How to make policy in a technological revolution

A few reflections from a talk I gave recently about End State. The audience were policy experts so I rattled through the intro to spend time on the implications for policy-making.

More interesting than my comments were the audience’s questions. I found these really thought-provoking so I’m sharing them here (all unattributed), along with abbreviated responses. (Skip to Implications for policymaking if you’ve read the book!)

Why I wrote End State

I started with a five minute opener on why I wrote End State. The TL;DR version is that the idea came to me in a meeting in Number 10 with Gordon Brown and Tim Berners-Lee. The meeting itself quickly faded from view but the image stuck with me. I had a sense that Brown was talking in the logic of the twentieth century state, asking questions like ‘what levers can we pull in Whitehall?’ Berners-Lee meanwhile talked to the logic of the internet, a world of networks and platforms. I later came to see this as a metaphor for an emerging incongruity between the twentieth-century state and the twenty-first century economy. It left me feeling uneasy — I worried that this gap would only widen.

Then, the decade from 2010 to 2020 unfolded. And things didn’t exactly go well. A host of new social problems emerged; ones that seemed to be characterised by this same sense of incongruity. There was insecurity in the gig economy, concern about the monopolies of big tech, and digital burnout/mental ill-health. Meanwhile there was rising public anger and frustration, manifesting in populism. The rage was ill-targeted but it also seemed to be onto something. I wondered if the widening disconnect between the state and the economy might help us understand what was going on.

Then my thinking took an optimistic turn. At Christmas one year, back in the UK from America, I went into a second hand bookshop on the Charing Cross Road. I stumbled on a small collection of newspaper articles, letters, and diary extracts from the time of the Industrial Revolution. What struck me were the eerie similarities between the industrial revolution and the digital revolution today. Back then, there was a sense that novel social problems were piling up unresolved: sewage in cities, railway monopolies, child labour in factories. And there was a fractious public mood as exasperation fed populism. I was reminded of that old phrase: history doesn’t repeat, but it rhymes. And strangely this left me feeling optimistic. After all, thanks to a generation of radical Victorian reformers, we made huge strides addressing those industrial-era social problems. We did this with ideas that seemed radical at first, but that were later accepted as a new common sense — ideas that became a new institutional settlement to govern an industrial world.

So this tees up the questions I answer in End State. How do we govern digital capitalism? What kind of state will we need? What can we (and what can’t we) learn from our response to the industrial revolution? And what are our generation’s set of radical ideas — ideas that seem implausible today but that will become a new common sense, and the way we govern a digital world?

Reformers at the 1900 Worlds Fare shared radical ideas for reform.

Implications for policy-making

So let’s assume for a minute that we buy this story (which you can literally do, since the paperback is out 19 May!) If a digital economy is leaving behind a twentieth century state, how should we approach policymaking? Here are three reflections on how to do policymaking during a technological revolution.

1. Go deeper

If what we need is a paradigm shift in public policy, incrementalism won’t be enough. If the problem sits at the level of the system’s logic, reflecting a disjoint between the state and the economy, then our solutions have to work at that deep layer; we can’t just tweak existing policies.

I think this failure to go deep enough helps to explain a lot of the struggles we have with policy today. We tend to work at the wrong level of abstraction. For example:

• Think about our policy response to the gig economy, and the public argument, playing out in lots of countries around the world, about whether Uber drivers are employed or self-employed.[1] The answer is they’re neither. These categories predate the concept of digital platforms, which work to a novel logic, so why would they be a useful framework within which to understand and govern platform-based work? It’s not until we grasp this, and stop trying to stuff new things in old boxes, that we’ll make progress.
• Or think about competition policy and specifically the mergers regime. Mergers policy was designed to stop a big company from securing market power by buying another big company in the same sector; it stops a merger of Tesco and Asda, which is how market power worked in the twentieth century. But a platform like Google doesn’t secure market power by buying Bing (who wants Bing?). They buy Doubleclick, or a startup in a nearby sector, and they ingest the capabilities into their digital ecosystem, which deepens their moat. So again the problem sits at a deep level of abstraction; the question isn’t ‘should we block more mergers?’ It’s not about tweaking some dials. We need a fresh diagnosis of the problem and new mechanisms. We need new dials.
• Or consider how we regulate algorithms, particularly when they’re developed with machine learning techniques. You can’t use prescriptive regulatory rules when the algorithm is inherently opaque; we need an outcome-based approach. So again, it’s not just about more or less regulation, or even about changing the rules within an old paradigm of prescription, it’s about moving from one regulatory paradigm to another.

So a good tip for policy-making in a technological revolution is: start by identifying the level of abstraction at which the problem sits.

If the problem runs deep, the work needs to be deep. It means, for example, there’s lots of value in just describing and defining new phenomena (if only we’d spent more time describing platform work and less time fighting over which outdated category to use). It means fresh problem-diagnosis is critical; don’t just assume an old diagnosis still applies. And it means we need space for creativity and Imagination, which is rare in policy work. In general, it means boldness is less risky than conservative incrementalism.

2. Be less technocratic

A related point: when we’re dealing with deep problems, we need to get beyond technocratic and economic reasoning.

This is partly because lots of the premises that underpin technocratic policy-making are themselves undermined by the digital revolution, so the tools have lost their usefulness.

But it’s also because deep problems tend to stray into political economy, even ethics. Notice how, during previous paradigm shifts, many of the biggest policy debates were ethical in nature. We didn’t ban child labour on the basis of a cost-benefit analysis; we did it because it was the right thing to do. Likewise we won’t develop governing arrangements for digital capitalism on the basis of technocratic or economic reasoning. Often it’s useful to take Keynes’ advice: start with ethics, and then bring in economics, rather than the other way around.

An example: should data sit in the commons and be governed as a public good? Is Google’s data really their data or is it our data, collectively? This is inescapably a question of political economy and ethics/philosophy.

We’re particularly ill-prepared for this in public policy today. In fact, one oddity of policy in the late twentieth century was that we came to rely very narrowly on economics as a discipline and on technocratic mindsets. We draw too little on wider disciplines/techniques like reasoning by historical analogy or ethical reasoning of experimentation in psychology. And so policymakers need to venture back out into this long-neglected intellectual terrain.

3. Be optimistic

Policymaking during a paradigm shift is hard work in quite a distinctive way. As one historian of the great Victorian reformers said, it feels like you’re running into the wind. The whole system around you is built on old premises, so it feels like you’re working against the grain. Your work environment — from its culture to its processes, habitual behaviours, and intellectual assumptions — was designed for a previous age, so it inhibits, rather than facilitates, the work that needs to be done. And systems don’t tend to be grateful to people who are trying to reform them.

The other thing that’s odd about long-term policy reform is a phenomenon I call progress amnesia. As soon as we make progress, we tend to forget what came before. Think about the way radical ideas emerge and get shot down as implausible, unaffordable, and dangerous, and then get accepted as common sense. Each time this happens, we forget almost immediately that the idea was ever seen to be radical. Think about how we dismissed public sewers, rules around sanitation/public health, the ban on child labour, job centres, social insurance, free education and healthcare, economic regulation and competition policy, and redistributive taxes. And then, after accepting these radical ideas, we didn’t really stop to reflect; we just kept on dismissing the next generation of radical ideas as they came around.

This has an implication that I find fascinating: there are ideas out there today that are currently thought of as implausibly radical that will be accepted as common sense. In fact this probably describes most of the ways in which we’ll respond to the digital revolution. So what are these ideas? That’s basically the mission I go on in End State — it’s a quest to find these ideas.

Anyway, so my final bit of advice is to read history. It helps to combat amnesia and it gives you energy to run into the wind. You see how far we’ve come and you see the process by which change plays out. This helps you see that the same process is happening now, live, and as reformers we’re part of it.

Questions and reflections

More interesting than my comments were the responses, some of which I’ve captured below, along with abbreviated responses.

1. To what extent does change depend on a crisis?

My favourite metaphor for change is a mudslide. The need for change builds up on the hillside — the mud of unsolved problems and frustration. Then a rainstorm unleashes the mud. We can all see the mud; we know something has to give. But no-one can predict the rainstorm — the liquefier — or the path the mud will take. WWII is an emblematic example in the UK; the war was the rainstorm that unleashed radical policy change, the need for which had been building since the 1920s and 1930s.

Wouldn’t it be nice if we didn’t need a crisis to unlock change? This is what draws me to Roberto Unger’s work. He argues that we should try to reduce the extent to which change depends on a crisis. His project is about creating the conditions in which transformative change is more normal; he wants us to reduce the gap between changes within the system and changes to the system. This means assessing policy initiatives not just on their first order effects but also on their effect on this relationship. In Unger’s words, “we should ask whether [the initiative] develops or undermines our attributes of agency, transcendence, futurity, and experimentalism.” Unger says one goal of policy work is “rob the structures we have created of their patina of naturalness and necessity.” To remind us that deep change is possible.

2. You say we need to go beyond economics, but at least economic tools like cost-benefit analysis are clear and practical. History, philosophy, sociology, not so much.

Recently I wrote a blog post criticising economists. Afterwards I got lots of DMs from economists saying I was being unfair, which I now think I was (well, partly anyway). In particular, I misdirected my aim; the real problem isn’t so much with academic economists as with non-academic economic advisers who often apply a reductive reading of economics to policy problems, and who also often stretch the insights of economics too far, into domains where they’re not supposed to go. (I still think academic economists bear a lot of the responsibility for this, having taught most of these economic advisers!)

One respondent to my post made a particularly good point, which feels similar to the point in your question: they said the reason we over-rely on economists is that economists are really good at developing tools for thinking, and academics in other disciplines are much less good at this. Whatever you think of cost-benefit analysis, as a tool it’s really clear and easy to apply. In a sense, this captures both the benefit and the danger of economics; it’s very easy to use — and also very easy to mis-use.

The work of philosophers and sociologists is harder to apply in public policy, and academics in these disciplines bear some of the blame for this. So I guess alongside my pop at economists there’s a good critique to be made of historians, psychologists, philosophers, for not being more useful. And there’s work to be done — including by policymakers — to make these disciplines useful; i.e. as policymakers we should (a) read widely, beyond economics but we should also (b) see what we can do to translate this wider thinking into useful policy tools.

3. Will populism get worse before it gets better?

I guess my reading of history is ‘yes’. (Sorry!) One thing that’s striking in history is just how bad social problems have to get before something is done. The sewage on the banks of the Thames was six feet deep before we built public sewers, and the breakthrough only came because MPs in the House of Commons were choking at the smell from the river outside. Another example: there were thirty breaks of gauge on Britain’s railway network before the British state mustered the boldness to legislate for and enforce a standard railway gauge.

One of my worries with populism today, which I think suggests our debate is still immature, is that we still treat it as a mostly cultural/anthropological phenomenon. People write about Trump voters as if they’re a strange species behaving in irrational ways. I’m not saying people are right to vote Trump but they’re right to think something is systematically broken in our governing arrangements. In other words, people who are angry are onto something. So until we get beyond the effort to explain away populism with anthropology, and address the underlying policy substance — e.g. the unequal distribution of esteem and earnings in our labour market — I suspect we’re still on the downward part of the curve; i.e. things will get worse before they get better.

4. Technology is speeding up, so is policy-making too slow? And if we rely too much on economics, what other disciplines do we need?

On the second part of the question, I think we should draw more on history, design, and psychology. But the bigger thing is that we need far more cross-disciplinary working. One of the most outdated aspects of policy-making is the way the various disciplines sit in silos, which is a way of working that the tech sector left behind a long time ago. To respond to digital-era problems with agility we need multi-disciplinary teams; policy experts, lawyers, statisticians, and even frontline staff working hand-in-hand in a permanent team setup to solve a problem.

On the first part of your question, I also think our mechanisms for policy development are too linear and run to a dangerously slow cadence. We research, consult, decide, legislate, and then implement, in that order, and the end-to-end process can take years; the Queen’s speech is a painful annual reminder of this. It’s the kind of linear or waterfall mindset that prompted the software crisis of the 1960s, when industrial-era methods proved incapable of managing the complexity and fluidity of software.

Perhaps most importantly, our linear methods for policy-making fail entirely to integrate learning and doing, which is the most profound shift of capability encapsulated by agile methods. So there are lots of flavours of outmodedness in our policy-making process, not just an over-reliance on economics; the system in which the economists are working is itself outmoded.

So my personal sense is that we’re now experiencing a software crisis but for governance, as the linear/waterfall methods of the state, and its cumbersome hierarchical institutional forms, struggle to cope with the pace of change in a digital age.

5. Lots of digital problems (e.g. online fraud) aren’t really ‘new’, they’re just new versions of old problems. So isn’t there a danger that we throw everything up in the air and act like everything has changed?

This reminds me of a comment I once heard from the philosopher Tim Scanlon. Someone in the class asked him ‘what’s the point of philosophy?’ and he responded by saying: ‘there are distinctions to be made, and it’s important to make them well.’

When we’re living in a period of rapid change, it’s easy to get breathless. This is captured by the phrase beloved of tech companies ‘this changes everything’. Every time I see that phrase — normally on an ad for yet another food delivery app, or a slightly better camera on an iPhone — I think ‘does it? really? everything?’.

It’s important at a time like this to make distinctions. I agree we need to distinguish new problems from old problems, and I try to focus on the former — i.e. cases where the digital revolution gives rise to a qualitatively new phenomenon, or a problem that foxes our existing governing arrangements. An example is the new dynamics we’re seeing in digital markets — winner-takes-all effects, and the novel social and economic grammar of digital platforms. This seems to me a qualitatively new thing, marking a discontinuity in economic and social history. Online fraud, I agree, by contrast, isn’t all that new.

Finally, I also think it’s important to distinguish justified concerns from froth. By froth I mean worries like ‘AI will take all our jobs’ or ‘social media will corrupt our kids’. This is no more true than ‘looms will take all our jobs’ or ‘the novel/radio/TV/video games will corrupt our kids’. So what we need is well-targeted radicalism; it’s not about throwing everything up in the air.

Footnotes

1. Or the eventual answer for categorising Uber drivers that we came up with in Britain, which is to use a foggy third category called ‘workers’, which feels to me more like saying ‘none of the above’ than actually answering the question of how to govern platform work.

This post is part of a year-long series on how we govern the future. To read along, follow me on Medium here or support the project for £3 a month on Substack. For the big story behind all this, the paperback of End State is out on 19 May 2022.