Joe Tomlinson: Government as Skunk Works

The legislative programme of this Government has—unsurprisingly, given its widely stated reformist ambitions—attracted close attention from those interested in constitutional matters. Several bills have been introduced into this Parliament that have (rightly) been the subject of extensive commentary and interest in this respect, including the Judicial Review and Courts Bill, the Dissolution and Calling of Parliament Bill, and the European Union (Withdrawal Agreement) Act 2020. However, there is another legislative initiative which has attracted very little commentary from those interested in constitutional matters, which may take us closer to understanding an important driving force behind the period of constitutional instability and change we are experiencing: the Advanced Research and Invention Agency (ARIA) Bill

The ARIA Bill was the pet project of Dominic Cummings. Controversial leader of the Vote Leave campaign during the Brexit referendum and then Chief Adviser to Prime Minister Boris Johnson, Cummings left Downing Street in November 2020 and has since become a vocal critic of the government he served. The myths and the reality surrounding Cummings are difficult to separate, but he has clearly exerted great influence on the politics of the last decade, even if the precise extent of that influence—and whether it is a welcome one—may be disputed. However, there are at least two consistent themes that Cummings has espoused for many years, and long before the word “Brexit” was even part of the mainstream lexicon. The first is that the UK should seek to become the world leader in scientific research and innovation. The second is that the machinery of government in the UK does not work and is incapable of tackling the pressing problems facing society. Both of these themes were on full display in a lecture Cummings gave at the Institute for Public Policy Research in 2014, entitled The Hollow Men (also see here).

In The Hollow Men, Cummings sets out his view that there has been sustained and catastrophic failure in government decision-making since the 1860s. This, in his view, is because “those at the apex of British politics made colossal error after error.” He provides harsh and wide-ranging criticism. Thatcher is fleetingly praised for dealing with “some of the worst excesses of accumulated errors and weakness” but is still assessed to have failed on a number of fronts, including “monetary policy, Europe, health, education, and welfare.” Ahead of the 2015 election, Cummings concluded it did not matter if voters chose Ed Miliband, then Leader of the Labour Party, or David Cameron, the incumbent Prime Minister. To Cummings, the result made no real difference as those leaders and those surrounding them were “hollow men”—a reference to the opening phrase of the T.S. Elliot poem of the same name. According to Cummings, these “hollow men,” from across the political spectrum, fail to “develop political institutions able to think wisely about the biggest problems in order to pre-empt some crises.” He concludes that future reforms should be underpinned by “a new national goal and organising principle” to “focus on making ourselves the leading country for education and science.” This goal, he claims, should shape the entire policy agenda and determine resource allocation, as well as providing a frame through which to approach the reconstruction of state institutions.

In The Hollow Men, the role of public law may be said to arise surprisingly often for a policy lecture. Law, lawyers, and courts are positioned by Cummings as a sustaining feature of what he perceives to be the catastrophic failure of our system of government. There are broad, abstract references to the need to reform public law, the constitution, and government systems. For instance, Cummings claims that the “new national goal and organising principle” (i.e. “making ourselves the leading country for education and science”) would “require and enable fundamental changes to how the constitution, Parliament, and Whitehall work.” He also provides various personal anecdotes of frustration from his time as a Special Adviser or ‘SPAD’ in the Department for Education, working under Michael Gove. For instance, he provides this example of an exchange about procurement with an official:

[Bang bang on the glass door, another official enters looking nervous, glances at the second official…] 

Official: Err, Dom, you know that contract we were talking about yesterday?

Me: Don’t tell me the tests have gone haywire.

Official: Yes they have but that’s not what I mean – I mean that Academy procurement process.

Me: Yes.

Official: Well, the legal advice says – if we go ahead, we’ll get JRd [judicially reviewed] and lose but if we stop and reboot we’ll also get JRd and lose.

Me: So we’re screwed whatever we do.

Official: Seems like it.

A second example is given in the context of a broad critique of “bureaucracies” that “cannot reliably do the simplest things.” Included in a string of complaints—alongside the claim that “spreadsheet skills were so lacking that financial models and budgets could never be trusted and almost every figure released to the media or Parliament was wrong”—is that “[l]egal advice was unreliable and government lawyers are also given the wrong incentive (they are told to prioritise never going to court, which is stupid).” According to Cummings, the wide array of competency issues within bureaucracies are:

[C]ompounded by a combination of the growth of public law, judicial review, EU regulation, and the ECHR/HRA, which have added cost, complexity, and uncertainty. There is no objective view of ‘what the law is’ in many circumstances so management decisions are undermined many times per day by advice to do things ‘to avoid losing a judicial review’ the risks of which are impossible to analyse clearly. Legal advice is offered saying that both doing X and not doing X could be ‘illegal’ leading to Kafka-esque discussions and pseudo-‘fair processes’ (like ‘consultations’) designed only to be evidence in court. Internal legal advice makes discussion of regulatory trade-offs tortuous and wasteful; it is always easier to urge ‘caution’ and ‘we’ll lose a JR’ is an easy way across Whitehall to delay or block change.

This is public law and the constitution cast in the role of undermining the way government ought to work and prohibiting it from taking on the problems that it should be tackling.

Given Cummings’ views on the role of science, innovation, and the functioning of public law and government, it should be no surprise that he told the Science and Technology Select Committee, on the topic of being asked to join the No. 10 team by Mr Johnson, that his response was

Yes, if – first of all – you’re deadly serious about actually getting Brexit done and avoiding a second referendum. Second, double the science budget, third, create some ARPA-like entity and, fourth, support me in trying to change how Whitehall works and the Cabinet Office work because it’s a disaster zone.

As a result, he then spent a significant amount of his time in No. 10 working on ARIA and the underpinning Bill.

The ARIA Bill establishes a new research funding agency, with a focus on providing long-term support for “high-risk, high-pay off” research. It promotes research which may not have obvious, immediate application but has the potential to be transformative over time. As part of the March 2020 Budget, a commitment was made to invest “at least £800 million” in ARIA, which formed a part of raising broader investment in public funding for research and development to £22bn by 2024/2025. 

The Statement of Policy Intent published alongside the Bill stressed ARIA should have certain distinctive organisational features. First, ARIA will exclusively focus on projects with the potential to produce transformative technological change, or a paradigm-shift in an area of science. While it is anticipated that most programmes may fail in achieving their ambitious aims, the theory is that those which succeed may have a profound and positive impact on society. The bulk of the money awarded by this body is therefore not likely or expected to lead to the desired outcomes and official decisions will be made within that framework. Second, ARIA will have extensive strategic, scientific, and cultural autonomy. This means it will have “maximum autonomy over its research and project choice; its procedures; and its institutional culture.” Decisions on the programme portfolio will be set by ARIA, not ministers, and allocation of funding to research projects will be decided by those with relevant technical expertise. Third, ARIA will give freedom and control to a small number of the “highest-calibre” researchers, who will be taken from both the public and private spheres. These individuals, filling the “Programme Manager” position, will be “empowered to dynamically channel funding, shift project objectives and milestones, and manage risks, to keep their overall research programme focussed around a coherent but evolving vision.” Fourth, ARIA will have extensive financial and operational freedom. This means there will be a focus on minimising hurdles across a typical project lifecycle to create an “agile” and efficient funding body. For instance, it will likely issue small grants rapidly without lengthy, open competitions.

Overall, ARIA is based on an organisational model where a small number of expert people have an enormous degree of operational freedom and discretion. The ARIA Bill, which is relatively short, reflects this policy intent and organisational design. It establishes the Agency, sets out its functions as conducting and commissioning scientific research, sharing findings, and exploiting scientific knowledge. It also explicitly states that ARIA is permitted to undertake “ambitious” research projects with a high tolerance for project failure. Amongst other provisions, the Bill seeks to exclude the Agency from the requirements of the Freedom of Information Act 2000 and the Public Contract Regulations 2015. While the principle of further investment in research, science, and innovation has been broadly welcomed across the political spectrum, there have been concerns expressed in Parliament about the lack of accountability within AIRA and the concentration of power within a few individuals (for an overview of the response in Parliament, see here).

The model being adopted for ARIA reflects and is derived from experiences of organisations responsible for effective and transformative innovation in science and technology in the US. Two are particularly prominent. The first is Skunk Works—the nickname for Lockheed Martin’s Advanced Development Programs, which was responsible for a series of transformative technological breakthroughs, particularly through their work in the field of aeronautical engineering (for helpful accounts, see here and here). Skunk Works has now become shorthand for a particular way of setting up an organisation: where a relatively small and loosely structured group of people, with a great degree of operational freedom, develop a project in the pursuit of radical innovation. Skunk Works founder, the famed engineer Kelly Johnson, is now widely credited with creating a new form of organisational management. Sometimes cited as the creator the so-called KISS principle (“keep it simple, stupid,”), amongst his “14 key rules” were: the Skunk Works manager must be delegated practically complete control of his work in all aspects; the number of people having any connection with the project must be restricted in an almost vicious manner, so use a small number of good people; and there must be a minimum number of reports required, but important work must be recorded thoroughly.

The second is the Advanced Research Projects Agency (ARPA)—which has been heavily cited in the UK government policy papers around the creation of ARIA (for commentary, see here). Established by President Eisenhower in 1958, ARPA’s mission was to make investments in breakthrough technologies for the purposes of national security (it later became DARPA, adding “Defense” explicitly to its name). It has become famous for work that enabled innovations such as the internet, GPS, and self-driving cars (for an account of its work, see here and here). Discussion around ARPA in the context of ARIA has often focused on the period between 1962 and 1975, and the work of Joseph Licklider in The Information Processing Techniques Office (a sub-division of the agency). During this period, ARPA is perceived to have been highly productive and this has been widely seen as a result of Licklider’s organisational strategy, which was akin to that within Skunk Works. It included reducing process, recruiting a small group of exceptional people, and giving those people significant operational freedom and control.

What does all of this tell us about present constitutional tensions? It is, of course, now trite to observe that politics, society, economy, and public law are inevitably interlinked, and that any vision for constitutional reform will also relate to those broader issues. However, another important influence on constitutional change—and one that is often understated—is the preferred organisational configuration of administrators. The form of administrative systems evolves over time and a government’s desire to adopt a particular organisational model can create constitutional friction. It is in recognising this dynamic that the ARIA Bill’s constitutional significance ought to become more apparent.

In narrow terms, ARIA is the most explicit attempt to create a public body in the style of organisation favoured by Cummings and questions are arising about the accountability structures of the Agency itself. However, the wider point is that, for Cummings, Skunk Works and ARPA are not just models for research and innovation bodies, but they also hold more broadly applicable lessons for building effective government institutions that tackle the big problems and achieve outcomes. Given the nature of that model, it is hardly surprising that it sits uncomfortably with those committed to something akin to the UK’s contemporary public law system.

How much government more widely has absorbed interest in and even commitment to this organisational model, beyond Cummings and ARIA, is yet to be seen. If its promotion was the result of the position and world view of one individual (or a small number), the connections sketched out here may be limited to a small contribution to understanding a short and volatile period of recent constitutional history, and the emergence of an experimental public body. Yet, there are already signs that commitment to this model—or at least some of its ways of viewing what is required to deliver good government—has lasted beyond the person who was its main exponent while working in No. 10, not least in the growing expression of political desire to set up government to “gets things done” and frustration with traditional administrative process. If the influence of the model is wider, the ARIA Bill may be of greater importance to understanding our contemporary constitutional conditions than has been recognised thus far.

I am grateful to Colm O’Cinneide, Elizabeth O’Loughlin, and Lee Marsons for helpful comments on various drafts.

Dr Joe Tomlinson is Senior Lecturer in Public Law at the University of York

(Suggested citation: J. Tomlinson, ‘Government as Skunk Works’, U.K. Const. L. Blog (13th October 2021) (available at