ScaleAndStructureOfSocialGroups

OK, a number of ideas came together in my head this evening.

• RossMayfieldsThreeScalesOfNetwork suggests that there are different social networks with different link distributions and values at different scales.

• ClayShirky reminded me (on SituatedSoftware) that there were other characteristics of groups than size. Some kind of internal structure. The question this raises is whether such structure is just particular to each group, or are there any other dimensions which we can make interesting generalizations over and can play the same role as scale. If Ross is right, then one kind of structure, that of link distribution, is a function of, or predictable from, size. But are there other independent, but reliable properties of such networks?

• BillSeitz asked if EveryoneWillBeFamousForFifteenPeople was talking about Mayfield's small scale CreativeNetwork, and suggested it wasn't. That immedietely gives a second kind of structure to a small network, that of MicroFanBaseOfABoutiqueArtist, a structure which is presumably star shaped, with maybe some minor connection between fans. But if this structure exists, this is problematic for a (naive) Mayfield position which would assume that size determines link-structure.

• Bill also raised the idea of a FullTimeNetwork which reminded me of TheHereWeb, one of several, slightly miscellaneous kinds of web taxonomized by BillJoy (See BillJoysSixWebs)

• Thinking about the three scales, reminded me of something else I've connected with it before : TomCoatesTwoDimensions of the role of the thing which organizes or is the focus of a group.

: Hmmm, that's good. Tom's extra dimension might give us something else to chromatographically separate groups of the same scale such as Shirky's people in the subway car, the fans of the boutique artist and the highly connected creative team. It's to do with the role of the thing which is the reason to be there. Maybe something to do with the degree of interaction it allows? Or the degree of engagement the group have with it?

• What is this it which organizes a group? Perhaps AddressableThings is relevant here. And it's really a kind of addressable-thing? Or is it more an engagable thing? The subway car doesn't have much of that quality, although the car serves a useful purpose to the passengers. The boutique artist has some of it, because he uses it to make himself the centre of the star. The project of the creative team has a lot of it. OK, so this mythical X-factor is a bit vague and undefined. And if it wasn't for ChristopherAlexander talking about TheQualityWithoutAName I'd be more suspicious of such things.

:: I believe that thing is just Cooperation. People have very strong psychological mechanisms to support Cooperation and when you analyze the examples you can see that where there is need for much of cooperation the team is magically formed. There is not much Cooperation in the subway case, but there is much in the project case. – ZbigniewLukasiak

• As Clay introduced the term "situated" which reminded me of SituatedRobotics and ALife, I guess I can use the word affordance (see also OnAffordance). There are some things which have a kind of affordance for a group to form around. It can be a star of some kind, or a shared purpose. The shape of the group will be partly determined by the group-forming power that the thing has. And maybe the technology ... nah, I'm losing it here. It's getting tautological. Go to bed Phil ...

The array of right structures of the groups are probably best documented and studied within the context of a PatternLanguageForTheSocialNetwork.

Decision making process as another dimension

Oli says :

You're looking for extra dimensions: how about decision making process/structure?

Some groups are partly defined by their heirarchical decision making roles A is boss of B, B is boss of C ...etc...etc.. Then on the loosely structured end of the scale you can imagine some groups having more of a PeerToPeer trend based or adoption based decision making process. E.g. which instant messaging client has been 'choosen' by a large P2P collection of people will be an organically 'decided' choice driven by adoption rates and the network effect. Obviously there are many more such structures and process.

So, once a group has been measured for scale and structure - then maybe you could identify approaches to software that are going to work best for that type of group. Maybe large organisations with tight hierarchical structure will be able to build and deploy very large scale complex pieces of software. Indeed, maybe there's some need for the complexity of the software to somehow mirror the complexity and structure of the organisation. If the organisation is P2P in structure .. then P2P software architecture will work well. If, however, there is a strong center to the organisation that does something different to all of the subordinate 'nodes' - then maybe the software has to reflect that structure.

So, maybe the choice of software reflects the owner, as supposedly dogs do.

– OliSharpe

Good points. Decision-making or power structure is another dimension. (Compare : EmergentDemocracy, CollectiveActionProblems, OnPower, TypesOfHierarchy)

Maybe we're looking at a classification system for groups with these dimensions :

|| Size || "Role" or engagement of the community || Internal Power || Enabling Technology ||

Clearly these aren't wholly independent. Some possible interdependencies :

• The technology can influence the internal power structure : better communication -> more chance of central decision-making working? (Compare IncaRoadNetworks). At the same time, it can enable decentralized.

• The size can influence the AttentionFlow and hence the power-structure : original PowerLaw discussion.

• The Role can influence the attention flow and hence the power-structure : some organizing things engage and distribute attention in different ways.

• Power-structure can affect the size : sometimes a hierarchy can scale up to a certain size better than networks (HierarchiesBeatNetworks); although at the massive scale it seems P2P takes over again (eg. the internet, the global market), presumably because no node has the AbsoluteLinkCarryingCapacity for the number of links necessary to be a hub to such a large organization. And too many intermediete steps between the centre and the edge, attenuate the signal / noise ratio of communication between them.

• Power-structure can affect the engagement. Clearly when attention-flow is channeled to a few super-nodes, others can get disengaged, alienated and disconnected.

So these aren't really orthogonal dimensions, but nor can we collapse or reduce them.

Oli, I'd really like to understand you're thoughts better on this : "If, however, there is a strong center to the organisation that does something different to all of the subordinate 'nodes' - then maybe the software has to reflect that structure. "

Are you implying that a more centralized hierarchical structure is a prerequisite for managing heterogenious "children" differently? Or are you just saying that treating different things differently requires support in the technology?

(This connects to the extremely vague TypesOfComplexSystem, but I'm out of my depth. Oli, as you're a mathematician, do you know anything about the real GraphTheory investigations into this? )

Oli replies:

My comments aren't driven by a mathematical angle, rather just simple observations about what kind of software different kinds of organisations (indeed functions within organisations) will require. You could imagine a large corporation having a whole series of organically connected wikis that are driving important knowledge sharing within the whole organisation. At the same time the supply chain software and accounting software is going to require audit trails, accountability and reporting in ways that are likely to reflect the hierarchical structure of the company.

Indeed, knowledge sharing via wikis is a good example, because I would imagine that in many cases a 'knowledge sharing initiative' within a organisation will precisely seek to break down barriers between parts of the organisation .. get people speaking directly with people they usually wouldn't ... you want to break the structure .. you want organic self-organising. So wiki would be fantastic. However, it would be an 'interesting' choice to try to run the accounts of a large company using only wikis !!!!

So, it's not even that one company will have one type of software that fits. Each 'function' within the company may have a different complexity and structure which has to be reflected in the software used. The larger the organisation, the larger the range of possible complexities.

–OliSharpe

Software reflects it's owner?

"So, maybe the choice of software reflects the owner, as supposedly dogs do."

Clearly, I choose wiki because it reflects and supports my disorganized way of thinking far better than any other software I've ever come-across :-)

Though maybe it re-enforces my HypertextSickness :-(

– PhilJones