Ryan Bourne holds the R Evan Scharf Chair in Public Understanding of Economics at the Cato Institute.

It is fast becoming one of the most discussed employment advertisements ever. Dominic Cummings’ call for No. 10 job applications from data scientists and software developers, quirky economists and policy experts, and other assorted “weirdos and misfits” has drawn reaction from global public intellectuals, former civil servants, and experience-weary ex-government digital experts.

Now, civil service employment practices, SpAds’ competences, and government project management, are outside my “circle of competence,” as Cummings might say. Without in-house experience, I’ll avoid passing judgment on whether better physics and maths skills or expertise in the economics of auctions might improve Downing Street’s performance. Others who I respect seem positive about his proposals, particularly as it pertains to quantitative skills and institutionalising analysis of uncertainty.

Likewise, some of Cummings’ broad proposals for Civil Service reform seem sensible as an outsider: “red teams” to push back on conventional wisdom; rewards for officials actually seeing through on delivering government projects; redundancy for poorly performing officials; more specialism, less generalism. All have clear rationales, though are easier said than delivered. And some tensions clearly exist between them. Greater longevity for brilliantly specialist civil servants, for example, surely creates an institutional impediment to radically adaptive change.

Given all the ink spilt debating these ideas, however, an obvious point has been missed. Cummings’ ideas here are for personnel and structural changes to a technocracy. For “better” management and people to deliver systems for a large enterprise (the state). They do not tell us anything, per se, about what he perceives to be the correct role of government – of when it should act, what it should do, and why. Yet without knowing what recruits and the civil service will be working on, it’s impossible to assess claims of the supposed “trillions of dollars lying on the sidewalk” from the “low-hanging fruit” of improved government performance.

Yes, yes, we have breadcrumbs signalling towards certain “ends.” This site’s editor thinks Cummings seeks a world of politics as “enterprise association,” harnessing AI, science, big data, cutting-edge communications in pursuit of regional rebalancing, science-led industrial strategies, and value-for-money procurement. Brexit, as Cummings acknowledges, brings necessary major policy change in other areas too, not least the promised immigration system.

But reading Cummings’ blog suggests a more romantic and expansive view of what an effective technocracy can achieve. He places central importance on “people,” reading as if tons of government failures would dissipate, and other projects become viable, if only more brilliant physicists, data scientists, or mathematicians, armed with cutting-edges models of uncertainty and understanding of non-linearities, were in government. Policy failure and other challenges, in other words, are downstream. “Bad management” or “the wrong skills” or “incompetent people” are held up as the root cause of bad government; better rational planners could be transformative.

My central gripe is that I doubt this is true. Government action ultimately reflects a decision to deliver collective action through the political process. And politics causes a range of structural problems that explain government failure, particularly on economics, irrespective of the brilliance of officials and project managers:

  • Political incentives and short-termism: civil servants ultimately work for politicians, and politicians can be myopic and ignorant, while wanting results conducive to re-election or pleasing interest groups. How else to explain prestige projects such as HS2 when other transport projects clearly could deliver better bang for the buck? Or moving from hugging huskies to denouncing “green crap” to meeting Greta within a decade? Or police spending levels with inflexion points at elections? If civil servants come and go, so do Ministers. There have been five transport secretaries alone since 2010. It’s all very well lamenting a lack of error correction in the civil service, but what about politicians continually demanding things with little record of success in their role as local champions (see current debates about high streets and activist government regeneration).
  • Knowledge problem: Data can help inform better policy, of course. But some significant economic problems are complex and intractable, even to the smartest brains or the newest methods. Politicians and planners seek “a solution,” often creating huge unintended consequences. Markets, by being open forums to fulfil individual wants and needs, instead find tailored solutions for different people. Economies are not predictable systems – if they were, then machine learning could make socialism a reality. Even “the best people” or “the best models” can’t forecast the macroeconomy with decent accuracy because “the British economy” is really 66.4 million people and 6 million businesses, each acting relatively freely.
  • Centralisation: Cummings might want to replicate successful private sector innovation. But market-based activity tends to start small and expand when signals like profits or consumer surveys suggest success. The public sector usually doesn’t have these signals. They could be mimicked by experimentation at local level, or hospital level, or school, with best practice spreading organically. That though, means decentralising power and accepting “post-code lotteries,” which governments are reluctant to do. Instead, project failure is met with new money and large-scale solutions. Without profit and loss, and the threat of financial failure, finding the correct “efficient scale” for much government activity is difficult, no matter what modeling or expertise you have.
  • Scope: Government engages in an extraordinarily diverse range of activities. Yes, individual-focused projects, such as the Apollo programme Cummings highlights, can be successful; but healthcare is more complex. Different policy areas often have conflicting objectives too (see the lower VAT rate on domestic fuel vs. policies to make fossil fuels less attractive). Reformers constantly run into Chesterton Fences – not least because no man can account for all of what the state does. Having a framework of what constitutes core activities and why (whether it’s delivering public goods, solving other market failures, redistributing or more) is, therefore, an important prerequisite for the type of management, resources and approach required.
  • Crowd out: government projects or the hiring of more “brilliant people” would suck individuals and resources out of the private sector, where they could benefit society more. It also disincentives individuals and businesses from finding their own solutions to problems, often creating de-facto monopolies less responsive to users/consumers and technological change.

Now, if Cummings is just laser-focused on improving delivery of core functions or projects, necessary Brexit-related change, or solving market failures, then this critique is neutered somewhat. His ideas could well generate improvements to delivery of activities government would be undertaking anyway.
But my fear, reading between the lines, is that these hires reflect an ambition for projects encompassing greater government economic activism. In that case, it’s worth revisiting why governments fail where markets succeed. There are limits to what talented weirdos and misfits can achieve.