Thursday, 23 October 2008

Collaborative software development & social networks

Dr Daniela Damian gave a rushed seminar today on her research that used social network analysis to study collaboration on software development projects within global software teams. They had access to text data so had a record of all the communication between members of teams, which enabled them to analyse between members.

My research requires analysis across teams, both the software team and the client-management team. I don't know what access I'll get to IT projects, but not all IT projects concentrate on software development, so Damian's work is only a bit relevant. But it is relevant.

They asked
  1. what the information flow patterns were, and
  2. if there was a good communication structure that fostered effective collaboration.
They identified different communication structures:
  • hierarchical
  • brokers where there was a go-between
  • completely dense networks
They used probability (Bayes theorem) to predict failing software-build results.

Implications
  • communication does matter
  • you can predict results using social network analysis
Her team had electronic data to work with. I noticed that one of the projects they analysed had only around 15 people in it, which might be what I might get access to. But I'm looking for more than electronic data (wouldn't it be great if I had access to electronic data). I'd like to know about face to face communication as well as telephone, email and any other. And I'd like to know where and when it happens. Damian's team analysed where in the world people were, and looked at the when from the perspective of time differences. But I want to photograph the rooms or car parks or caf├ęs where people meet.

Rushed? She spoke quickly and question time was limited because she had to catch a train.

No comments: