Thursday, April 28, 2011

Beyond Small Groups - Organizational Growth and Phase Changes

In my earlier post, Small Group Size Limits and Self-Reinforcing Feedback Loops, I talked about internal forces limiting the number of active members of a group.  While I said that meetings never got much turnout, that doesn't mean that membership necessarily dwindled.  I've been involved in organizations with a local dues-paying membership of 20, 30, or even 250 for a non-revolutionary union.  But still the primary decision making meetings had the same limited turnout, time after time.  We could plan actions and various interesting events and get sizable numbers to show up as consumers, but it was generally the same few people doing the behind-the-scenes work and making decisions.

Here is an example diagram of such an organization, with 6 highly connected active members (red), 16 slightly connected members (green), and 25 distant members with little or no connection (blue).

That's not democratic and it's not sustainable.  The few people making the decisions and doing the work eventually get burned out.  Meanwhile, the rest of the membership isn't being plugged in effectively and aren't developing needed skills and confidence.

Using the concentric circles model from The Purpose Driven Church, such a structure has an overworked core and an uninvolved crowd, with no congregation or committed.  The gap between the core and the crowd makes it that much more difficult to build the core.

So what do we do about all this?  Fortunately people have been researching how groups work and we can learn something from that.  In Discrete Hierarchical Organization of Social Group Sizes, researchers show that "rather than a single or a continuous spectrum of group sizes, humans spontaneously form groups of preferred sizes organized in a geometrical series approximating 3, 9, 27, ..."

What does that mean?  Essentially, if you took various established groups of people and listed out all of the sizes, you might expect that you'd find groups of every imaginable size.  Or you might expect that maybe they wouldn't all be equally common, but at least there'd be a simple pattern in the probability density function, like small groups are more common than larger groups, but that other than a sloped pattern of some sort, the graph would be nice and smooth, with no bumps.  But, no, as you can see below, it's all full of bumps.  Groups tend to be tightly clustered around specific sizes, with sizes between the bumps not very common at all.  It's not quite as simple as 3, 9, 27, ..., but the pattern is roughly that the popular sizes are powers of 3.

As the researchers point out, "It may be that the absolute values of the group sizes are less important than the ratios between successive group sizes."  So perhaps the underlying force driving this pattern has no preference for 3, 9, 27 vs. 2, 6, 18.  This is interesting, because when I looked at my past experience, starting with 3 didn't really match, but starting with 2.55 (or 0.85 instead of 1) actually matched my personal experience very well.

If you start with 0.85 and successively triple it, and then take a range on either side by scaling by 3^(-0.33) and 3^(0.33), you get the following set of numbers and ranges.  What's interesting is that not only does this series match my experience, but the specific ranges seem to match fundamentally different structures of groups.

  • 1:     1       Lone individual
  • 3:     2-4     Informal group - formal structure and even voting make little sense
  • 8:     5-11    Formal process needed to function smoothly, voting makes sense
  • 23:    16-33   Full group interaction impossible, communication needs to be limited, subgroups make sense
  • 69:    48-99   Subgroups essential, full deliberative meetings essentially impossible
  • 207:   144-297 Larger than Dunbar's number - impossible to know everyone, coordination of subgroups is complex

What about the gaps, though?  Presumably there are natural forces that tend to move groups in the gap to a nearby common group size.  Likely down, since these are likely sizes where no structure works very well, causing people to leave until a more workable group size is reached.  Or up if the group sees the problem and fixes the structure and recruits new members.  According to this table, we'd expect that groups of 10 people might be stable, but a group of 13 would tend downwards to 8 people.  In this case, it makes sense and matches my anecdotal experience.  Meetings of 10 people are doable with some good process, but a meeting of 13 would either be long or the process would have to stifle a few people almost completely.  Either way, you'd probably lose a couple people.  And adding complex structure to a group of 13 would seem forced and overly bureaucratic, unless of course, you recruited a few more people.

The research paper, being from evolutionary anthropologists, suggests that the reason for the gaps might have to do with some fundamental limits of the human brain that limit the way we can relate to other people, and the numbers of people we can maintain specific types of relationships with.  They also say that "at present, there is no obvious reason why a ratio of 3 should be important."  Presumably this ratio has something to do with either a limit in human brains or some more fundamental aspect of networks, where a smaller branching factor such as 2 or a larger factor such as 4 would be less functional or efficient.

Regardless, though, how does this help us with our problem?  How do we take a group of more than 15 people but have the same benefits of a smaller, "tight" group?  How can we maintain participation, democracy, social bonds, and leadership development?  And do it efficiently so that people don't flee the organization?  And can we do all that while still growing the organization exponentially?

I'll leave thoughts about growth for another post.  For now, let's look at the simpler problem of a sustainable, more static structure.  Looking at the different group sizes above, I think it's clear that no single structure will accomodate all of them.  In fact, I think a single structure can only really work within one of those ranges.  Each tripling in size leads to a phase change requiring a fundamentally different structure, and these changes probably need to be done in discrete jumps rather than a lot of smooth, incremental changes.

Now, a lot of people already have experience with small groups in the 5-11 range or smaller.  And the problems with keeping such a group functioning are more basic issues of meeting process and group discipline and purpose.  These issues aren't trivial, but I think they apply in all group sizes and don't fundamentally alter the group structure.

But what about a group of 19 active people?  Well, already, for full involvement to be possible, the primary way these people participate in the group can't be in a large full-membership meeting.  They will have little opportunity to express themselves, to contribute, and to learn.  A full-membership meeting is still necessary to maintain group cohesion, but otherwise people need to be primarily involved in the group through a subgroup of some kind.  There can be a few of them, each one probably in the 5-11 range, maybe a bit smaller.

Here is an example diagram of such an organization, broken up into 3 cells with 19 active members (red) and 27 slightly connected members (green).

These subgroups could then coordinate directly with each other or use some time at the full-membership meeting to coordinate.  What's nice about this is rather than drifting towards a small group of leaders and a much larger group of passive members, far more people get a chance for leadership development and making a contribution to the group.

This is the basic idea behind a concept I've been reading about called "Cell Churches".  These are evangelical churches that treat small groups as the primary structure, with larger congregations bringing everyone together for group cohesion.

For a larger group of 69, small groups would still be the primary way for people to be involved in the organization.  The difference from a group of 19 is that there would be far more small groups, so many that simple ad hoc methods for coordinating them would fall down, and coordinating at the full-membership gathering would be too time-consuming, so some additional structure would need to be created to facilitate that.

Likewise with much larger groups, it would make sense to divide up on some criteria such as geographic or language or whatever makes sense based on the group's purpose and strategy.  It would still be one organization, with city-wide, organization-wide projects and struggles, but member's direct involvement would through a smaller section.

This sounds complicated, though.  And if growth requires fundamentally altering the structure repeatedly, how could we make and track plans?  I think I have some possible solutions for this, but that'll be for another post.

Next: Logarithmic Planning for Exponential Growth.

No comments:

Post a Comment