As the UK prepares to launch a new high-risk, high-reward research agency, Eoin O'Sullivan, director of the IfM’s Centre for Science, Technology & Innovation Policy (CSTI), speaks to Jason Naselli about what it can learn from the ARPA model in the US and globally, and why technology and innovation management is the key to success.
The UK government has announced the establishment of a new technical innovation research agency, the Advanced Research and Invention Agency (ARIA), which will be ‘tasked with funding high-risk research that offers the chance of high rewards, supporting ground-breaking discoveries that could transform people’s lives for the better’.
Based on models from other countries, primarily the US Advanced Research Projects Agency (ARPA) model, the government says that the new agency will aim ‘to deliver funding to the UK’s most pioneering researchers flexibly and at speed, in a way that best supports their work and avoids unnecessary bureaucracy’.
Eoin O’Sullivan discusses what elements are important to ensuring the new agency is successful, what can be learned from other models around the world, and what the IfM and CIIP are doing to further the discussion around challenge-led research agencies, in conversation with Jason Naselli.
What can challenge-led agencies like ARIA do that other research and development approaches cannot?
Well, the high-level view is often that if you pull back the bureaucracy, if you get rid of red tape, if you allow a greater tolerance for failure and risk, and if you can empower charismatic, smart people, you can get more technological impact more quickly.
Countries which have adopted this approach, most notably the US Department of Defense agency DARPA [Defense Advanced Research Projects Agency], have made remarkable technological breakthroughs, gained competitive advantage and achieved economic impact in key emerging technology domains.
But how accurate is that story?
All of those things – low bureaucracy, tolerance of risk, empowered programme managers, missions – are necessary and useful in enabling conditions that can help catalyse technological innovation. But that’s only part of the story.
Looking at this issue from the IfM point of view, through our research in technology and innovation management, we can see that these R&D project management agencies are doing other things differently as well; they have different types of people, and the way they’re setting goals, engaging with the research and innovation community, and managing portfolios of projects is different.
My argument would be that these agencies can make a difference for emerging technological innovation because their programmes are hitting different points of the technology life cycle. They’re nurturing R&D networks; they’re addressing key bottlenecks to translation and scale-up of technologies; and they’re revealing certain types of technology development pathways.
So, the unique thing is how well they do all that. Much of the genius of ARPA is in the innovation management details of how the run their programmes. For them to do that well, they need low bureaucracy. They need to be able to, for instance, commission a specific prototype quickly from the best team without having to go through a long conventional tender process.
But it’s not as simple as saying: remove the red tape, empower geniuses, give them a mission, and magic happens. In fact, there is something very particularly effective and transformative in the process of how they do technology and innovation management.
You’re saying that the actual project management approach of successful agencies is different and important.
Yes. Through the frameworks we’ve developed to understand innovation, what stands out is how the US ARPA agencies really make a difference at a certain phase of the emerging technologies life cycle. If you look at the big successes they’ve had, what they’ve managed to do is take a promising new area of science and technology, where people are just starting to explore how it might be applied and exploited, and they do something extraordinary at that point where exploration and exploitation activities are strongly coupled.
This is the stage where researchers are still discovering new science, but the people trying to apply it are also coming up against lots of new science. Those doing the early application are coming up against the limit of our understanding of the science, so they need to go back to the discovery people to ask further questions about the underpinning science governing why and how the technology works.
This creates really fast feedback loops between those doing the early application and the people still doing the science, where their expertise is necessary to diagnose any barriers to application that the technology development people are encountering. This is a real sweet spot in the life cycle of a technology, and it’s where this type of agency programme can make a real difference in accelerating technology impact.
It’s important to understand that in order to identify which challenges are particularly suited to the challenge-led innovation approach. If you’re looking for more applied science and early technology research, universities are doing this very well. If you’re looking too far forward in commercial application, where the science and understanding of a technology is much more stable, industry R&D labs can do that. It’s not clear to me that all exciting science and technology problems necessarily lend themselves to what the ARPA-like agencies have been able to do.
It’s not like high-risk, high-reward programmes haven’t been tried before; it’s not like there haven’t been other attempts to give more freedom to programme directors and remove red tape. Many other countries have tried this as well. And it’s clearly not always sufficient.
Missions, low bureaucracy, tolerance of risk… these are clearly necessary enabling conditions. But a key element seems to be successful innovation management of these technology transitions.
When you look at some of the ARPA agencies in the US, they all have a particular focus, such as defence, energy or intelligence. But you’re talking about how important it is to understand which problems are able to be addressed by an agency like this. How does ARIA go about defining a focus and a mission?
One of the complications is that how you pick your focus operates at different levels. The mission is important. But missions by themselves are not magical things.
Why do missions matter? A critical thing they seem to provide is bringing coherence to a complex endeavour. Part of what’s going on is that you are exploring multiple innovation pathways, looking for all the bottlenecks that might be stopping the translation of these new technologies into challenge solutions. You’re nurturing new research and innovation networks that are configured in different ways.
So if you have a mission, as everyone decides what their piece of the puzzle is, they can check that against the mission. The mission brings coherence to this activity, helps makes the process more joined up and the embryonic innovation system more connected and efficient.
And, of course, there are counterfactuals to the ARPA model. The most obvious is Bell Labs – the great corporate R&D lab of the AT&T corporation, where the transistor and laser were invented, whose researchers received eight Nobel Prizes for physics.
Bell Labs didn’t have a public mission, but they had coherence that came from being part of the dominant, vertically integrated telecoms giant (and a monopoly for much of the 20th century). They effectively had all of the pieces of that industry either within their supply chain or within the large AT&T system. So, the coherence came from the goals of the industry and the clarity of purpose of the corporation, rather than a policy or socioeconomic mission.
One could make a convincing argument that Bell Labs had far more impact – in terms of important technological breakthroughs – in the latter half of the 20th century than DARPA did. So mission matters, but partly because coherence matters – coherence of vision for the future technological system, as well the wherewithal to pull the technologies through to deployment within that system.
What are other important elements beyond coherence?
Another important factor is the need to have a client or a community or set of organisations that can pull the technology through. In this sense, DARPA is maybe not the best analogue, because it is part of the US Department of Defense, which has an unbelievably large development, deployment, scale-up and industrialisation budget. DARPA is a small part of that, but at the other side of DARPA is a huge budget to pull the technology through and demonstrate it and deploy it.
That’s why we should look at other ARPA clones like ARPA-E, which is part of the US Department of Energy. The Department of Energy doesn’t have that same pull-through budget, so ARPA-E has had to work a lot harder to understand how to, for instance, engage with venture capital. They had to stand up a technology-to-market team that would work on a lot of these issues. And it really became a technology and innovation management problem to ensure they had impact.
Some of the work we’re doing at the IfM on understanding private sector investment in technology and issues of technology development transitions shows how important this is. For ARIA to work, those issues need to be managed.
It’s interesting to focus on clones like ARPA-E, which are somewhat less well known. The trailing promotion for these agencies is the big breakthroughs and transformational technologies, such as the early internet, which came out of DARPA. But ARPA-E and others, while they have their successes, don’t have those big world-changing breakthroughs to point to. So, if ARIA is modelled more like them, do expectations need to be managed?
Let’s remember that DARPA has been around longer. There are many big DARPA successes to point to, but the biggest and most obvious one is the internet. They started working on that in the late 1960s, but it really only broke through in the early 1990s.
All technologies have that kind of life cycle. The reality is that if you look hard at it, even the original ARPA was building on research that had already been happening before it came into existence in the late 1950s. So, if the key phase in the emerging technology life cycle is 15 to 20 years or more, it’s too early to judge something like ARPA-E, which was set up in 2009.
You’ve got to have the urgency to pull the technologies through, but you have to understand that the impact will not be visible in the short term.
There of course is low-hanging fruit out there that could generate success sooner, but if you start chasing low-hanging fruit, you change the culture in a certain way. This is what we’re hearing in discussions with people who have worked with ARPA agencies in the past – the culture sets in very early. And as soon as you start feeding in goals to capture ‘quick wins’, you don’t set yourself up to have the really big impacts, which should be the main focus.
It’s interesting, because you began by saying that simply loosening bureaucracy and letting insightful, smart people work unfettered is not enough. But by the same token, it appears that leadership matters a lot in these agencies.
Absolutely. The key thing is that the management, governance and leadership really matters. And it matters at different levels.
It’s all about innovation management. These agencies are essentially innovation project management agencies, which makes it very interesting from the IfM perspective and is why we are focusing on furthering discussion and research in this area.
What conversations will you be having through CIIP and across the IfM in the coming months? What are the other issues you will be exploring?
An important initiative from CIIP has been the policy practitioner workshops we’ve convened. There aren’t enough forums for the team that is setting up ARIA, or other challenge initiatives like the Industrial Strategy Challenge Fund, to come together, to share lessons and practices and not have to reinvent the wheel.
There are a lot of lessons and experiences to be learned from around the world, not just from the US. Getting into the operational details with those who have been there – how you design a strategy, set priorities, manage projects, set goals – that’s really important. That’s one of the services we provide, along with the technology and innovation management capabilities we have ourselves.
We began this discussion at a roundtable event we held with practitioners in February, and a big question that came across was how to effectively engage with the rest of the research and innovation system in the UK.
In an effort to rationalise the existence of ARIA, the danger is that you try to put too much clear blue water between ARIA and the existing UK research infrastructure. Yes, ARIA needs to be insulated from day-to-day concerns of government, but their ability to do what they do comes from being well connected to the rest of the system – to know what other people are investing in, what bottlenecks they observe, what opportunities they have. ARIA will have to leverage capabilities and budget, facilities and resources across the system.
Again, this is an innovation management problem, and very much plays on the insights we know from the work the IfM does around technology management and research and innovation policy. In my research group, the Centre for Science, Technology and Innovation Policy (CSTI), we’ll be exploring approaches to operationalizing mission-oriented innovation policies and managing challenge-led technology R&D programmes. We’ll also be looking at adapting foresight and roadmapping frameworks and processes to support challenge-led R&D programme design and management.
Another issue is metrics – monitoring and evaluation on the right kind of timelines. How do you do this in any agency where you are putting a premium on low bureaucracy and empowering programme managers? How do you do that for an agency that is high-risk, high-reward, where impact is over a longer period of time? This deserves more attention.
Finally, a critical dimension is the people. The idea that you have to get the right people is so central to the whole conversation. Why are these challenge-led research agencies different? One way you can tell they are different is there are many different types of people working there. You consistently encounter people who understand the discovery and exploration process as well as the development and application process. Understanding the career experience, expertise and capabilities needed – this is another theme we’re very interested in and will be exploring in ongoing research projects.
The key for ARIA right now is the operational phase. The decision has been taken to do it – all these insights will now inform not just how it will develop but what is needed right now to ensure it is a success. That’s a conversation we will continue to have through CIIP and the IfM.
Read Eoin's submission to the House of Commons Science & Technology Committee inquiry into 'A new UK research funding agency'