On Sunday 19 October I took part in the Institute of Ideas' annual Battle of Ideas at the Barbican which aims to encourage free-thinking and open-ended public discussion on a range of topical subjects. PwC is a proud sponsor of this yearly event.
The other panellists for the ‘Big Data: Big Danger?’ session were Sheila Bird from the Cambridge Institute of Public Health, Marion Oswald from the University of Winchester, and Sandy Starr from the Progress Educational Trust. The session was chaired by Timandra Harkness.
The session looked at the danger and opportunities of Big Data. I focused on 4 topics: Definitions; Decisions; Next step; A new Data Deal.
Let’s start by the definition of Big Data, which is assumed often to be understood, but seems to mean different things to different people. There is no right or wrong definition. To me, big data is the ability to use lots of data to support decisions.
That begs 3 questions:
First question: What kind of data? Let’s start with where I live (GPS), who lives with me or next to me, what I read and buy on the internet (cookies), how I use my phone & iPad, my car (GPS), what I buy (credit cards), what we watch on TV (Set Top Box) or listen on radio (DAB), what our neighbours are like (post code socio demographic and behavioural data). And of course a very thorough picture of my health profile. There is one thing which isn’t known (or should not be known), it is my name, as it will have been taken out of the data and replaced by a number. Marion Oswald talked briefly about getting the right law and regulation. There was a great article in the Evening Standard developing these points.
Second and third questions: Data to do what? And for whom? We can use big data to improve health, to improve security and detect frauds. In which cases, the beneficiary overall is “society” – and that’s all of us. Big data can also be used to help me get a cheaper car insurance. It requires me fitting a “black box” monitor that allows my insurance company to know how I drive, which presumably will demonstrate with real time data that I respect speed limits and do not drive erratically. In that case, I benefit and the car insurance company too (and the manufacturer of the black boxes, and those employed to build them).
Big data has also come to imply that decisions are made by computer programmes, or with Artificial Intelligence, automatically. For some, alarm bells start ringing. Let’s look at two examples and check the benefits and the dangers:
Many organisations are already equipped with software that processes invoices automatically. The software checks: “Is this invoice matching a contract we have? Has the invoice been authorised by an approved person? Is the amount within the authorised payment bands? Do we have enough funds to pay now?” That saves the organisation time and improves accuracy, which means suppliers can be paid in a timely manner. When these software packages are well tuned, the entire payment process is more efficient and effective. That seems to be a good thing.
But some would say that isn’t really big data, but simply automation of repetitive, standardised tasks. That’s true. Big data comes in the next step: using all these payment transactions, over a few years, to help organisations manage the overall picture of what they spend, with whom and when. Big data can help describe what has been spent, then it can help predict where the spend will be. It can help detect “abnormal” payments. It can help predict whether there is too much risk concentrated into a few suppliers. Think Fukushima and the problems caused indirectly to Apple and its super-efficient but perhaps too concentrated supply chain.
The second example builds on my car insurance example. What happens if somebody in my family starts driving the same car with the black box, but starts speeding regularly, and has a few “erratic” driving episodes? The insurance company may reserve the right to terminate its coverage of my car. Not great – but at least I should have known.
Extending further to the health sector: What happens if personal health conditions are used to impact eligibility to jobs, promotions, healthcare coverage?
It is easy to predict that big data applications to automate decisions to improve efficiency and effectiveness will continue to grow. For example, one could imagine that IBM’s Watson computer will soon lead to create better call centres through the application of excellent Natural Language Processing proprietary software.
Next step: Bring individuals truly in
Will there be more big data applications where the stakes are more personal? It depends if the public and organisations can work together to improve Awareness and Choices and build Trust.
Do we know what data about us is stored by what organisation? I found that of the 20 or so respondents to an informal survey in the PwC Data Analytics community, about 50% felt there were at least 50 organisations holding some of their sensitive data. Personally I stopped the exercise at around 35, knowing full well that the total is probably north of 150.
And it is not easy to find out what is held about each of us, as it seems one needs to ask each organisation that has some of your date what they hold. Those organisations in the EU are legally bound to give you an answer.
Are we truly given a choice to let our data being collected (and how much), stored (where), aggregated (with what other data, from what sources), passed on /sold on (to whom and for what purpose)? My view is that organisations probably need to become more customer centric and much more explicit about what they do with “our” data … which they seem often to consider “their” data.
I have three adult children, all quite digital. And the debate has been raging. Should we or not be bothered about sharing our “data” and trusting organisations to use our data ethically? The traditional line is “if you don’t have anything to hide, why would you worry.” That seems to be generation dependent. I don’t believe I have much to hide. I am sure I don’t want to share all of my life with every organisation.
We all have organisations we trust more than others. Personally, I will tend to be more trusting of organisations with a public-interest purpose. For example, the BBC, the British Library and the British Museum have missions to expose people to old and new content. Would I trust them to help me achieve a goal (e.g., learn more about French Modern History)? If getting a BBC/BL/BM recommendation of books to read, shows to watch or listen to, and exhibitions to visit requires me to share my goals, my reading list and habits, my diary, my viewing and listening habits, then I would.
A new data deal
Alex Pentland at MIT talks about the need for a new data deal. My colleague, Carlo Gagliardi, talks about putting the customer in charge of their data in a blog for the Management Consultancies Association. Working on raising Awareness, increasing Choices and building Trust – or ACT as a simple acronym to remember our next steps – is today’s challenge for organisations wanting to take advantage of enormous opportunities with big data. Without active promotion of Awareness, Choices and Trust it will much more difficult to realise the phenomenal potential of big data, as it will be seen by many to present more danger than benefits.
One final observation if I may. When we stop exercising, our muscles become weaker. I wonder if each of us relies too much on big data to make decisions for us, we will lose the opportunity to “practice” making simple decisions, and thus lose some of our sharpness in making choices. Remember being able to read a map?