Friday 24 August 2012

Big Data Hubris

One of the most extraordinary things going on in technology at the moment is the realisation of the potential for the analysis of communications data. Having created a giant electronic network that includes everyone (the internet), we have created a situation where a significant percentage of our daily utterances are now made in an analysable form: the text that we send to each other, the transparency of those communications, the metadata that accompanies messages (like location and time, for example), the vast capacity for recall of histories and the tracing-back of conversations. Winograd and Flores argued (before the internet!) that computers were about communication (rather than data processing) and that their primary function was to record the speech acts that we make to one another (see I think they were clearly right. Moreover, the recording of speech acts is unprecedented in human history: it is not the 'connectedness' of people that is new with the internet (we have always been interconnected as human beings); it is the capacity for the strategic organisation, recall and manipulation of human commitments.

Literacy created shared memory and allowed for the expansion of culture beyond the confines of oral traditions. Writing, in all its manifestations, has provided a foundation upon which the cultures which surround us established themselves. The book has been the foundation of every cultural development in world history for at least 2500 years. The storage and manipulation of speech acts needs to be seen historically in that context. And just as those who in the early days of literacy must have wondered what it all meant, we should be similarly open-eyed and open-minded about what our current revolution might do for us. It is, after all, a very very recent development. It's worth remembering that in the ancient monasteries, it was the art of interpretation of texts, of asking 'what does it mean?' that occupied the scholars.

In which spirit I would prefer to see our current fascination with the analysis of what is being called 'big data'. I am certain that for all the excitement, there are big dangers. Technology can give us a false sense of confidence that somehow we know, or can calculate, what it means. Its hard, mathematical, logical designs leave little room for doubt and questioning. But the ones who say "This is what it means!" will be the tyrants.

As our educational institutions take more notice of the power of data analysis, there is a danger of hubris regarding the "uncontestable objectivity" of some data analysis. Some people will find this useful to their own ends and use it as ways of manipulating others. We will have managers buoyed-up by the apparent logical rationale and uncontestable analysis that supports their most devious plans. But their analysis can never be any closer to the realities of the daily experiences of those they manage than if they had simply asserted their will without any explanation. Data analysis should not be confused with explanation!

This is not to say that data analysis cannot be part of the process of explanation. It is here that I believe  the most productive outcomes of our 'big data' fascination might arise. But our problem, and it is the problem that's thrown into sharpest relief by the technology, is that we don't really understand what an explanation is. I think if the medieval obsession in the light of literacy was with hermeneutics - with understanding the meaning of things, then our obsession in the light of 'big data' should be with a deeper understanding of explanation. That is to understand how the games that we play with communications lead to learning, community and flourishing. But what we must realise in that process (and this is the antidote to hubris I think) is that however we look at this, the need is always for us to explain.

No comments: