I finally had occasion to re-read Christopher Allen's post on Dunbar's number and the size of effective human groups. It seems that 150 people is something of a maximum limit, although (according to Dunbar) a group of that size would require "42% of the total time budget to be devoted to social grooming". Dunbar sees boundaries at much smaller sizes: 4 is the most effective size for a group of people having a conversation; a person can maintain very intense relationships with only 10 to 12 other individuals at a time (e.g., a small, focused team that lacks hierarchy or lines of reporting and is based on strong mutual interest and interpersonal trust -- such as, in my experience, the early Jabber developers); 30 to 50 people can function as a small living group such as a "troop" with well-defined leadership and a clear division of responsibilities (this seems to have been the usual size of pre-tribal bands throughout human history); larger groups up to 150 can work, but only with increasingly large investments in socialization and management. Beyond that, you need to "divide and conquer" by forming workable sub-groups (early human tribes were probably formed for short periods from individual bands and did not work together constantly).
I'm especially interested in the smaller groups. One analogy that organizational psychologists and motivational speakers like is sports teams: the nine members of a baseball team, the five members of a basketball team, etc. (such teams have more members, but only a handful are fielded simultaneously). Another analogy I'd like to see pursued is that of musical groups. What's interesting to me about such groups is that they often sustain highly creative output for a long time (e.g., I noticed the other day that my favorite rock band, Yes, is currently on their 35th-anniversary tour). The most stable size of such bands seems to be five, but three (Rush), four (Beatles), and six (Grateful Dead) also seem workable. More than six seems too much, but that may be more for musical reasons than for organizational reasons. In general, Miller's insight from 1956 still seems to hold true ("The magic number seven plus or minus two: some limits on our capacity for processing information.", Psychological Review 63: 2-5). In my experience, though, it's more like five plus or minus two: I've found that it's hard to hold more than seven items in my memory at once, and that groups larger than seven become unwieldy.
My renewed thinking about this was inspired by a brief chat last week with Mikael Hallendal of Imendio, creators of two well-regarded Jabber projects: the Loudmouth C library and the Gossip client for Linux. It turns out that Imendio is a consulting duo based in Sweden -- something like the Simon & Garfunkel of the Jabber community. ;-) I don't know if Mikael and Richard plan to scale up their little group, but the sense I got from Mikael was that they won't scale it up very high (perhaps to that sweet spot of 10 to 12, but no more). Personally I find the idea of working in a small "group" or "band" such as that quite appealing. Sure, it would require a fairly large investment of time and you would definitely need to know and trust the other people to an uncommon degree, but I think such a small band could be deeply creative and productive. (Paradoxically, I've always been a lone singer-songwriter-composer musically, but I think I could work well in a technology group like this.)
It seems to me that Dunbar's research also has political implications, but I'll save that for another day...