Hell is a fully-connected network.

Occasionally I run into a particular situation that is based on a very fundamental misunderstanding of the function of social networks. In an effort to make an organization or industry more innovative, someone gets the idea that we need to get everyone talking to everyone else. In other words, the most innovative network is a fully connected network. If some connections are going to improve innovation then more conections are going to give you more innovations, right? Well, actually, the answer is no. In fact trying to achieve a fully connected network will inhibit innovation. There are probably three main reasons for this.

Dense Networks Can Be Insitutional Straight Jackets
Dense networks are really good at building and enforcing conformity. This has been a topic of interest for network researchers for many years and the theory behind it is very easy to understand. Think about it this way… what is the size of the smallest group that can enforce majority control? Anyone who grew up in a family with two other children can tell you that the answer is three. If there is a dispute, two can usually overpower the dissident. The underlying network structure of governance is therefore the triangle. Any way you draw it, the network of three is a triangle. A fully connected network is a network of triangles. You can see this clearly in the simple network below (from Chad Walker’s blog on network density in terrorist networks).

Fully-Connected Network

Information might flow quickly through the network but novelty and diversity are really going to struggle in this environment.

Real World Network Ties Aren’t Costless
The other problem relates to the cost of network ties. It takes time and effort to build and maintain relationships. Some relationships are going to be good for us but too many relationships are going to get us into trouble (something I’ve started to call the Tiger Woods principle). In one of our major studies of professional engineers in a very large business, reciprocity is the key dynamic that holds the network together. The connection is sustainble if both parties get someting out of the interaction. Past a certain point, it gets very hard to sustain too many interactions. In fact, trying to sustain the interactions will distract from the tasks associated with getting the job done.

null

We Can’t Actually Handle Very Much Information
My final point is that we really can’t deal with very much information at one time. Access to more information doesn’t usually result in better decissions because we tend to prioritse information that we are familiar with and there is usually a rush towards the first feasible option. If you want a thorough coverage of how our brains are generally bad decision makers then read Daniel Kahneman’s “Thinking Fast and Slow”.

As far as we can tell, the natural structure for collaboration networks is clusters of people with a few links between the clusters. This creates spaces for novelty as well as connections between the groups to allow for formation of novel connections that are the basis for innovation. To misquote a particular French philosopher, hell is a fully connected network.

Please note: I reserve the right to delete comments that are offensive or off-topic.

Leave a Reply

6 thoughts on “Hell is a fully-connected network.

    • Hi Ralph

      thanks for that find (yet again… you do a great job in bringing these gems to the blog).
      I’ll do a post on that on Wednesday.

      John

  1. Nice post John,

    There is another, reason why fully connected networks are bad. The only way to get one is by enforcing a static network and not allowing new people to interact with it (as soon as new people come in, the connectivity of the whole network immediately goes down).

    We see this a lot. Social clubs, small towns “not invented here” organizations. They seek to enforce norms through high connectivity, but that inevitably means keeping the network closed and not allowing new information in.

    Greg

    • Hi Greg

      Totally agree with you here. Dense networks reinforce indentity and cohesion but they also define who is in the ‘ingroup’ and the ‘outgroup’

      John