunch 13:30 – 15:30 2 hour slot 15:30 – 17:30 2 hour slot (including closing activity on last day(s)) (individual/small-group sightseeing occurs here and/or during the beginning of dinner) 19:00 — 21:00 dinner
2. Depending on availability and requirements, the papers' committee can approve sessions spanning multiple blocks.
3. Only part of the conference should be organised this way; the rest of the conference should be free-for-all/independent/catch-all events. LCA has two days of miniconfs (full-day, across 5–6 rooms, so 10–12 miniconfs), followed by three days of general talks. Attendees can still schedule last minute events in available rooms.
3b. Each day should still be started with a common event, e.g. welcome on the first day, and 4 keynotes. This space will be used for announcements, and maybe we can find sponsors for prizes to be given away to entice people to be there (this is what LCA does). The "next-debconf-presentation" could be one of those early events, or be the last event on days 3 or 4. The closing address should be fixed as the last event on day 5. TODO: update this to reflect that DebConf is 7 days long, not 5 (one of which is a day-trip, and one of which is the OpenDay, so effectively it works out the same — LCA has an OpenDay on day 6 too)
4. There is a limited number of sessions. Interested conference attendees are expected to apply to be a chair of these sessions. This should happen early, possibly before the actual CfP. Each session, once approved, will show up in a "list of sessions" displayed when attendants enter their events. There is no initial limit of sessions, but the limited slots shall later be populated with the best session proposals.
5. The call for sessions should thus precede the call for papers by a few weeks. It might also make sense to actively seek session leaders for certain topics that we want to see (see below).
6. There's a 1:∞ relationship between sessions and events, meaning that an event can only ever be associated with one session.
7. Event submitters can associate their events with sessions as they see fit, or because they have been invited by a session chair to submit a specific event. An event needs not be associated with a session, in which case it goes into the general pool.
8. Session chairs can pick events from the general pool, although this should probably be accompanied with some correspondence so as to prepare "squatting" ;)
8b. The papers' committee has final say over events and may at their own discretion move a particularly great proposal to the main track, or have it as a keynote.
9. The papers' committee vets all submissions for general quality, because they have a certain level of experience with that. This should be easier than it was in previous years since they need only focus on certain aspects, and can leave decisions over the applicability of contents etc. to the session chairs.
10. Before events are accepted, the sessions slots are filled with the best sessions. Events belonging to unpicked sessions default into the general pool.
11. The papers' committee puts together the schedule of the remaining days from the (vetted) general pool. It is up to the papers' committee to appoint "independent session leaders" to alleviate their burden too. Independent session leaders are responsible for clearly defined parts of the schedule and serve as single-point-of-contact for those blocks.
 To Do
- Mail lists related to the teams below, encourage them to find a sesion chair for their thing
- Draft announcement saying: suggest a talk to us, OR to your session leader
- Proceed mostly normally
- Find the best way to schedule sessions as time gets closer: Can we put one session per day?
 Proposed Sessions
(list comes from looking at DebConf9 schedule) (names and such need to be tweaked)
- Open Day (Get one person in charge of it, like a session)
- Debian Social Organization (lists, irc, keeping people moving smoothly, NM, ...)
- ftpmaster/package archive/licencing/
- Debian Deployments (e.g. "how we used Debian in this situation"/large scale deployments/in schools/)
- Packaging tools (debhelpr, VCSs, ...)
- Debian Infrastructure (lists, alioth, buildd, ...)
- Change Layout Debian Web, Wiki and others. Proposal: http://www.kalleswork.net/projects/debian/