In defence of Data Science, a reaction to Cambridge Analytica.

The intersection of data science and politics has hit the headlines this week with a whistleblower's account of how Cambridge Analytica used harvested Facebook data to build the psychological buckets they used for the 2016 Trump Campaign. Facebook have suspended Cambridge Analytica and their British parent SCL Group from Facebook whilst they investigate.

The fallout will continue, with the Guardian now setting up a dedicated section called 'the Cambridge Analytica Files' based on over a year of research and the extraordinary outpourings of whistleblower Chris Wylie. He gives a detailed account of how he used data harvested through a Facebook app developed by Global Science Research, a company founded by Cambridge University lecturer Dr Aleksandr Kogan (who for a while chose to call himself Spectre…over Blofeld, one presumes). But the core issues highlight exactly why Signify exist.

Signify was founded in reaction to what we saw as an increased abuse and misuse of data science, both in traditional advertising but especially in politics. The combination of big data and psychographics has triggered a weaponisation of data that was increasingly being used for fear and paranoia-based campaigning. Whether it was fear of immigration, crime or making teenagers paranoid about their bodies to sell more deodorant, data was and is being used to exacerbate our base fears with the intention of controlling our actions.

It is that word ‘control’ that has defined the massive over-claims by some data science players and led many traditional advertising firms into a data arms race. All advertising and campaign messaging is an attempt to influence and move a target audience from one position, belief or behaviour to an alternate one but the shift in what is being sold from influence to control has tipped the balance and led to the belief that success requires more and more data, that targeting must be more and more individualised and that messaging must be more and more extreme. A course that sees some players de-anonymising datasets and promising results that are running headlong into a brick wall in terms of GDPR, shifting regulatory frameworks, privacy laws and ability to deliver. A course which is also counter to a wealth of scientific evidence.

Signify believes in the power of data to improve communication, empathy and performance. We do this by illuminating issues and causes that people really care about then matching those to a candidate or client’s strengths. For instance, we ran a study of attitudes toward gun control in Virginia that conclusively proved what advocates had long suspected: the vast majority of voters (including NRA members) support tighter gun controls and the tiny proportion that don't (less than 2%) will never vote for a Democrat anyway. In terms of hard electoral numbers, there is zero risk and massive upside in being tough on guns. Events bore us out during the subsequent election in Virginia, as the DNC ended up running a slate of candidates who all had F-ratings from the NRA – and won.

We aim to encourage engagement rather than suppress motivation. We aim to help engineer solutions rather than exacerbate fears. We do this in the firm belief (backed by history and by data) that bringing candidates and electorates together in positive common cause to improve communities and nations is ultimately more powerful and lasting than seeking to divide and cow in fear.

The struggle for the soul and future of data science is real – with positive developments like the explosion of impact investing and micro-finance clouded by the emergence of a total surveillance economy in China, Russia and other states. But one of the most helpful facets of data science is its adherence to fact and truth. And it's a good moment for data scientists everywhere when charlatans see their work unravelling.

If there is a legacy to the exposure of the methodology and intention behind work Cambridge Analytica have done, we hope that it is a change in the way leaders in politics and business think about data and ethics. Signify and other companies who share our values are already working to change the course of data science and apply data and technology in a race to the top. Big data and data science are not going away. They have the potential to vastly improve the level of empathy and understanding within society. At the very least, the ways we generate, retain, protect and engage with data will only become more important.  We will get the data science and politics we deserve. Signify are working hard to ensure better data science and better politics. If you want to learn more about our work or how we can help you get in touch.

Guardian CA Files