Theme: Government & Parliament | Content Type: Blog

Cummings, Cognitive Bias, and the NHS

Jennifer Dixon


Christopher Boswell

| 20 mins read

While we decipher early signals on what the new government will bring, Dominic Cummings’ blog has become a must-read. Not because of the deliciously toxic anecdotes of his experience to date in government. Nor the amusing extended job/blog advert to attract clever ‘misfits’ and ‘assorted weirdos’ into Number 10, published in the (conveniently) media-fallow period last Christmas.

More interesting are the clues it gives as to the cognitive slant of a mind currently influential as Boris Johnson’s apparent guru-in-chief. In a potentially far-ranging portfolio, Cummings is rumoured to be focusing on boosting science and technology, and reforming the civil service. But with voters putting the NHS first in the list of big issues facing the country, could further reform of England’s largest organisation be in the sights of the new government? If so, what might Cummings’-thought signal?


The most useful leads come from an extended blog from October 2014 ‘The hollow men II: some reflections on Westminster and Whitehall dysfunction’. With many themes repeated in subsequent blogs, it is a howl against the civil service, in particular (according to Cummings) the short termism, lack of planning, lack of skills in science and maths, the Colonel Blimp quality of management, and of course the PPE public school educated over-confident bluffers colonising the civil service and political class.

Cummings notes ‘Ministers have little experience in well-managed complex organisations’, and that ‘Politics is a story of repeated administrative failure’. And, everywhere in sight, poor project management except for (this is my favourite) ‘the quiet calm capable women (aged 25–35) paid a fraction of the senior management, and without whom the entire Department for Education, and probably most of Whitehall would collapse’.

Cummings repeatedly admires the rapid progress in science and markets in contrast with the government’s poor ability to make accurate predictions and solve ‘ancient’ social problems. He notes, ‘When the institutional architectures of science and markets are working normally, they display self-correction at the edges of the network – they do not require wise chiefs at the top to decide and fix everything’.

Cummings’ exhibit one is the 60-year-old US Defense Advanced Research Projects Agency (DARPA), originally created to keep ahead of the Soviets in research and development of technology and science. The schtick is: specify broad objectives (not the process), free up the brightest and the best (in particular scientists, data scientists, mathematicians etc) to pursue high risk high gain ‘out there’ projects, ergo progress (AKA innovation) will be faster. This philosophy is clearly well worn in science – from Rutherford’s Cavendish laboratory in Cambridge in the 1930s to today’s Crick Institute – and underpins the government’s current Industrial Strategy. It is also prevalent in markets.

Put aside Cummings’ faith in the intrinsic motivation, virtue and self-regulating management of scientists (note the recent Wellcome Trust survey of research culture in science) and the hyper science-maths-weirdo elitism (phrase adapted from Giles Wilkes) evident in Cummings’ blog. His emphasis on freedom and innovation to make progress is clearly lesson 1 in the neoliberalism textbook, in which the hand of the state has minimal reach. Surely more of this will follow Brexit and the need for UK plc urgently to generate more enterprise. But while public services and the bureaucracies that shape them aren’t companies or science labs where discovery is the raison d’etre, might this prevailing wind reach the NHS with either another dose of market reform or a prising open of NHS assets (eg data) to help industry?

But then no, wait: ‘we also need institutional change to allow a reorganisation of expert attention on important problems’. Cummings argues that we need institutions to help science and markets to focus on the right things – foresee disasters and risk adapt, and foster international collaboration on the big issues. He continues, ‘Academia and markets are not aiming the most able people at our biggest problems’.

So there does need to be a guiding hand to set priorities. Science and markets if left by themselves produce unpredictable progress in a number of directions, which might suit science and markets, but not society. This is where government comes in. But here we are left with twin problems central to Cummings-rage: lack of skills and information in government; and inherent obstructing characteristics of bureaucracies.

Cummings on bureaucracy

On the lack of skills point, Cummings observes ‘a combustible mixture of ignorance and power’ at the heart of government, linked to the lack of ‘information aggregation’ to assess what is needed to set the right priorities for now and the longer term. On bureaucracies, his sharp observations on Whitehall are of short-term political fixes being vastly prized over good management. He also notes a lack of piloting, testing and transparent learning from policy successes and (in particular) blunders – in part because of a lack of incentives to do so – and, of course, rules rules rules on process and red tape. And the endemic short-termism ‘The effective planning horizon at Number 10 is ten days at best (72 hours often)’’.

The result is apparently near permanent omnishambles leading to omnidefeatism groupthink, a flight of entrepreneurs and other creative-genius types who might help discover new solutions, and poor grasp of the bigger long-term issues facing the UK. All avoidable in Cummings’ view.

His solutions are far reaching, but boil down to more rationality plus added spice from ‘misfit’ entrepreneurs. Rationality takes the form of more specialist skills (maths and science of course), better planning and management, more transparent learning, more data, more prediction based on quantitative predictive models, and more focus on the longer term. His prescription is for government and the civil service, and shows his belief in a wiser state. And as bureaucracies kill innovation, hence the need for ‘assorted weirdos’.

The plea for more rationality in bureaucracies is respectable enough, and if not new maybe now more in vogue. In its long Christmas essay, The Economist examined if the future would hold more command economies, directed by an information rich analytical control centre and a cybernetic dictatorship, with market economies (like democracy) considered to be messy, unpredictable features of the past. Nowadays the reach of government into supporting British enterprise – for example via state aid or softening domestic anti-trust laws – appears more acceptable to traditional anti-state conservatives.

The tension Cummings is describing between control and freedom, state and markets, the short and long term, planning and unpredictability, is as old as the hills. Cummings understandably wants the best of both sides. But what if anything might his musings mean for the NHS, which is far bigger, riskier, vastly more complex and difficult to change than a government department? It is also not a vast immoveable ‘blob’ (as Cummings dubbed the Department for Education): the 2012 Health and Social Care Act put its national agencies such as NHS England, at arm’s length from its sponsoring Department of Health. While these national agencies are a cluster of mini-bureaucracies, NHS Trusts are more like mini-franchises, and general practices are small private partnerships under contract to the NHS. And the specialists do not cluster atop the hierarchy, but on the shop floor as front line workers.

More innovation and competition?

So in a Johnson-Cummings era, with the NHS promised an extra £20.5bn by 2024 (real terms), is Britain’s largest and favourite organisation about to be subject to more competition? Or, in contrast, greater centralisation and grip run by a giant data-fed geek-led control room?

An obvious point first. The government‘s agenda is currently, er, rather large – how to survive and thrive after Brexit being front of mind. The likelihood of government having the appetite or bandwidth to entertain major reform of something as risky as the NHS seems slim, unless of course a series of events rattle the public and threaten to unhinge it.

On developing more market-style incentives in the NHS, evidence to date shows no great appetite among policymakers. The impact on performance of previous waves of reforms over the last 30 years is underwhelming at best, considering the political capital expended in implementing them. And collaboration across providers to coordinate complex care better is now all the rage not just in the UK but in most western nations grappling with caring for the rising tide of older people with multiple chronic conditions. There is more interest in using innovative private suppliers of data and tech services to disrupt slow-to-change patterns of care, such as online digital primary care or other smart telehealth to keep people at home living independently for longer. What is less known is how a central asset of the NHS – data, and specifically a longitudinal record for every patient cradle to grave – will be intelligently exploited, maybe with industry, for collective gain.

No, the big hope is in tech. The science-data-tech-entrepreneur-innovation industrial complex in health seems to be in full flight. There’s huge investment in the Industrial Strategy and the life sciences sector deals, the setting up of UK Research and Innovation, funds for a national AI hub, plentiful venture capital support of entrepreneurs and a Secretary of State as keen as mustard. Priorities for funding have been set but the extent to which these are skewed towards boosting science and UK enterprise, or the needs of the population and the NHS to function better is moot.

So, a lot of heat and light is currently being generated on Cummings’ ‘science and freedom’ domain. But innovations to be effective have to be implemented and spread by humans working in organisations, sometimes large and providing a ‘people service’ like the NHS. And the hand of the state needs guide the tech towards those ancient social problems, not just where the profit is.

Better management?

What is more interesting then, is how his thoughts on ‘bureaucracy’ may play out for the NHS in future. Three thoughts on that.

First, the contrast Cummings repeatedly draws between the breathless progress of science and markets and the sluggishness of bureaucracies is telling. On the face of it, fair enough you might think: after all, the stimulus for discovery in science and markets (discover or be annihilated) may be stronger, and more economic than social. And in any case finding a new drug in the lab is easier than addressing those complex and ancient social problems that public services provide for.

But this can bias us towards thinking that a new technology, drug or any other new kit transforms care faster than staff changing services incrementally and on the job. Innovation over improvement in short, and the balance of investment and policy attention shows it. Leaving aside the empirical question as to whether the sum total of a lot of little changes – say at the frontline of NHS care – may result in more progress than a number of innovative technologies, the smart (extra) money and talent at the moment is too skewed to the latter. Speeding up those incremental improvement changes requires what Cummings also values: good systematic management and transparent learning from trial and error requiring, yes, better information evaluation and data analysis. It also requires and a talent for understanding and managing human behaviour. Just the things he observes are in short supply, at least in Whitehall, as they are in parts of the NHS and all large organisations.

The irony is that without them, fancy new tech can’t be implemented either fast or well.

But how? It’s a long haul not a quick fix, nor more hyper-top-down management fed by a big-data central control room. It is about good management of people: constancy of purpose, linking top management to front-line workers with a joint mission for quality and productivity, a systematic approach to improvement understood by all. It is about driving out fear of failure, encouraging constant discovery and ongoing course correction fed by better information on tests and errors. Yes, and also another of Cummings’ points, devolved control to specialists (front line workers) to have the agency make changes. All known, just go back to W. Edwards Deming’s work in the 1950’s which helped transform Japanese manufacturing.

So in contrast to most previous NHS reforms, many focusing on administrative anatomy and tectonic structural reform, might a future wave affect its management physiology? Cummings hints at this when he writes, ‘we need decentralised coordination to tackle hard problems that existing institutions are incapable of’. He recognises that science and technology underlie complex interdependent networks (which are a feature of NHS – think clinician networks) and that progress is about ‘people, ideas, machines, in that order’ (quoting US military strategist John Boyd).

Second, Cummings rails against the focus on process in bureaucracies (the ‘horrors of HR’), an overemphasis on process over the outcomes, which kills creativity. Clearly with important outcomes at stake, a person’s health, health care is inherently risky. Process matters. The process manual for workers even at McDonald’s is large, and health care is more complex than flipping burgers. Given his interest in science and technology this might point to a new focus on cutting regulation, and in particular of new innovations. Already the prevailing wind in the entrepreneur-industrial complex is for organisations like the NHS to fast track trialling of innovations such as new digital apps and promising cancer drugs in situ, in the wild as it were, rather than wait for more traditional (longer) evaluation before introduction. This may suit entrepreneurs and allow patients faster access to innovations. But while large live datasets in the NHS will allow the real-world testing to happen fast, this is some way off. Investment and attention here perhaps.

Third, Cummings’ points on big bureaucracies having a failure of information aggregation, combined with a lack of data scientists, analysts and modellers to predict the future to help set long term priorities. In short, he notes bureaucracies cannot deal with complexity well and that politicians aren’t CEOs and haven’t generally managed large complex organisations. A recipe then at least for keeping a priority-setting Department of Health separate from professional management of the NHS, and for NHS Trusts and GPs not to be throttled by too many central targets from NHS management HQ.

Cummings signals an old lesson ‘grasp what it is reasonable to attempt to manage…while devolving other things and adapting fast to inevitable errors’. For businesses, this might work, but try telling that to PMs or ministers that want a risk-free NHS under their watch. And even for top management in the NHS it is easier said than done. The paradox is that more transparent and faster error recognition load risk onto ministers and erode the political cover needed to devolve management. Experience shows in the NHS, any such devolution (even if helped by political devolution such as Greater Manchester) will be short lived if pressures mount. Central ‘command’ and grip will then surely be the atavistic response.

On predicting and planning for the longer term (not 72 hours but say 5 years plus), it’s clear from many reports that government could do a lot better if it thought this a priority. ‘What’s the counterfactual?’ (of not doing this) remarked one senior Whitehall (Department of Health) insider to me recently, unconvinced of the benefits of more effort. I was too polite to suggest: Brexit, flooding, failures of major outsourced government contracts, or a failure to plan the NHS workforce resulting in widespread staff shortages. As Cummings ruefully notes, the information and talent are there, but where are the incentives? Here the NHS does have a good story to tell, having put together a comprehensive Long Term Plan based on evidence, convincing enough to the Treasury to lever that £20.5bn extra funding promised by 2024. But more future thinking and planning would surely help.


Back to reality. It is difficult to see if the wind from any of Cummings’ breezy blogs will blow beyond trying to reform Whitehall and reach the NHS. In any case, reform is unlikely to be framed in Cummings’ rather abstract terms, markets versus bureaucracies, geeks versus bluffers, but much more practically to resonate with voters. And Cummings also may not last as the PM’s advisor – he may be a ‘mutant virus’ soon to be expelled from the government’s immune system (to adapt a phrase from his DfE colleague), or expel himself in frustration like other entrepreneurs.

After the Brexit referendum, some simply won’t engage with Cummings on principle. That would be a great pity. Like them or not, his ideas are modern takes on old tensions, and well worth pondering over.