Design notebook: Gaming and the Dunbar number

[evilhat]

I'm hoping to make a series of short posts on "how MMO design influenced the design of The Road to Amber MUSH". A lot of research went into our previous commercial MMO design foray, and a number of things we learned are applicable to a broad spectrum of games, not just MMOs.
This first post is on the Dunbar number, group size, and its applicability

Comments

[thenetweasel]

There's an interesting correlation we've found with software development. The optimum team size (allowing for some specialization) is effectively 6-7. Teams don't hit their real scale factor again until you have 5-6 of these size teams, or effectively large enough for heirarchy to work.

There's plenty of statistical data to back these findings up as well, particularly from companies like QSM. I'll have to read through the original reference and see if it jibes with the anecdotal experience with development teams.

Something to consider though - while those software teams become more effective in their specific area over time, they stagnate and develop tunnel vision (effectively horizontal silos) if left together too long. You really need to be able to swap out about 33-50% of each team every 6 months or so. So regarding your concern for cliques - is there a game mechanism that can encourage that sort of "promotion" or "exchange"?

One final thought: People don't gather on inclines.

[debela]

But they do gather on steps. :)

One of the things that interested me about this is that we strive, as a business unit, to have no more than 7 direct reports to a given manager. My team is actually three teams, and I have, I admit, shamelessly overlapped them in places to take advantage of a given skillset and to prevent the tunnel vision effect. The other problem is that they can end up thinking that they have no promotional path outside their tunnel, and it shocks them to learn that they could move up in other teams.

[bhoneydew]

There's a thick stack of literature on this regarding span of control. 7 or 8 is the general limit; more recent work puts the "crisis capacity" at 3.

Maybe it's coincidental that that number (7) appears in the "how many digits can you remember" type of tests.

The problem of control gap has existed in at least two of my management experiences: the no-man's land between a team of 8 and a team of 35 or so. Between there, there's not enough to justify the hierarchy and too much to run it all centrally.

Unfortunately that's exactly where IT's headcount seems to be for any small-midsize company (between 150 and 600 FTEs).

[rfrancis]

One final thought: People don't gather on inclines.

Unless it's the Fort Ramp on PernMUSH. :)

[malkin]

This is why I laugh my ass off, when EA says that a particular game was made by a "small" team of "only" 100 people.