From 'disaster' to digital | University of Oxford
A mass of social media 'thumbs up' icons
A mass of social media 'thumbs up' icons

From 'disaster' to digital

Helen Margetts was called ‘bonkers’ by her colleagues when, aged 29, she packed in a high-flying corporate career to study for an MSc in politics.

She is now Professor of Society and the Internet at the University of Oxford and director of the Public Policy Programme at The Alan Turing Institute, the UK’s national institute for data science and artificial intelligence. A Fellow of the British Academy and former director of the Oxford Internet Institute, she shares her research and expertise on digitalisation and data science with parliamentary committees, government departments and other organisations across the public and private sectors.

Iain McLean, Professor of Politics at the University of Oxford, spoke to Helen about how she challenged governments' use of IT and data, and became a leading expert on issues of data science, ethics and AI.

Iain McLean - So, Helen, can you tell us more about your work relating to policy, at the Turing Institute and beyond?

Helen Margetts - I set up and direct the Turing’s Public Policy Programme, which we founded in 2018 to develop data-driven public services and innovation to solve policy problems, and the ethical foundations for data science and AI in policy-making. We work with policy-makers in over 70 organisations across the public sector, including 13 government departments, and the programme now comprises over 25 research projects.

One of these projects, for which I am the Principal Investigator, is all about building new ways of detecting, measuring – and ultimately countering – hate speech.

This growing phenomenon threatens democracy and social structures more generally. For example, several female MPs stood down at the last election citing hate speech as one factor in their decision, and some argue it could discourage a whole generation of women from entering public life. But we currently know rather little about the scale and scope of online hate speech, and there is hardly any official data collected. In fact, the UK Home Office collects no statistics on misogynistic hate speech at all, because hate speech (that doesn’t contain illegal content such as death or rape threats) towards women is not a crime. We have just produced a policy briefing which reviews systematically the available evidence, and are developing an Online Hate Monitor to track the phenomenon going forward.

We also do a lot of work on ethics in the public policy programme, including ethical frameworks for the use of data-intensive technologies in government. For example, my team wrote an ethics guide for the use of AI in the public sector, which was launched by the Minister for the Cabinet Office, Oliver Dowden, in June 2019. I hope that guidance like this can avoid incidents like the government's plan for electronic health records back in the early 2000s. That was a complete disaster because the public felt, rightly, that the government was planning to sell the data, and it was a real illustration of what happens when the ethics goes wrong.

I sit on the government’s Digital Economy Council, which brings together government and major players from the tech community to implement the UK Digital Strategy. It is chaired by the Secretary of State for Digital, Culture, Media and Sport. And I regularly give evidence to parliamentary committees, most recently to the House of Lords Committee on democracy and digital technologies, where I talked about our evidence from the hate speech project.

There is in general a lack of evidence about the impact of digital phenomena. For example, with computational propaganda and misinformation, we are starting to know something about the scale of it, but we know very little about the actual influence it’s had on people’s voting behaviour. We know more about the influence of the Sun newspaper on the 1992 election than we know about whether and how much Cambridge Analytica and similar organisations influenced recent elections.

Can you tell us a bit about your background?

Sure. My first career track was as a computer programmer and a systems analyst in the private sector. Companies I worked for included Rank Xerox, and an oil company.

I was on a successful career track. But I was very interested in politics and wanted to write an essay, something that hadn’t been part of my mathematics degree.

So, I did something which my employers at the time thought was bonkers, and applied to do a Master’s degree in public policy and public administration at the LSE. They held my job open for me, but I never went back to the private sector. I worked full-time as a researcher at LSE and studied for a PhD with Patrick Dunleavy and Christopher Hood.

It took me about 15 years to earn as much money as I had been earning when I left, never mind what I would have been earning had I stayed – but I never regretted it.

And what was your early involvement with policy?

At the time – the beginning of the 1990s - I was amazed to discover that nobody ever mentioned computers on my courses in politics or public policy.

But in those days, the Comptroller and Auditor General, Sir John Bourn was a visiting professor at the LSE Government Department. In one lecture he gave he talked about a disastrous project to computerise social security in the 1980s, called the Operational Strategy. He explained how challenging it was for the National Audit Office (NAO) to do a policy evaluation of something like that.

And for me it was fascinating. I ended up writing my Master’s dissertation about why that project failed. Why was it so disastrous? I looked at the relationship between government and technology in what was then Department of Social Security. I turned it into an essay and entered it in a competition run by the Royal Institute of Public Administration. It won the Haldane Medal and was then published.

After that, I got a research assistant post at the London School of Economics, working with Patrick Dunleavy, and I started doing a PhD on the same topic but much wider – it was about IT in government in Britain and America.

So, how did you build upon that, and start to engage more with policy?

Government had contracted out IT work in a completely crazy way. Some of its IT contracts were among the biggest in the world.

While writing my Master's thesis, I interviewed Sir John Bourn. And at the end of the interview, when it didn't matter so much about annoying him, I said: ‘What would the NAO do if one company had about 70 or 80% of the large value contracts?’ At the time, that was true. The company Electronic Data Systems, later taken over by HP, was such a company.

He grinned and said: ‘Oh, we’d get someone like you to look at it, Helen.’ I didn't think it was a very good answer for the whole National Audit Office, which had 700 staff at the time. And there was just me.

I guess that might have sort of sown a bit of an idea in everybody’s mind.

Shortly after I finished by PhD, in 1999, the NAO outsourced delivery of a major Value for Money study to a joint team led by Patrick Dunleavy and myself, looking at Government on the Web. We carried out two further reports in 2002 and 2007 (Government on the Web 2 and Government on the Internet). Together they represent the first systematic study of the development of digital government in the UK, and have been officially credited with major improvements and cost savings to government, something of which we were very proud.

Can you tell us more about the process of delivering that report?

We wrote the report, did all the interviews, and cleared the report with the civil servants. It was a real insight into the policy world.

Before we started talking to civil servants, we spoke to private sector firms at the cutting edge of digitalisation, such as the Financial Times, DHL, and Dell, in particular to see what was possible. And I was so glad we did that, because when government said certain things were impossible, we knew they were wrong.

We had some push back on clearance with certain departments. It was normal for them to have lots of meetings and for multiple versions to come back.

I remember one Friday afternoon, we said we’d like to stay on site until it had been cleared. They said there would be lots of changes to make, so it wouldn’t be possible. So I said brightly, ‘Oh, it's OK, I’ve got my laptop’ – probably the first laptop to appear in those particular offices.

Obviously in the end, what you produce is a compromise in all sorts of ways. But when the 1999 study was finished, I won an ESRC grant (also with Patrick Dunleavy) to study policy transfer and digital government. That allowed us to put the intellectual meat onto all the empirical policy work we had done, developing a framework for understanding digital government which later was published as the book Digital Era Governance and related articles that now have 1000s of citations.

How did you come to be involved with the Turing Institute?

As a social scientist with a mathematical and computer science background, I’ve always been interested in very quantitative methodologies – though I use qualitative ones, as well: my thesis was based entirely on interviews.

In 2011 I applied for an ESRC professorial fellowship, and looking back on it they took a bold step in giving me money to collect 'big data' for social sciences research before the term had even been invented.

So, when the idea for the Alan Turing Institute came along, I joined with two people in Maths and Computer Science to lead Oxford’s successful bid to be part of it.

I later proposed to the director that we should have a public policy programme, dedicated to using data science to make government better in areas such as policy making and resource allocation. Because until then, all the attention for big data and data science was on the private sector – generating profits and entrepreneurialism.

Of course, it had to be funded by Turing core funding. I’m very grateful that they took it on board, although we have other public funding now.

You say government has had some disastrous experiences with IT and data. Can your programme within the Turing Institute make things better in the working life that you have remaining?

I remember in one interview for my PhD, I asked the Head of a Treasury expenditure division about the Inland Revenue outsourcing contract, ultimately worth about two billion pounds, which his division would have had to approve.

When I asked what expertise he had for that, he said, ‘We don’t have any IT expertise, no.’ He was really anxious to assure me that they didn’t know anything about it. 

We are dealing with years and years of people not thinking this is important, not realising that the computer systems, the digital front end, and the data is what government IS, really.

The more that the British government in particular struggled with technology, the more the civil servants would try and keep away from it, because it was like career suicide to get involved in a big IT project.

We are dealing with years and years of people not thinking this is important, not realising that the combination of computer systems, the digital front end, and the data is what government IS, really.

But I have hope. So, the digital infrastructure is sadly lacking in a lot of departments, because legacy systems are still there. They’re using complex rats’ nests of code left over from the 80s and 90s. The Universal Credit programme was initially paralysed by legacy systems like this.

What finally made government go digital was the fact that everybody was using Amazon and eBay and Facebook and so on. People weren’t willing any more to interact with government by letter or cheque.

And there’s also a recognition now that you can’t just buy IT off the shelf. If you want to integrate data intensive technology into policy making, you have to understand it.

What are some of the lessons have you learned during your academic career?

I came into academia a bit late. I did a maths degree, never a politics undergraduate degree; and I was 29 by the time I finished a Master's degree. I still feel sometimes the lack of that grounding at undergraduate level.

But a mathematical background has given me a lot of advantages. People were less likely to question me from a quantitative point of view, just because of that maths degree.

I'd worked in real organisations and understand all that brings, which can be difficult for students of public administration who haven't worked yet.

What advice would you give to early career researchers? 

If you’ve discovered how to cure cancer, then you don’t have a problem finding a policy maker who wants to talk. We need to view topics like hate speech as a sort of cancer of society. 

Throw yourself into every opportunity and as well as developing academic excellence, focus on research which aligns with solving real problems – the two activities can go hand in hand.

We also have to make social science as attractive to policy makers as other areas of research, like medicine. If you’ve discovered how to cure cancer, then you don’t have a problem finding a policy maker who wants to talk. We need to view topics like hate speech as a sort of cancer of society. Social ills, social policy problems, and social issues are things we could be tackling in this way.