UK ARPA: An experiment in science policy?

This essay was published yesterday as part of a collection called “Visions of ARPA”, by the think-tank Policy Exchange, in response to the commitment of the UK government to introduce a new science funding agency devoted to high risk, high return projects, modelled on the US agency DARPA (originally ARPA). All the essays are well worth reading; the other authors are William Bonvillian, Julia King (Baroness Brown), two former science ministers, David Willetts and Jo Johnson, Nancy Rothwell and Luke Georghiou, and Tim Bradshaw. My thanks to Iain Sinclair for editing.

The UK’s research and innovation funding agency – UKRI – currently spends £7 billion a year supporting R&D in universities, public sector research establishments and private industry [1]. The Queen’s Speech in December set out an intention to increase substantially public funding for R&D, with the goal of raising the R&D intensity of the UK economy – including public and private spending – from its current level of 1.7% of GDP to a target of 2.4%. It’s in this context that we should judge the Government’s intention to introduce a new approach, providing “long term funding to support visionary high-risk, high-pay off scientific, engineering, and technology ideas”. What might this new approach – inevitably described as a British version of the legendary US funding agency DARPA – look like?

If we want to support visionary research, whose applications may be 10-20 years away, we should be prepared to be innovative – even experimental – in the way we fund research. And just as we need to be prepared for research not to work out as planned, we should be prepared to take some risks in the way we support it, especially if the result is less bureaucracy. There are some lessons to take from the long (and, it needs to be stressed, not always successful) history of ARPA/DARPA. To start with its operating philosophy, an agency inspired by ARPA should be built around the vision of the programme managers. But the operating philosophy needs to be underpinned by as enduring mission and clarity about who the primary beneficiaries of the research should be. And finally, there needs to be a deep understanding of how the agency fits into a wider innovation landscape.

Most funding for science – in the UK and elsewhere – is based on one of two philosophies. The most common is based on competitive funding for projects. A researcher or group of researchers put in a research proposal outlining a project of work for a few years (somewhere between 2 and 5, typically), and this is judged by other scientists, in competition with other proposals. The topic may be completely open, or focused on some priority area. Most research council funding takes this form.

The other approach focuses on funding people – individual scientists. A judgement is made on the quality of a scientist applying for funding based on their track record of previous discoveries, and on this basis funding is awarded – usually for a relatively long time, and with a relatively open programme, with the expectation that the scientist will explore whatever avenues turn out to be interesting and productive. This is largely the approach used by the European Research Council, which has gained a very high reputation in its 12-year history, with the quality of applicants driven up by continent-wide competition. The longer-established Royal Society Research Chairs – whose previous holders include Sir Andrew Wiles, a mathematician notable for proving Fermat’s last theorem, and Sir Konstantin Novoselov, Nobel Laureate and co-discoverer of graphene – operate on similar principles.

The US defense research agency ARPA – whether by accident or design – hit on a third philosophy. J.C.R. Licklider ran the information technology programme at ARPA between 1962 and 1964, when the research he funded laid down most of the foundations for modern computing, including networking that led to the internet, and the principles of human/computer interaction that were further developed at the XEROX PARC laboratory to give us mice and graphic interfaces.

This third philosophy, then, focuses on the programme manager, or perhaps better, a programme leader. These are individuals who come with a vision of the future of some field of science of engineering, they develop and elaborate that vision – in collaboration with others – into a research programme, and then they create and support a research community to realise the vision.

But even before we define an operating philosophy, we should understand what the fundamental purpose of any new agency is, and who or what it should serve.

Part of ARPA’s success is complete clarity on both fronts. Its purpose was, and is, to ensure the technological superiority of the US armed forces. The people it serves – its customers or beneficiaries, if you will – are soldiers, sailors and airmen of those forces, as they carry out their task of projecting the military power of the USA. Others may well benefit from its investments – defense contractors, industry more widely, the whole economy, and ultimately everyone in the world who has benefitted from the internet, but these weren’t ARPA’s primary purpose.

So who should the primary customers of a UK version of ARPA be? It’s easiest to start by listing who should not be those primary customers. It shouldn’t be academia; this should not be about maintaining the health of the conventional disciplines. This is not to say that it isn’t important to maintain the continuing progress of scientific fields and the training of researchers in those fields – this is crucially important as the foundation for all scientific and technological progress. But this is the role of the research councils and funding agencies in partnership with the universities. The academic scientists and technologists who will generate the ideas and carrying out much of the research for our UK ARPA will emerge from this foundation.

Neither should industry be thought of as a primary beneficiary. Expertise from the private sector will be needed, and ultimately the ideas should create significant new commercial value – but this should be for tomorrow’s industries and tomorrow’s companies, rather than for the near term benefit of the current incumbents. Once again, there is a place for the state to support the nearer-term R&D of industry, but this should be carried out directly through innovation grants, and indirectly through R&D tax credits – together with much smarter use of government procurement to drive innovation.

This still leads us in search of the fundamental purpose of the UK ARPA – the equivalent of ARPA’s mission to ensure the technological supremacy of the US armed forces. This example suggests that to have longevity and political staying power, the purpose needs to be closely coupled to the strategic goals of the nation.

The cold war provided an existential threat to the US nation; the environment from which ARPA emerged was the shock that Sputnik posed to the USA’s technological self-confidence. The cold war is over, but the world remains a dangerous place and the future is even more uncertain than ever.

Direct threats to security from hostile state and non-state actors remain important. Moreover, new threats will emerge; we live at a time when the world is more connected than ever, so events in distant parts of the world could have direct, yet difficult to predict, consequences for the UK. Rapid changes in our natural environment, together with a paradoxical blend of economic stagnation and rapid technological change could combine with uncertain consequences.

The global physical environment is changing fast; anthropogenic climate change will certainly result in significant warming, while tail-risks from more extreme scenarios need to be taken seriously. How, for example, would the UK react to a powerful state in another part of the world unilaterally implementing geo-engineering measures that mitigate the effects of climate change there, but impact deleteriously on the UK’s own climate?

The UK has committed to reducing its own net greenhouse gas emissions to zero by 2050. This is the right thing to do, but there’s not a wide enough understanding of how demanding this target is for an advanced economy like the UK, whose economy still depends on fossil fuels for about 80% of its essential energy needs. And even if the UK did reach this target, what good would that do if major emerging economies insist on prioritising widening access to energy over decarbonising the sources of that energy? The solution to this dilemma can only come from innovation to drive down the cost and increase the scale of low carbon energy sources, and increase the efficiency with which it is used.

There are socio-economic threats too. We have seen a decade of stagnation in productivity growth unprecedented in the UK for more than a century, yet we have no consensus on how to end that. We have no intuition as to how the economy and society of the UK might respond to a no-growth world (we may need to look at the experiences of countries like Argentina to understand the threats that emerge in the scenario of continuing stagnation). There are specific technological issues that need to be faced. The unpredictable effects of rapid technological change in some areas – machine learning and ubiquitous computing – are widely discussed, but equally or even more important will be the consequences of the apparent slow-down of technological progress in some areas – the end of half a century of exponential growth in computing power, the dramatic slow-down in the rate of creation of new medicines[2].

An effective government needs to sponsor serious long-term thinking about the changing world and the UK’s place in it. There will be no better time to do this now, as the UK leaves the EU. Such a long-term look ahead would provide a framework for identifying the areas of greatest need for technological innovation, and joining up this kind of thinking should be a priority for this new government. The enduring goal of UK ARPA should emerge from this analysis.

But we can’t wait for this kind of thorough strategic forward-look at the threats the UK faces, necessary though it is. What practical steps do we need to do to get our UK version of ARPA up and running within a year or so, and what relationship should it have to the existing umbrella organisation for science and innovation funding, UK Research and Innovation?

There’s no compelling reason why the UK ARPA shouldn’t be set up within UKRI, and some good reasons why it should, particularly in the first instance. Most importantly, it can be done without delay – under the Higher Education and Research Act the Secretary of State has the power to instruct UKRI and provide new funds conditional on these instructions. Moreover, UKRI already has the necessary governance and administrative infrastructure to issue grants, and there’s no point rebuilding that. The existing processes can, with imagination and will, be easily be streamlined to reduce bureaucracy.

After being established in UKRI, should the UK ARPA subsequently be spun-out as an entity separate from UKRI? I would turn the question round – UKRI is a young organisation which itself needs to evolve further. Whether UKRI can comfortably accommodate an experimental and risk-tolerant organisation like UK ARPA will be a good test of its own evolving responsiveness and flexibility.

But the approach of a UK ARPA can and should be quite different to existing funding mechanisms within UKRI, especially the classic project-based funding used by the research councils. It is possible that, notwithstanding the already significant powers of the Secretary of State to change the structures of UKRI by regulation, further legislation will be needed to allow UK ARPA to adopt such mechanisms within the overall framework of UKRI, and if so the government should not hesitate to introduce this.

The priority should be to recruit programme leaders in a few areas. These leaders will need to have deep technical expertise, a vision for their research area, and the personal skills to build a wider research community around this vision. It should be obvious that such individuals will be rare and at a very high level – these are people who will probably be very accomplished academic scientists or technology industry leaders. The main task of the new director of the UK ARPA will be to design the conditions that will make these positions an attractive career move for such exceptional individuals.

The programmes that each of these individuals lead will probably be at the scale that would provide support for perhaps half a dozen research groups, with that support at a level that their leaders will be able to devote most of their time to the problem. The level of support should be sufficient to attract the most talented researchers from overseas, and this internationalism should of course be welcomed and facilitated. Over 5 years, each programme might have a budget of order £20-30 million if it is focused on laboratory science, rather more if it involves any technology demonstrators.

Setting up these programmes should be accompanied by considerable rigour and scrutiny, but once established the programme leaders should have very considerable discretion in assigning and redirecting the funding without further bureaucratic overheads. There will be risks in this approach that need to be foreseen and guarded against – programme leaders might have conflicts of interest or they might not look widely enough to build their research communities, relying too much on existing networks.

Finally we need to think about the place of UK ARPA within the UK’s wider innovation landscape. That landscape needs to change, to achieve the widely shared goal of making the UK a significantly more R&D intensive economy.

It’s important to remember that the UK ARPA will always be a relatively small part of a much bigger system, just as DARPA itself is only a small part of a much bigger US innovation system. DARPA’s 2015 budget was $2.9 billion out of a total R&D expenditure by the US Federal Government of $121 billion . In round numbers, UKRI needs to have something like a £10 billion budget if the UK is to meet its 2.4% R&D intensity target, so a £200 million budget for UK ARPA would make it occupy a comparable niche in the UK system.

DARPA, like any funding agency, relies on other people and organisations to actually to the work that its success depends on. In the US, this involves a rich system, including university researchers, government laboratories, but also a strong degree of private sector involvement, including big companies like the defense contractors which in the US carry out so much directly federally funded R&D. But there is also an important tier of what are essentially research SMEs – private, sometimes not-for-profit, contract R&D laboratories like SRI International (formerly the Stanford Research Institute). These have no direct UK counterpart, though the R&D consultancies and design houses like Cambridge Consultants are perhaps the closest analogues[3].

The UK’s innovation system needs to change, to reflect some wider shortcomings of the existing landscape – in particular a relative neglect of translational research, and an unhealthy degree of geographical concentration that contributes significantly to the UK’s regional economic imbalances . But as we change the landscape, those changes can be shaped to make the UK ARPA more effective.

A well-designed UK ARPA could be an important part of the wider innovation system, part of a necessary process of being more experimental in the way the UK government supports science and technology. We should learn from the best of the US experience, focusing on visionary programme leaders given the freedom to create and support research communities to develop those visions. The agency should have a strong and enduring clarity of purpose, set within a long-term view of the strategic threats and opportunities facing the nation. The development of the new agency should also be set into a context of a wider reshaping of the UK’s innovation system, with more emphasis on translational research and less geographical concentration.

[1] https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/731507/research-innovation-funding-allocation-2017-2021.pdf

[2] See e.g. Bloom N., Jones, C.I., van Reenan, J., Webb M., Are ideas getting harder to find?
https://web.stanford.edu/~chadj/IdeaPF.pdf,
Jones R.A.L., and Wilsdon, J., The Biomedical
Bubble: Why UK research and innovation needs a greater diversity of priorities, politics, places and people, NESTA 2018, https://media.nesta.org.uk/documents/The_Biomedical_Bubble_v6.pdf,

[3] See e.g. Connell, D. and Probert, J. (2010), Exploding the Myths of UK Innovation Policy: How ‘Soft Companies’ and R&D Contracts for Customers Drive the Growth of the Hi-Tech Economy, CBR, University of Cambridge.
https://www.cbr.cam.ac.uk/fileadmin/user_upload/centre-for-business-research/downloads/special-reports/specialreport-explodingthemyths.pdf

3 thoughts on “UK ARPA: An experiment in science policy?”

  1. Richard, this is great stuff and, you are right: the full compendium is definitely worth reading. Lots of historical background, a strong feeling of enthusiasm and urgency and, of course, very articulate. (do you think the sentence on p.19 “Do we need this last sentence, or should it be rephrased?” was included intentionally to give it a hot off the press feel?)

    The idea of making the NHS and regional councils the “customers” seemed pretty wild at first. But in some ways it makes a lot of sense, certainly ticks various political boxes.

    The point David Willetts made about making students specialise narrowly at 16 leads to bad politicians and civil servants was very persuasive. I definitely don’t get the impression that the UK universities are pushing for a more general education – they seem to want the opposite.

  2. I hope that the new government will lead the UK to a brighter future.

    Thank as always for your commentary.

  3. Thanks Patrick. But which of us has not had an editing glitch working to a deadline? It really was done very quickly so no need to make intentional mistakes!

    You’re quite right about Universities – science and engineering departments are particularly bad at wanting students who are already focused and specialised.

Comments are closed.