I’ve been looking for information about W.L. Gore & Associates because of a connection with Requisite Organization research. In my search, I came across an interesting discussion about management styles by instrument scientist Eric Nehrlich. He directs us to a very useful case study about Gore (which will be dealt with in a later post) but he also mentions the idea of Dunbar numbers. I figured I’d note here it and a couple of things Elliott Jaques and David Billis have said in a similar vein.
Dunbar numbers are quite “meme-y” right now amongst the cognoscenti, and it provided me an excellent illustration why lots of publishing doesn’t necessarily make for better information. Run a Dogpile seach on Dunbar numbers and you get an inordinate amount of blogs where people make wildly unsubstantiated comments about how this idea lines up with their experience. It sounds a lot like Jung talking about how when he thinks about a red car and then sees it, it’s something special (synchronicity) rather than simply that odds are he is likely to see a red car when thinking about one simply by chance. Or the common misconception that there are more problems on nursing wards on full moons: nurses have a busy night, look out and see something that they think is full, and say “Q.E.D.!” We all do it, which is why we have developed science. I’ve been knocked at least twice on this blog about unsubstantiated or testable claims, and deservedly so. Anyway, the only good way to get any information on The Dunbar Number (150, by the way) is to read his article.
Dunbar argues, in “Coevolution of neocortical size, group size and language in humans” [Behavior and Brain Studies, 16(4):681-735 (1993)] that the size of the neocortex has a direct relationship on the average size of societies in primates. Primate species with larger neocortexes (neocortices?) have larger bands. Apes have larger groups than monkey, and humans, with a massive neocortex, have larger social groups than the apes. If the averages for species holds, then humans should have an average upper limit of about 150 people per social group. It works out to perhaps be larger or smaller than that in practice. Read his article, or the books (1996 and 2001, I think). Human groups in “primitive” societies (which may of course simply mean a lot of things that aren’t so simple) apparently hover below 150. Bigger than that and they fall apart, like Baptists. Supposedly, early agricultural communities in prehistory also were less than this. Which is why Çatalhöyük is so incredible with it being 10,000 years old and housing thousands. Bryan McNair has an interesting review of the salient points from Dunbar’s Grooming, Gossip, and the Evolution of Language (1996) that covers the issues probably pretty well. Like I would know. At least it isn’t obsequious.
Julian Fairfield mentions this monkeybrain concept in his talk on learning. So it ties back once again…
If Dunbar is correct (and I’ll take a peer-reviewed article that tries the underlying science or that tests the idea over a good sized sample of groups), then Jaques is totally wrong.. Except of course that he wasn’t.
[You know, the whole 150 and the group falls apart sounds so Tower of Babel. God’s curse comes back on us. Anyone want to run with that?]
Obviously, humans have achieved greater communities than 150, even if we tend to create subcommunities in order to get our personal experience down to that number. How did we do it? Why can we achieve larger social groups that still maintain some social cohesion? Think of the Moonies, Baptists, Christian Reformed people (especially if they are Dutch), Brittany-wannnabes…
Well, apparently Elliott Jaques discovered some of the self-organizing priniciples of human groups that allow them to become larger than their apparently hard-wired neocortextual limit. (Not that I entirely buy this whole evolutionary biology angle anyway.) Groups that were able to get beyond this 150 person limit were more likely to succeed simply because they could do more. We started organizing hierarchically. This isn’t some manmade creation: nope, we seem to be hardwired to organize ourselved hierarchically. By doing so in a natural or Requisite way, we allow ourselves to unleash the power of thousands.
What Jaques discovered was that hierarchies are a self-organizing principle of human groups. The hierarchy allows for much larger groups to follow a set of ideas and accomplish joint goals. Not any hierarchy, though: only natural or requisite ones. When we allow ourselves to organize naturally, following these self-organizing patterns, we have something totally different and more robust than what we have normally experienced in hierarchies.
Even companies that eschew “manager” and “subordinate” still seem to produce these self-emergent hierarchies. I’m not totally sure, but I suspect from the various things in their own descriptions that W.L. Gore & Associates, the penultimate flat company, is actually requisitely organized if you took a look at it ethnographically. (Extant organization for you heads.) We all know that just because the org chart says “X” — or in this case, “flat” — doesn’t mean that it’s reality.
This is what I was talking about when I said that Jaques’s requisite organizations are networks. I meant they have interesting power laws and emergence. Boom and they’re there. And with David Billis’s work, I think that we can see the dual powers (powers of 2 on money and time).
Hierarchies are really an amazing innovation because when requisite, they self-organize.
In order to test this, we can hypothesize that any group that has exceeded its 150 Dunbar number would have to have some form of hierarchy (extant). We would surmise that groups that do not have a leader who is at least Stratum III — for a variety of reasons, I’m assuming that what we call “leadership” actually occurs at the first layer of true managers rather than at Stratum II — would simply divide. Groups with two leaders of equal stratum would also divide. I gotta work this out but I think we can actually run this as an experiment.
Requisite organization, by layering in natural complexity strata, allows for increasingly complex information and relationships to occur while maintaining a social cohesion. Relationships at the top are more difficult because they have more distance to them. At the line level, the relationships are physically constrained within the shop floor. Distance attenuates this social grooming: you can’t maintain as large a coherent group. This may spell something for at what stratum you can have distance management.
I’ll clean up what I’m trying to say, especially about Gore & Associates, later. This sure doesn’t make a lot of sense the way I’m saying it, but I wanted to start writing it down. Smash these two things together and you get Reese’s Peanutbutter Cups to my mind.
Christopher Allen’s discussion of the problem of Dunbar Numbers at Life With Alacrity
Women workers employed as wipers in the roundhouse having lunch in their rest room, C&NW RR-1943 Clinton, IA (LOC). By Jack Delano
Some of these logic connections appear very presumptuous. There are nevertheless some compatible relationships that appear obvious between Jaques and the numbers being referenced to Dunbar (which in fact might support that Jaques was right and not wrong).
The reference which indicates that information dissemination to groups larger than 150 is ineffective can be supported by Jaques assertion which denotes the significance of cascading three stratum context setting, In addition the three strata hierarchy is also referred to as a “mutual recognition unit” or MRU and while it might in fact run numbers that are in the order of 300 as opposed to 150 the principle is that in order for the hierarchy to be effective there needs to be mutual recognition capability among the MOR, his or her subordinates and his or her SORs (suboordinates once removed). The hierarchical society as a result can grow significantly beyond 150 )or 300 considering that the organization can effectively structure itself to 8 strata. If we considered that each immediate manager had 10 subordinates (the numbers will be less when we consider that some roles are individual contributor roles) a stratum 8 organization potentially can accommodate 11,111,111 employees within, in this example 111 person MRUs.
Bear in mind that the size of the clan, (organization, etc.) has its relationship with the CIP of the leader so in its relationship with these Dunbar numbers if within the population the highest CIP was stratum III we would expect to see the clan divide and form two separate units when the numbers were getting too high. Similarly, if the highest CIP were stratum II they would divide when at a corresponding number. The opposite is also true where a stratum IV was available one would expect to see two or more clans merging.
Think about the significance of this in the context of corporate mergers, which often fail. Corporations are off seeking synergistic value propositions and we need to consider that the corporation pre merger reflected the size of the CEO and unless one or the other has CIP higher than the size of the individual corporations post merger, or they recruit a new leader, that the merger is apt to be a failure and the organization will shrink back down to the size of the CEO.
You’ve put it much clearer than I had. Thanks.