Big data: trust is not an algorithm
Lecture date: 29 Sep 2013
Tremendous interest has been shown in Big Data, starting with the NSA and their gigantic facilities for data collection and storage: “Total information awareness” is perceived as a state that would somehow lead to total knowledge. A little surprisingly, the “security industry” has been joined by companies delivering products and services, who hope that collecting vast amounts of data about spending patterns – and perhaps analysing message traffic for keywords? – will enable them to devise new products and services we mere consumers don’t yet even realise we need. Big Data as a substitute for understanding and imagination?
Is digital data the sum total of what is important to our daily lives? Can it be? The answer has to be No, because relationships, trust and personal exchanges are built on all manner of information that cannot be digitised, from the familiar but unquantifiable concept “he has a nice face” to the quality of relationship – not just the measurable length of time – behind “she and I have always worked well together.” Failing to understand this leads to a danger of too much emphasis on digital data and to disregarding the significance of the non-digital. Increasing the emphasis on the sort of data that can be counted and analysed tends to a corresponding increase in procedures, laws and regulations.
Technology hopefully proposes “if you record the way people behave, using Big Data, you can anticipate their future needs and even their future behaviour.” Human activity (including trust!) cannot be reduced to numerical data. As we evolve away from our (atypical) transaction-based society, not least in response to the current financial crisis, sharing should become more important than purchasing. Sharing relies on trust. Trust, being embodied, cannot be digitised.
Human behaviour: “the gift” has been the notion underlying around 95% of human interaction. Private interest, payment, formal monetarised exchanges … probably no more than 5% of the total. We may think, in 2013, that we are discovering the “sharing economy” as something new and exciting, but in fact it has been the norm for most exchanges for tens of centuries before us. It is not well served by Big Data, that reduces individuals to the record of their digital activities and cannot track the truth of their social interactions.
There is a terrifying expectation that Big Data will result in better risk management. In fact, most catastrophic changes can be observed and examined after the event without revealing what actually triggered them. Big Data can be mined for patterns, but “phase shift” changes are a disruption to a pattern rather than the end of a gradual change. Knowing all there is to know about who is telephoning whom not necessarily increase our safety by giving us information about imminent attacks.
Large institutions/data repositories could work well when we want to consider what works and what doesn’t: collating experiences about medical treatments, for example. There is no guarantee that the information collected will support presuppositions, though, and the particular case of medical treatments both ignores and highlights the fact that most medical care is delivered outside the context of hospitals, medical centres and doctor’s practices. The quality of the results any treatment delivers is much harder to quantify than the number of “transactions” (hospital visits, for example) or the details of medication or therapy.
If we live in a world where resources for, say, medical treatment are coming under pressure, we need to pay more attention to the 95% of the time where healthcare is a question of caring for each other. When someone cares for an elderly relative at home, this is rarely assigned a monetary value; the same applies to the help someone else can give by allowing the carer time off. It’s not an easy argument to sell: it’s hard to record informal home care as an economic activity, even though it does have a major impact on our lives and (even) our economic productivity. Informal care – which tends to run on a longer time-scale – does more to determine quality of life and long-term outcomes than formal interventions do.
Large organisations like Big Data: “if it’s something we have a lot of, it must be valuable.” This is not logical! How can we reconcile stewardship of large amounts of data with the trend towards a sharing economy? only by using the content of these data repositories to support and create social and ecological values and solutions. Organisations need to ask “what infrastructure can we provide that will help people help themselves with the knowledge we have accumulated?” This is the opposite of the “why change something that works” attitude. Take water/sewage infrastructure as a real-world example: “why change something that works?” leads to large-scale, high-tech, high-price replacements of elderly systems with a new version of the same thing. Sharing information – scientific, engineering, social – about what the systems need to do leads to community-scale, low-tech, thrifty installations that respond to old problems in a new way. It’s harder to do – but the clever companies will embrace the complicated solution.
Download the interview (mp3)