Hewlett-Packard discovers communities of practice by analysing intenal e-mail exchange
E-mail reveals real leaders [via Column Two]: how Hewlett-Packard discovers communities of practice by analysing intenal e-mail exchange. Contains reference to the paper, which describes the algorithm in more details:
Tyler, J. R., Wilkinson, D. M. & Huberman, B. A. Email as spectroscopy: automated discovery of community structure within organizations. Preprint http://xxx.lanl.gov/abs/cond-mat/0303264, (2003).
Later, via comments to this post
In two recent projects I noticed a strong bias when comparing e-mail derived mappings against ethnographic observations. e-mail 'tend to favor distance, the exchanges are far more reflective and structured, the level of trust and open sharing is mostly lower and the boundaries are different, i.e. e-mail identified communties have a tighter core and almost no periphery.
Question the claims of 'mining tacit knowledge' via e-mail analysis and experience suggests you will get a highly skewed 'picture'. Network data analysis can be very different depending on local practices and habits e.g. subscription to listservs, use of BCC and 'passing' (forwarding) from node to node.
Good communicators (inc real leaders and managers) tend to maximise individual telecons or face to face contact. E-Mail traffic is skewed based on people's between-the-lines motives for using e-mail. Self-aggrandisement, lack of courage and arse-covering are three to think about.