I recently wrote a piece arguing that most countries were too small or too stingy to successfully set up their own version of DARPA.

But it’s very easy to tell people what not to do. But having smugly said that DARPA-clones are generally not a good idea, I ought to at least try to provide an alternative.

So then, if a government or department wants to make a big push on innovation, doesn’t have enough money to set up a DARPA ($3bn+), and is willing to consider new institutions – what should it do instead?

Let’s consider some of the appealing aspects of DARPA we might want to copy in whatever we do:

• Firstly, it invests in innovative, often radical, ideas.
• Secondly, it connects the part of government it works with (for DARPA, the defence sector) with unusual and disruptive technologies – it creates porosity in a sector that can be quite insular.
• Thirdly, it thinks beyond pure research. By connecting innovations to major public and private supply chains, it provides a potential source of demand for innovations (market shaping rather than just market-failure-fixing, in Mariana Mazzucato’s words.)
• Fourthly, it uses a range of innovative methods. DARPA was an early user of challenge prizes, and is known to be a tough and hands-on manager of grants.

At the same time, we can’t run an operation as big and as blue-sky as DARPA itself, since (as we discussed in the last post), this requires a scale and a risk tolerance that most governments don’t have.

Thinking about this, we came up with a more practical alternative, which we’ve tentatively called the Advanced Systems Agency (suggestions for a better name gladly received).

The point of an Advanced Systems Agency is (a) to tie together innovation policy efforts in a particular department or field with wider policy goals, to encourage radical innovation and (b) to build links between a department’s long-term strategy and procurement plans with innovations and developments going on in the wider world.

The key elements are:

• Set up a unit whose responsibility is to promote innovation to meet the department’s needs, and to grow the ability of businesses (and civil society) to help meet them.
• The unit should be directly in charge of innovation funding: research funding, innovative procurement (like SBRI) and other projects like business acceleration efforts or Enterprise Capital Funds. (It’s important that at least some of these are designed to make the agency receptive to external ideas: the model used by Entrepreneur First, IDEALondon and Hack Train that combines acceleration with open innovation is one way of doing this; futures exercises and challenge prizes are others.)
• The unit should be closely connected the department’s long-term strategic plans, both so that they can design innovation interventions that meet the department’s needs, and so that they can help shape the strategy by providing an idea of the potential of future innovations. While there’s a case for doing funding some innovation in a wide range of areas, it should focus a decent chunk of its resources in areas linked to the big bets the department is implicitly making.
• To do this well, it needs a good basis of innovation analytics – an awareness of where good research and business innovation is going on, capability to think about the future and to use that to prioritise and be open to disruptive shocks (I find the use of innovation analytics tools like those from Quid interesting here).
• The unit should have a good link to policymakers crafting regulation in the sector: this is an important driver of innovation but generally suffers from a lack of innovation input.

It’s not just what the organization does that matters, it’s also how it does it. This requires political support and the right skills.

Political endorsement is important if the agency is to be able to work effective with the Department’s wider plans and strategy. Otherwise it’s likely to be ignored. So it’s really only worth attempting if innovation is a priority for the Secretary of State responsible for the area, and if s/he is willing to give it this kind of support.

It also matters because innovative procurement and research involve some risk – both of being seen to have wasted money if things don’t work out (which innovation sometimes doesn’t), or of having procured something improperly (since delivering innovation is not the primary objective of procurement law).

What’s more, this kind of systemic innovation policy is relatively long-term relatively hard to evidence, at least compared to more mechanical interventions like R&D tax credits. This takes some political confidence: if all you have from your finance ministry is a year’s cash and a demand for a successful RCT within 12 months, this is not the policy for you.

The team also needs the right skills. It goes without saying that technical expertise is indispensable. But a team of only technical experts will not do a good job of running and Advanced Systems Agency.

The team also needs to be good at building networks (Oli Barrett’s idea of a government networker springs to mind here), and at plugging into the wider strategy of the department and the sector, which requires policy skills. The mix of government insiders and outsiders brought together in the Government Digital Service, or, looking further back, in Partnerships UK, could be one way to do this.

So there it is: if you want a DARPA, but can’t afford the lavish expenditure required, don’t build a half-heartedly copy. Try an Advanced Systems Agency instead. Though perhaps you may want to think up a better name for it.

This kind of policymaking is very difficult to evidence. New innovation institutions are complex interventions, and therefore hard to evaluate. (If you’re interested in the overall evidence for innovation policies, check out the review we worked on with the University of Manchester.) So this proposal is necessarily somewhat speculative.