June 14, 2007

...Learn TDD with Codemanship

Social Cohesion in the Agile Alliance

Simon Baker and I have exchanged a few emails on the subject of whether or not the Agile community is a clique.

The ideas are still forming, but we're quite quickly moving towards formulating a little experiment that might shed some light - perhaps.

The question seems to be one of social cohesion. A group of people are said to be 100% a clique if everyone in that group knows everyone else. Such a group is totally cohesive.

So the degree to which a group is a clique could be measured by calculating what percentage of possible connections between members actually exist.

The example network above has 10 possible connections between its members, of which 6 exist. The group has 60% cohesion.

For a social network as big as, say, the Agile Alliance, we would expect cohesion to much lower than 60% - I certainly don't know 60% of the members! But how much of a clique is the Agile Alliance compared to similar groups in the software development community?

An experiment to discover this might require 3 components:

1. Capturing the social network of a number of groups within the software development community
2. Measuring cohesion for each group
3. Benchmarking to ascertain relative cohesion

To keep costs down and to save time, we could exploit an existing social networking infrastructure. It would need to be one whose custodians would be willing to share (possibly anonymised) data so we could perform the calculations. I have a web site in mind, but would need to do some not inconsiderable back-scratching to make it happen.

Getting the support of the membership of the Agile Alliance will be difficult. Getting the support of other social groups in the wider community will probably be harder.

Performing the calculation would require a bit of coding, too.

It may never happen, but at least now I know what might need to be done to make it happen - which brings us a step closer. Let's see where the rabbit hole goes.

(The blog network route might prove to be a much bigger source of data, of course, but the technical work required is not inconsiderable.)
Posted 14 years, 1 month ago on June 14, 2007