As usual, the PlayOn blog has got some interesting data.
Long ago, in my Small Worlds presentation, I wondered whether optimal guild size (as evidenced in UO veteran member stats) was around 60, since that was the bucket in the histogram that showed the most dramatic spike. Chris Allen expanded on this quite a bit, picking 50 as his figure. Based on the data from PARC, looks like he had it closer to correct. (You can also check his response to the new data here.)
Implications for online game design? Back in 2003, I said that guild support should extend to at least Dunbar’s Number of 150. Now I’d add that perhaps we should design for groups of 50-60 as a target ideal size; if that’s the most cohesive group, might it also be the most retentive?