September 22, 2021

I, Science

The science magazine of Imperial College

The weekend of panel discussions tackled issues like the value of data modelling and the ethics of Big Data ...

battle_of_ideas_main

The annual Battle of Ideas at the Barbican took place over a weekend in late October and included around 80 panel discussions on a wide variety of subjects from morality to generation wars, and from art in schools to the battle for scientific information.

Adversaries met on one of Sunday’s main talks to joust on ‘Number crunching and ethics in the era of Big Data’, which was especially relevant in light of Edward Snowden’s revelations about the wide extent of mass surveillance undertaken by the United States government.

On the side of limited acquisition and use was Dr Martyn Thomas, vice-president for external affairs at the Royal Academy of Engineering, who felt people have a right to privacy. He said that many respectable members of society have good reason to keep a part of their lives secret – for instance, HIV carriers may fear prejudice, or witness protection could force a person’s past to be withheld.

Dr Tim Hubbard, senior group leader on a Genome Analysis Project at the Wellcome Trust Sanger Institute, said he recognised that questioning large groups of specially selected people is likely to get better results than behaviour analysis using big data, but pointed out that large group surveys take time. He felt big data provides value because it can deliver results in real time.

Hubbard and Thomas also locked horns on the issue of using genome data. Hubbard supports a future where everyone has a right to have their genome sequenced, but where they choose what type of genetic abnormalities, if any, their doctor would tell them about. Thomas pointed out that this is unfair to the relatives: perhaps a sibling wouldn’t want to know if they’re predisposed to an incurable genetic disorder, but their brother or sister then chose to be informed about that predisposition. In this case, the sibling’s chance of having the disorder has increased exponentially even though their personal decision was to live happily unaware. Is disclosure of certain disorders a decision that needs to be taken on a society level, rather than an individual level?

A sobering middle ground was offered by Bryan Joseph from PwC, who pointed out that we’ve already been using mass data for centuries. He said if we go back to the 15th century, before printing, monks had a monopoly on the written word and they hated losing that monopoly, in the same way that the people who controlled data before the widespread use of databanks will hate handing over that control. He doesn’t see a ‘problem’ of Big Data, but just a situation of more data, and said we are at the start of a journey that society has been through a number of times before: “it will just take time to make new rules and reset ourselves,” he said.

The PwC partner also found fault with one of the major problems cited by Dr Hubbard, which is that we have begun to rely on the data to make decisions. He said more data will reduce the reliance on human judgement in the decision making process, but that there must always be a human who is ultimately responsible for decisions, even if it’s the programmer of the algorithm that crunched the data. The US press recently used the term ‘rogue algorithm’ in reference to a financial mishap, as if the algorithm itself was responsible and could somehow be blamed. Joseph drew a parallel with a fictional motorist who follows the directions of his satnav past a ‘Road Closed’ sign – who’s to blame for the resulting accident? The satnav, or the user who just drove into a lake?

Data was a running theme of the science and technology panel discussions. In another on data modelling, all three speakers noted that there were weaknesses in the modelling process. Dr Robin Purshouse from Sheffield University felt models are always wrong to some extent but can be useful, while Hilary Salt, the founder of First Actuarial, felt ‘anchoring’ damaged modelling. This is where a model is built with an expected output in mind that can influence the outcome. It’s like the placebo effect – even when you’re aware of it you can fall into it. An example symptom would be not bothering to go back and check results if the first output feels right.

There were dozens of other talks over the weekend on topics such as ‘Science journalism: the tyranny of evidence’ and ‘3D printing: a new industrial revolution?’ Videos and podcasts of many of them will be appearing on the Battle of Ideas website over the next weeks and months so check them out to catch up on what the specialists said when the gloves came off at this year’s Battle of Ideas.

The Battle of Ideas took place in various rooms at the Barbican Centre over the weekend of 19/20 October 2013.