News Reader

The age of the superforecasters: Should data science shape the way we do government?

Can data scientists change the way we do government? Senior figures in the UK government are certainly putting their faith in data as a way of revolutionising the approach that bureaucrats take towards making, shaping and delivering policies fit for the digital age.

The move is being driven by The Prime Minister’s chief adviser – the controversial Dominic Cummings, the man credited by many with masterminding the campaign to take the UK out of the European Union. Indeed, many feel his success on that front relied on an understanding of the power of data, AI and social media.

Now, he’s looking to change the profile at the very top in the corridors of power. With bureaucrats and ministers typically graduating with arts degrees, Cummings (who read Ancient and Modern History) believes a fresh perspective is required. Somewhat unflatteringly, he’d previously said he was looking for ‘super-talented weirdos’ to help him drive that change.

Data science ‘skunkworks’ to innovate

A new job advertisement for a £135,000 expert has now shed light on how data plays into this thinking. The Guardian reported how the advert explained: “The analytical unit, known as 10 ‘data science’ or ‘10ds’ is a pseudo startup within No 10 designed to drive forward the quantitative revolution. The current plan is to establish a data engineering team, data science team, a skunkworks and an analytical deep dive unit.” 

Skunkworks is an unusual term originally used to refer to innovative work carried out by an aircraft manufacturer free from the constraints – reflective of the way Cummings and co want to pioneer a new approach.

The advert goes on to explain: “The unit will ensure that No 10 is an intelligent customer of analysis, providing challenge and feedback across government.”

‘Government needs to evaluate data more rigorously’

This vision for a government built on data science was expanded further by Michael Gove – a Cummings ally and senior figure in the cabinet.

In a Ditchley Park speech in July, he explained: “Government needs to evaluate data more rigorously and that means opening up data so others can judge the effectiveness of programmes as well. We need proper challenge from qualified outsiders.

“If Government ensures its departments and agencies share and publish data far more, then data analytics specialists can help us more rigorously to evaluate policy successes and delivery failures. People’s privacy of course must be protected. But once suitably anonymised, it is imperative that we learn the hugely valuable lessons that lie buried in our data.”

He went on to explain that this is a challenge that will mean new skills have to be brought in to the civil service, adding: “That means we need to reform not just recruitment, but training. We need to ensure more policy makers and decision makers feel comfortable discussing the Monte Carlo method or Bayesian statistics, more of those in Government are equipped to read a balance sheet and discuss what constitutes an appropriate return on investment, more are conversant with the commercial practices of those from whom we procure services and can negotiate the right contracts and enforce them appropriately.”

Focus on superforecasting

Cummings has also asked government officials to read up on superforecasting as part of a drive to improve their own thinking – prompting a flurry of interest in the discipline from the commentariat.

Superforecasting is the brainchild of Canadian-American professor Philip Tetlock, who demonstrated that some people have the aptitude to make better predictions. Superforecasters dispassionately use statistical and analytical practices to make forecasting a more scientific data-led pursuit – attaching the sort of rigour that has helped to professionalise political polling, for example.

‘Tournaments’ pitted people against one another to find the person who could accurately calculate the probability of something happening. Superforecasters then refine and amend their predictions based on the course of events, leaving ideology and bias to one side – and often outperforming pundits who struggle to see beyond the ever-changing news cycle.

Superforecasters already apply their skills in a range of ways. Some are adept at the markets – applying the sorts of lessons that seasoned traders have also written about to explain currency movements across the globe. While that’s clearly of interest to politicians, so too is the ability to spot civil unrest or the onset of war. That’s already happening for people high up at Nato, while aid agencies are able to rely on such techniques to predict and prepare for famines or other natural disasters that will require an emergency response.

There’s clearly, therefore, value in adding cutting edge data science into the mix at government level – especially in administrations where the skillset is lacking. It can certainly help to fuel an informed debate that isn’t led by emotion or political preference – and pinpointing a developing social policy crisis can be every bit as valuable as the work to predict war or famine.

Superforecasting’s image problem?

Yet Tetlock himself is a little worried that his vision might start to be associated with a particular ideology or creed – especially given the polarising nature of Cummings as such a public advocate of his methods.

In a tweet, he addressed this image problem and wrote: “So ironic if many thoughtful UK readers rejected the book because a man they reject likes it. The core thesis of the book is to be wary of simple heuristics & stale status hierarchies of expertise.”

He expanded on that in an interview with The New Statesman, adding: “It would be unfortunate if forecasting tournaments or superforecasting came to be linked in the public mind with a particular political point of view. Regardless of ideology, surely we can all agree that we want our leaders to have ready access to the most accurate possible probability estimates of the consequences of the courses of action they are considering. After all, who wants to be led off a cliff?”

Political pundits are keen to tell us all that we’re in unprecedented times – yet governments aren’t completely powerless in the face of challenges. Through a more analytical and neutral approach to data, statistics and forecasting, the next set of challenges can be more accurately predicted and prepared for. The recovery from a global pandemic might well represent a good time to start. It’s likely that governments from across the globe will be keen to see how well superforecasting and a data science ‘skunkworks’ is able to do this in the UK, as policymakers ponder their own approach.

Subscribe to Vulcan Post Newsletter

Stay updated with our weekly curated news and updates.
 
Read more about our privacy policy here.