Introducing Science Works

Why I think a new organisation is needed to tackle the most pressing challenges facing British science and tech

tl;dr

• I’ve co-founded Science Works, a new research and policy studio focused on accelerating progress in British science and technology. This blog post explains why.

• Last week we published our first essay, Linear Regression by Will Stone – a systematic critique of the linear model of innovation, and why government needs to move past it.

• This week we are launching Reorganising Research: an initiative to work out what mix of research institutions Britain actually needs, and how to build them. It opens with a £3,000 essay prize for new thinking on how research should be organised, and a call for evidence to find the institutional experiments already underway. Both are worth your attention – and, if you know people who should see them, your sharing.

• Have a look at our website and sign up to the newsletter to follow our work.

• And if any of this resonates, drop me a line – I’d like to hear from you.

~

The case for a new organisation focused on accelerating British science and technology rests on two things. The first is about the moment in which we find ourselves as a country. The second is a bit more personal. This post attempts to weave them together. As a result, some of this you’ll recognise and agree with, some of it you absolutely will want to challenge me on, and some of it you’ll have to take as personal reflections.

In the spirit of Tom Forth’s excellent recent piece:

“Intelligent and considered people will disagree strongly, but they deserve my blunt clarity so that they know what they disagree with. We must later find a nuanced compromise as part of fixing our country together but I urge those who disagree to avoid reaching for nuance too soon. Everything is “a little bit more complicated than that”. There is no insight in saying so too early.”

For a while now, the UK has been betting heavily on research and innovation as a critical input to achieving growth and prosperity. The R&D budget has risen steadily, from around £10bn pa in the early 2010s (when I started my career in research policy) to over £20bn pa now. Alongside this – or perhaps as part of this – research policy, technology policy and industrial strategy have become ever more intertwined, from Willetts’ Eight Great Technologies to the S&T Framework to Vallance’s buckets. A procession of ministers and prime ministers has come and gone, each praising Britain’s ingenuity, spirit of inventiveness, or our current/future status as a scientific superpower.

Few among us would say this attention has been unwelcome. But even fewer of us would say that it has added up to a step change by any real metric. What do I mean by this? Productivity remains stubbornly flat, as does GDP growth. Business dynamism is weak; private investment is comparatively meagre; our position in certain global supply chains is more precarious than ever. Pouring money into the research base was meant to change all of this. But our research institutions hardly feel triumphant. Rather, they are under considerable strain – prey to exogenous shocks, and facing a funding crisis all of their own.

There is something else here, not about funding, or shocks, or policy more broadly. More of a growing sense of a bigger threat to science itself. For a while now, we’ve enjoyed a reliable political consensus that research and innovation are worth backing. This now seems to be gradually draining away. We need to be absolutely clear-eyed about this – the things we hold most dear today may carry little weight with a government of tomorrow, and this should worry anyone who cares about the future of British research. For me, it convinces me that we have much (hard and necessary) work to do, much of which simply can’t wait.

So – has science failed to deliver? Is that the real problem? No. This would be a wrong and dangerous lesson to learn. The macro case for the positive impact of R&D spending on productivity is inarguable. And perhaps more importantly, Britain continues to be a great place where great research gets done by great people. When we succeed, we succeed spectacularly. And an enormous amount of good flows from the R&D system – not just economically, but societally, culturally, and diplomatically. We also have to be prepared to defend scientific progress in its own terms – not for instrumentalist reasons, not for political expediency, but for just how cool it is that humans have this one weird trick called ‘research’ that can uncover deep truths about the life and the universe.

I doubt anyone reading this disagrees with the paragraph above. Actually, the nub of the argument is so familiar that it’s become rather boring to say it. It is deployed in every sector submission to a spending review, every UUK or Royal Society ‘manifesto’, every public campaign against potential cuts. The sector makes its argument for ever greater government support on the basis that the system is working as intended.

So if the argument is no longer working, that’s a problem – and perhaps that’s on us. At least part of this is down to changed expectations: we would be fools to think that additional R&D money doesn’t come with a stronger request to address the government’s policy priorities. This includes reckoning with where we should improve and reform ourselves as a sector. I distinctly remember this kind of ‘quid pro quo’ being in the air in No10 discussions around doubling the science budget. Have we done all that much anything to reflect on where we might improve? Or is it a case of ‘give us more inputs and we’ll keep producing the outputs’?

I reflect on the Tickell review of research bureaucracy as a major intervention here. That review was meant to lead to a dramatic reduction in the amount of unnecessary bureaucracy clogging up the system. But, rather than clearing the pipes, the report instead seems to have been simply co-opted by the system, dissolved into working groups and engagement exercises that apparently lead nowhere. Or let’s consider the great promise of the People, Culture and Environment reforms to the REF – that we could use research assessment as a tool for genuinely transforming researchers’ experiences for the better. We instead led the sector down a dark culture war path that is now having to be unpicked – at the expense of the very agenda we were trying to advance.

What if the public thinks we’re stuck in our ways, serving our own interests, and, consequently, hugely disconnected from their needs? If that’s the case, the only way we are going to dig ourselves out of the mess we are in is by making some more fundamental changes to what we do and how we do it. To be clear – this isn’t ‘science failing to deliver’. It’s closer to a collective failure of the system to do the difficult work to consider what’s best for the science base itself.

I am keen that we start by asking hard questions where our own unwillingness to be honest with ourselves is already costing us, and where the work that needs doing is simply not happening at the depth that’s needed.

Melchior Küsel’s Scene Showing the Construction of a Building, ca.1670

The first set of hard questions is around resource allocation. For all that we spend on science, we have a remarkably poor picture of what that money buys. We do not know in any objective sense which areas of research we are genuinely good at, what it actually costs to do research in a specific and justifiable way, or how funding flows through the system at an institutional and sub-institutional level, or where scientific talent gathers and where it drains away. What is the right balance between early-stage and translational research? When do international collaborations serve us well, and when do we need to prioritise domestic capability? What is the right balance between investing in broad ecosystems versus backing the strongest players within those ecosystems? When should we double down on our strengths, and when should we invest in the capability to grow new strengths?

I could never, ever get the answers to these questions when I was in government and the lack of clarity left a gaping hole in the argument for public investment. It made every decision infinitely more difficult, and the system ever more stuck. The Treasury is also deeply aware of this too, by the way. So, my call to the R&D sector to “know thyself” is not navel-gazing; it is the precondition for defending science at all. We need far better evidence about what works in research – better research on research, and better analysis of the whole economy of research and the systems by which we allocate resources – because without it we can neither improve the system nor defend it.

This is a major piece of work that, with regret, the sector cannot be trusted to do by itself, because – deservedly or not – the stuff trotted out by sector bodies is tainted by the bad smell of self-interest and status-quo preservation. What’s needed is an honest and objective picture. To that end, last week we published a piece taking on one of the most stubborn fictions about R&D policy: the linear model of innovation, which holds that ideas flow through to impact via a neat sequence of steps. The model is wrong and it would be foolhardy to pursue it slavishly.

The second set of hard questions is around AI, which is coming for the science base whether we’re ready or not. This is a transition that we are woefully unprepared for, and where our position is sharply double-edged: the UK has a tremendous amount to gain from harnessing AI in our research base – perhaps more than anyone – but we also have much more to lose.

I look at the advances in deep learning for protein science and what’s coming for materials science and mathematical research, and I think we could be on the verge of an explosion in research productivity that ushers in a new technological dawn. The UK could be at the heart of this – we already have amazing capabilities and successes to build on. If we handle the transition effectively, we could use AI as a solvent to scrub off huge amounts of bureaucratic grime from the research process, automating away repetitive and boring tasks, and speeding up research by 10x or even 100x. The UK can’t win on frontier model development, but we potentially have something just as precious: a science base that can be the obvious and best place for capabilities to be adopted, tested, leveraged for accelerated discovery. With all the talk about AI sovereignty, perhaps our science base is one genuine point of leverage.

On the other hand, there’s a huge amount to lose if we get this wrong. An intuition pump: consider you are running a science funding agency in a country with a much weaker science base than the UK’s, but you have a strong sense of urgency, and much less reason to be complacent. Does AI-for-science look like a threat to you? Or does it look like something you can harness, making the gamble that you can leap a generation ahead?

The flip side is that the very technology that could elevate research to new heights is already eating away at parts of it, from grant-writing to peer review, at a way faster rate than we are equipped to deal with. Never mind my intuition pump. We are already past the point where we can wait and see which future turns up. We need to build the capacity to steer ourselves one way or another. At the moment, I don’t think we are acting like a country that understands the stakes.

My third set of hard questions sits around our institutional setup. It is now almost cliche to note that the UK is unusually reliant on a particular model for doing publicly funded research – university academia. That model is now under considerable stress. I do not want to say that these stresses should not be relieved – they absolutely should be. But I also want to state clearly that our overreliance on one single model makes us much more vulnerable than we ought to be. As universities up and down the UK are backing out of projects, shuttering facilities and making staff redundant, there is no other structure to absorb the shock, and so the whole system contracts at once, with capability lost and talent gone. This has all the hallmarks of a monoculture under stress – when the model falters, the whole system falters together.

None of this is the fault of the people inside the system who are (mostly) doing their best to operate effectively inside a system that they did not design (and probably wouldn’t design this way anyway). The problem is that a system running on a single, strained operating model can’t run the variation and experimentation that allows for self-learning, nor can it flex to meet unexpected challenges, and, as it battens down the hatches, risks becoming inhospitable to the outliers that so often produce the most impactful results. We haven’t chosen this system, of course – we have merely inherited it. But we have failed to adequately question it. A more diverse system, where there are more kinds of institutional setup, more extensive models for partnership with industry, more varied routes to funding interesting things, more ways to organise science to optimise for specific risk appetites and cultures… this would be much more resilient to shocks, much broader in what it could attempt, and probably quite a bit more exciting to work in.

This post might be read as a swingeing attack on the existing system. It’s not meant to be. It is, though, intended to be a shake of the shoulders. I’ve spent most of my career thinking about these sorts of systemic problems and wondering why the system can’t act decisively to address them. I now realise that the reforms that are needed are unlikely to be driven by the institutions we already have. Not because the people there lack talent – far from it; these are some of the most talent-dense institutions on the planet! – but because the dynamics of the existing system fundamentally constrain what’s possible.

Science Works is the thing I’m building to try to make a dent on these problems. This is of course a bit of a punt: I’m putting the bulk of a good, comfortable position to one side to bootstrap a new venture, on a shoestring, to work on trying to solve some of the toughest problems in British science and tech. What makes me think we can succeed? First – we have the pedigree. We’ve got a superb team, a talented group of contributors and a seriously good advisory board, bringing considerable experience of thinking about and working on these exact problems. Second, we have a time-limited opportunity to shape policy in a way that could be genuinely transformative – doing meaty work that (current and future) decision-makers can act on, and leveraging our connections and networks to build reforms from the ground up.

Most importantly, the changes that I think are badly needed are not going to come from any single organisation. Reforming how science happens doesn’t happen with a single policy paper or a software tool or a pilot. It requires a shift in what the whole ecosystem of researchers, funders, builders, and policymakers consider to be doable and worth doing. Nobody does that alone and we don’t want to try to.

So if you think this direction is worth pursuing, here’s a concrete way that you can help. This week we launched Reorganising Research – a new initiative to work on defining and catalysing the institutional dynamism that we need in the UK. We have announced a £3,000 essay prize for new thinking on how research can best be organised, and a call for evidence (in partnership with Wellcome) to uncover examples of institutional diversity that goes under-recognised within our existing system, that we can learn from and might be deserving of wider attention and support.

There is more to come. Subscribe to our newsletter to follow along, and if our problems are your problems drop us a line.

Ben