How Big Data affects our lives

Big Data is as much of a threat to civil liberties as the misuse of genetic information and should be subject to exactly the same kind of ethical debate.

Professor Frank Pasquale, speaking at the RSA today, gave chilling examples of firms insisting employees use tracking (phones, FitBits etc) to track their lifestyles and then correlate the data with work performance. This, he says, is a big threat.

His argument focusses on three problems:

Collection. Technologies are adopted because they are slightly better than existing techniques, Pasquale argues, but they are still poor. That leads to frequent errors and If that happens  correcting it at source is difficulty. And even if it is corrected at source that doesn’t mean it will be corrected everywhere. This needs to changed, he argues, so there is a robust method for correcting errors everywhere they have been used. That is difficult, he acknowledged, but critical.


Analysis
.  We are rapidly becoming an administered society, Pasquale believes, and phenomena such as Red/Yellow/Green pre-classifications are commonplace but opaque.  There are numerous examples of abuse, he says, such as people’s credit scores being penalised because they tried to dispute their credit score. “This is a Black Box problem,”he says.  The problem is that these systems are serving those who are scoring consumers not the consumers themselves. “It has become almost a quasi-judicial role,” he says. “A kind of Big Data Star Chamber.”

Uses of data. Credit card companies analysed their data and found something very interesting- people paying for marriage counselling on a card are more likely to subsequently default. This finding can then be used to raise rates or lower credit limits. This he says is very troubling as effectively we are penalising people who seek marriage guidance – a perverse outcome. The answer is to eliminate certain types of sensitive data such as health and sensitive behavioural data.

Pasquale pointed out that with the genetic revolution society decided that the issues were so important that we decided we needed serious ethical debate. Big Data, he says, is of a similar magnitude as an issue and needs a similar ethical debate.

One of the problems in this space is that national jurisdictions vary so much in how they handle the privacy issue. Pasquale says governments use this lack of standardisation sometimes to get around their own country’s rules. The answer, he believes, is more harmonisation of standards globally.

There are some very powerful uses that Big Data could be put to which would benefit humankind – improving response to illness, for example, by analysis of large anonymised data sets of medical data. And he points to another positive example in the treatment of returning veterans of the US Army. Veterans traditionally have a much higher suicide rate than the general population. The Durkheim Project aims to monitor the social networks of returning veterans and perform sentiment analysis to compare with patterns detected in the social networks of those who have previously committed suicide. This is done with full consent and is, he says, an excellent example of the positive use of Big Data Analytics.

Frank Pasquale is Professor of Law at the University of Maryland and an Affiliate Fellow at Yale Law School’s Information Society Project.

Can altruism be more effective?

  Moral philosopher Peter Singer gave a compelling argument today at the RSA for the utilitarian approach to charity. He believes that not only should we give more of our incomes away but we should make sure it is as effective as possible. 

You get much better value for your money investing in developing countries than in developed. It costs much more money to make an equivalent difference in Britain than in a developing country. 

He is a supporter of the Effective Altruism movement which he says originated in Oxford and has been growing in influence for the past decade. He cited organisations such as Giving What We Can which encourage  people to give 10% of salary but also promote an analytical approach to its allocation. 

The idea which initially sounds very appealing isn’t without its problems, however. Singer himself raised several questions: How do you decide what the best thing to do? How do you compare cataract removal with maleria net? How to treat non-human animals compared to humans?  How do you measure the value of giving to organisations which try to change government policy rather than directly intervene? 

Despite these challenges he believes trying to allocate resources based on effectiveness could transform philanthropy. 

“Most people don’t research and even those who do mostly do very little,” he said. Any improvement would be a good thing. Similarly he resists giving out a target for the percentage of income which should be given. Much better to start somewhere and then review. 

His ideas create discomfort though. Do they mean we shouldn’t invest in charities serving developed countries? Isn’t the logical conclusion a reductionist one where the single most effective organisation gets all the money? How do you value volunteering rather than cash? 

His answer is again pragmatic – applying an evidence-based approach to giving (time or money) will make the world better, even if it is far from perfectly implemented. 

The future of jobs

Two of the speakers at this year’s Thinking Digital conference, Tony Hey a former vice president of Microsoft Research and Luciano Floridi Professor of Philosophy and Ethics of Information at the University of Oxford, both argued that fears over AI are overblown.

But interestingly they both believe increasing automation will result in a very significant and permanent rise in unemployment as more and more jobs are rendered obsolete by technology. They both argue that society will need to deal with this through some kind of basic income mechanism as well as a big pick up in educational attainment.

It seems like the basic income is an idea whose time is fast approaching.

Why information grows

A short history of information from the Big Bang to the modern complex economy. That was basically the entertaining talk Cesar Hidalgo, who leads the Macro Connections group at MIT Media Lab gave at the RSA last  Thursday.

He started with the universe made up of matter, energy and information. Information is expressed by how physical things are arranged and you need to spend energy to create information. Take a deck of cards – shuffle the deck and you have the same matter, but arranged differently. That is information. DNA is also information. For information to exist it needs to be embedded in matter.

However, for information to grow – which is the subject of his investigation – it has to develop the capacity to compute. Computation is much more widespread that we usual think. For example trees are able to compute, he says – they know which way to grow their roots and their branches, when to shed and when to grow their leaves and how to fight off diseases.

 

So living organisms can grow information but they are limited by their own internal physical properties as defined by their genetic codes.  Humans are the only species to break through that limit. By using computation to make our imaginations real we grow the total amount of information. “Products are embodiments of imagination and information.” In other words we “crystallise imagination”.

So why do we do this? In order to transmit the skill of others who you may never have met in the form of products. He provides a good example of how much value imagination provides. Take a supercar such as the Lamborghini Veneno which costs $3.9m. If you rearrange the matter in the car by crashing it into a tree at speed you still have exactly the same amount of matter, but after the crash the idea of the car is totally different and so is the value.

So, humans found a way to crystallise information but we have a finite capacity to accumulate knowledge. So we need to organise the work in a network so that different people provide different parts of the puzzle. But embodying information in a network of people is hard and there are limits on the size of the effective network, trust being the key ingredient.  “Trust reduces the cost of transactions,” as he puts it.

Societies differ in their intrinsic levels of trust. Familial economies are low trust societies where only family members can be trusted. These societies therefore have small networks and basic industries and they tend to want the state to step in to solve all problems.

High trust societies on the other hand have networks of people organised in firms and have larger and more complex industries (aircraft manufacture, for instance) . They tend to self-invent institutions to help them where they need it.

Therefore the differences in income are differences in computational abilities, Hidalgo argues.

As societies get more complex the next step up from networks of people in the growth of information is networks of firms.

 

There is what Hidalgo calls a re-embodiment of computation  in ever more complex structures. And economies highly nested and activities cluster together.

It is therefore possible to predict quite accurately which categories of exports countries will move into by studying what they are exporting now.

Thus he believes the information theory of economics is much more useful as an indicator of economic strength than GDP.