People behave differently in groups!

To what extent can social software applications be used to observe, engineer and predict social behaviour? Looking at management theory, organizational behaviour and psychology there seems to be a great body of work associated with the behaviour of groups. Groups are dramatically different to individuals. The focus of marketing practice today is primarily around understanding individuals, but the focus should be on groups. Tools, metrics and methods for groups simply don't exist in the mainstream.

I've observed patterns of emergent online group behaviour that repeat so frequently that they seem be to almost inevitable. Perhaps they are, and these behaviours are driven by human nature. Equally, the way that these groups form, grow and interact does seem too influenced by the interaction grammar of the application. The question I get frequently asked by brands/businesses is a point of view on how to control online social groups. I've tried to put together a theory that distinguishes between the things that can be influenced and the things that are unpreventable. So far, the best thinking on online group behaviour I can find is by Clay Shirky.

Online communities are far older than most people think, stretching back to the early 70's. So it would be very interesting to place the behaviours seen on facebook, bebo etc in a historical context. Direct experimentation to demonstrate theories should actually be quite easy. The really fascinating related area is the network analysis methods being developed by intelligence agencies to spot terrorist cells. Like the original birth of the internet in military technology, I see this type of analysis of online behaviour as potentially the most valuable to businesses in the long term. It would also cause people to re-evaluate their view of privacy. If you knew that the way you interact could be used to predict the nature and strength of your beliefs, values and allegiances, you would probably be far less keen about web 2.0 'contribute' culture.