What is the Ideal Conference Size?

View of the Golden Gate from SSL

I recently attended the Flux Emergence Workshop 2011 at the Space Sciences Laboratory on UC-Berkeley's campus, which was a new experience in my conference-going career.  As the acronym implies, FEW is designed for a very small group (~25 attendees).  Everyone that attends is given a 45 minute time-slot and asked to prepare a 30 minute talk.  The other 15 minutes are designed to be used for questions and discussion both during and after each speaker's presentation.  It is also highly focused on a single topic - transport of magnetic fields from the solar interior to the solar atmosphere.  It was an excellent conference and easily the most productive meeting I have ever attended from a research standpoint.

This has led me to think about the ideal conference size and how to improve the conference going experience.  First, let me divide conference into two groups:  community conferences and topical conferences.  Community conferences are things like the AAS meetings, the APS March, April, and regional meetings, and the AGU.  These meetings are extremely broad and are designed to reinforce the sense of community felt within the broad confines of an academic discipline.  Let me set aside those types of meetings for the rest of this discussion.

The other type of meeting is what I call a topical conference, and I mean that very broadly.  This is any type of meeting that is designed to address some sub-field.  For me that would cover everything from the Solar Physics Division meeting, which covers all of solar physics and has roughly 300 attendees (depending on the location), to the Flux Emergence Workshop. Any conference focused on a given instrument, observatory, or technique would also count. I imagine that most if not all grad students have attended several such meetings.

There are a lot of variable that control the conference experience, but I want to focus on just one, namely number of attendees.  I have attended 2 conference with less than 40 attendees, 2 conferences with between 40 and 80 attendees, 3 conference in the 80 to 150 attendee range, and 2 conferences with more than 150 attendees.  Here is my ranking of conference size based on my experiences, from best to worst:

1. Less than 40:  This size encourages extensive interaction with the entire group.  The group is small enough that everyone knows everyone by the end of the week and there is ample opportunity to discuss the details of each person's research.  This size also forces a strong topical focus.
2. 80 to 150:  This size means that the meeting has some topical focus but is big enough to bring in a large cross-section of the sub-field.  With a group this size you are likely to have a few people whose research is extremely relevant to yours to talk to.
3. More than 150:  This size means that pretty much every major research group in your sub-field will be represented.  The major challenge is finding the people you want to talk to and actually finding time to have a meaningful conversation.
4. 40 to 80:  This size of conference comes in last because these are generally too large for an extremely focused meeting but not large enough to bring in a good cross section of the sub-field.

Of course my experience comes from a relatively small sub-field, so maybe the numbers don't transfer to other fields, but I am interested in the commentariat's views.