ARIA is Dominic Cummings life’s work. His legacy.

What became apparent through the hearing was that the creation of a science funding organisation with “extreme freedom” has been the end goal of a storied and bizarre political career.

Why Vote Leave? – well, only by leaving the EU could we avoid the bureaucracy of EU procurement rules. Why become Boris’ right-hand man? To secure the doubling of the science budget, changing the civil service to allow for a new approach to science funding, and establishing an ARPA-like entity. Oh, and committing to actually leave the EU. These were the four promises Cummings extracted from Boris Johnson shortly after he became prime minister.

Cummings’ story is now our story. The last five years of British history have led inexorably to the Advanced Research and Invention Agency. The diagram of the blob inside a larger blob (representing the need to “sample the wider space” of innovation methodologies) is the blueprint that brought us to the current political reality.

(“Sample the wider space”, incidentally, is that rarest of beasts. A googlewhack. It has never been said before today.)

Taking back control

In Cummings’ telling, most of the work he did at number 10 was around science funding policy. He detailed the battles with the Treasury over the possibility of an agency that worked outside of normal procurement rules and the need for the primary legislation we now have before us. He’s not as happy with the bill as it stands – too much ministerial involvement.

But he also put in what he would probably call “the hard yards” with conventional science funding – numerous meetings with UKRI (Ottoline Leyser got a number of namechecks) yielded the end of the impact assessment.

Cummings was scathing about the way “the system” (yes, we are at that level of analysis) has captured science, funding, and science administration. At one point, he told the committee that not one person has the power to stop or change the research funding system – even the Prime Minister and his office. This may come as a surprise to Kwasi Kwarteng, who appears to be undertaking another of these indeterminable reviews of bureaucracy in science funding as we speak.

And the “sixties ARPA model” is the solution. Or part of the solution, we never really got an answer as to why Cummings would as a condition of his employment demand that billions of pounds is poured into a system that is not in his eyes fit for purpose.

Ahistorical innovation

As I’ve discussed elsewhere, even narrowing your net to “sixties ARPA” captures a number of sins. The main interest of the organisation at that point what some highly dubious actions in Vietnam – Joseph Licklider’s work on networking and computer science (conducted, I feel obliged to add, in collaboration with industry and academia from the start and with a sizable conceptual debt to Vannever Bush’s 1945 “As we may think”) was a poorly-regarded sideline for an agency primarily focused at the time on counter-insurgency.

The central problem with “extreme freedom” is that it doesn’t exist – even an organisation as loosely defined as ARIA will generate cultures and eventually processes to steer development in ways deemed to be useful. It is possible – like Licklider – to sneak in other priorities under the radar, but the same is true in any organisation.

Cumming’s ahistoricism is backed up with an equal ignorance of modern science funding – he told the committee, apparently in all seriousness, that current systems were unable to fund anything with an output that is not an academic paper. To pick an example of recent UK research output at random, the Oxford/Astra-Zeneca Covid-19 vaccine is not an academic paper and was developed by the Jenner Institute which is funded by – among others – MRC, the European Union, and the Department of Health.

He was clear that ARIA needed to be seperate from UKRI – going as far to say that even a remote association would make it better not to establish the new organisation at all, and adding that Leyser agreed with this assessment. Commendably, he was equally clear that he did not seek and would not accept a role for Dominic Cummings in ARIA – naming instead a range of his scientific heroes (Michael Nielsen, Tim Gowers…) as potential leaders. A “bog standard” vice chancellor would be a bad idea.

Cummings’ final discussion with Boris was on science policy – he got another promise, this time that the Cabinet Office would take a more active central interest.

The science of politics

There was a lot else in there for seasoned Cummings-watchers – Department of Health procurement as a “smouldering disaster”, a nuanced defence of his apparent pay-rise. Nothing, oddly, on research ethics or the links between defence and research. “The Manhattan Project”, we learned, “had a clear target”. Hiroshima?

The £800m for ARIA was only a starting point, fine to begin with but he’d expect £3-5bn in future. And 66 per cent of projects – possibly more – would be expected to fail. All this should come alongside the expansion of research funding, but the latter should be conditional on both the system more generally and universities themselves cutting barriers and bureaucracy. He warned against “accountancy gimmicks” in double counting research funding – he was clear that all increases need to be real increases.

Kwasi Kwarteng and Jo Shanmugalingam were the support act – confirming that both research funding targets (2.4% of GDP by 2027 and £22bn by 2024) still stood. However, discussions about the UKRI budget are currently stalled between BEIS and the Treasury, with the spectre of taking Horizon affiliation costs out of the current budget very much still on the cards. We’ll hear in the next two weeks – the whole episode proving an early Cummings point about the need for long term planning and stability in research funding.

Kwarteng is less absolutist about ARIA than Cummings – he supported the need for departmental and parliamentary oversight in section 1. Recruitment for the Chair of ARIA will start in April or May – there are no preconceptions about who a suitable candidate may be. He was clear that there needs to be a skills balance between the Chair and Chief Executive, backed up by the rest of the board. Although ARIA could be “slightly whacky” there is still a need for corporate governance structures.

The new agency can set its own priorities, but BEIS would set “tramlines” based on the existing “grand challenges”. ARIA could then choose to adopt, modify, or ignore them. But hang on to your hats, because Dom promised another blog post.