20.11.20 Big Tech

How Dominic Cummings wasted the greatest opportunity of his life

Boris Johnson’s chief adviser had a vision of government by data. But when the pandemic gave him a chance to remake the British state, he flunked it

When Dominic Cummings took the stage in London at the Institute for Public Policy Research, a thinktank, in November 2014, few outside Westminster’s political bubble had heard of him. Dressed in a plain black suit with a skinny black tie, he looked like he’d walked off the set of Reservoir Dogs. He was introduced as a man with “interesting and controversial views”. Though he would later gain a reputation as a snarling hatchet-man, he laid out in measured tones his views about “fundamental aspects of how the world works”. This involved a helter-skelter tour of disciplines such as astronomy, entomology and neurology (Stars! Ants! Neurons!). Quite what this amounted to was hard to say, but he spoke with the unsmiling confidence of a man who thinks that everyone appreciates how right he is.

The nub of his lecture was a screed against the institutions that comprise the British state. Nothing about them, Cummings said, was good. There are no quantitative skills in Westminster, no management skills, no ambition. Incentives are misaligned, goals are unclear, failure is normal. Decision making, he said, was “almost random” and “largely rubbish”.

Cummings never mentioned the demands of ordinary people, whom good government is supposed to serve. To him, the state was a machine, and its smooth-running was inherently desirable. Until recently he had served as a special adviser in the Department for Education, yet he seemed uninterested in the subject except for the extent to which it might equip graduates with the skills necessary to run the country. The impression Cummings gave was that the British state was a disaster. His solution was to tear it down and start again.

The British public would soon get to know Cummings, first as the architect of Brexit and the “Vote Leave” campaign, then, after Boris Johnson entered Downing Street in July 2019, as the prime minister’s chief adviser, or “brain”. During the pandemic he became notorious as the man who broke lockdown rules to drive to Barnard Castle, apparently in order to test his eyesight. This escapade led one British tabloid to call him “a cheat” and another to label him “the weirdo blamed for destroying lockdown”.

Liberal opinion towards Cummings has often veered between awe at his supposed power and contempt at his views. The media presented him as unstable and Machiavellian, a political wizard in a beanie, stumbling between genius and incoherence. Yet the stridency with which he seems happy to dismiss other people as “thick as mince” or “charlatans” is at odds with the vagueness of his own thinking: it’s hard to work out exactly what Dominic Cummings believes. Though his writings are legion, he is incapable of expressing himself systematically or concisely. He tears others to shreds, but is less good at expressing a positive view about almost anything.

One thing that Cummings has consistently been in favour of is data – more specifically, the use of data science to improve the performance of government. He has written of the benefits of “Seeing Rooms”: control centres that analyse streams of data to support government decision-makers. He has repeatedly pointed out that the machinery of the British state, which has changed surprisingly little since the Victorian era, was hopelessly inadequate for dealing with the complexities of the networked world. Pandemics are one of the challenges that a state would have to face, a danger that Cummings had pondered on his blog (he was castigated for later inserting a reference to “Sars coronavirus” into a post he had written in March 2019) .

There was nothing inherently ideological about Cummings’ ambition to remake the state. A left-wing version might envisage a government that used data to deliver services to citizens at a lower cost than the private sector, in the mould of Project Kangaroo, the BBC’s abortive attempt to build a state-owned Netflix. A right-wing variant might use data to shrink the state to its barest extent, while simultaneously knowing more about its subjects’ lives than ever before. Through this step-change of knowledge the state would be able to navigate its place in the world and its relationship with its subjects skillfully but subtly.

Cummings first came to the attention of the British public as the director of Vote Leave Matt Cardy/Getty Images

Cummings’ view seemed to have been vindicated by his successful campaign to take Britain out of the EU. David Cameron’s recklessness in calling the referendum exemplified how even the most powerful man in the country failed to understand the minds of Britons. (The two were not friends: Cummings described Cameron as “spectacularly” incurious, among his more polite expressions; Cameron labelled Cummings a “career psychopath” who dripped poison.)

This transformation would not be easy. Cummings knew that hundreds of thousands of civil servants stood in his way, a group he famously described as “hollow men”. They operated the machinery of government and, in his eyes, had a vested interest in perpetuating the status quo. “Although [Whitehall] is systemically incompetent viz policy and implementation, its real focus is on its own power, jobs, and money,” he wrote on his blog.

In his digital utopia, these pettifogging mandarins would be swept aside, replaced by computer networks that spanned government departments and fed data into algorithms. Cummings included himself in the ranks of the dispensable. He repeatedly claimed that he had joined the government merely to set things in motion and that he would leave within a year. Like the apocryphal actuary who writes computer programmes to automate his work, then leans back and collects his paycheck, Cummings reckoned that historic changes to the machinery of the state could be achieved with the seeding of a few smart ideas. After that, the data man could ride off into the sunset with his rucksack, his work done.

In an investigation in partnership with 1843, the Bureau of Investigative Journalism spoke to 17 IT contractors, former government employees and members of Cummings’ team at the heart of power. According to them, Cummings planned to create a new unit in No 10. This team would devise policies based on data and manage the information flow itself – the latter task would be accomplished by data scientists who would analyse data sets from across the spheres of government. The idea was to create a virtuous circle by which real-time monitoring of departments would inform policy and enable the prime minister and his advisers to identify underperforming areas. Cummings is said to have wanted to entirely eliminate permanent secretaries – the apolitical heads of civil-service departments – and shrink the cabinet from 22 to a mere six or seven secretaries of state.

If successful, this would radically transform both the power of the executive, and the dynamics of that authority. Though the prime minister is the nominal head of the British government, the Treasury has long been a significant influence on all areas of policy because it determines the extent and purpose of funds that each department receives. Data would allow the prime minister to take back control. Tellingly, neither Cummings nor his team ever spelled out what kind of policies his data-led government might pursue.

To build a fairer, more democratic society, we all need to know what isn’t working and who is responsible. Our investigations expose failings and wrongdoing so those with power can be held to account. We have more stories like this and you can help us investigate

Yes! I will support the Bureau

The Covid-19 pandemic has killed more than 50,000 people in Britain, left untold numbers debilitated and disrupted everyone’s lives. But for Cummings it presented a golden opportunity to impose his vision of digital governance. In a public health emergency you need to centralise control and annex large datasets – actions that would meet with significant resistance if they weren’t happening in the name of fighting a pandemic. Among officials and technology contractors there was widespread talk of not letting this opportunity for digital transformation go to waste, according to a number of people involved in the government’s coronavirus response. This was Cummings’ chance to prove the effectiveness of a vision he had long extolled: to have “seeing rooms” at the heart of the panopticon, watching over and controlling the activities of the furthest reaches of government, augmented by digital technology.

Cummings had focused on digital matters since becoming the prime minister’s chief adviser in July 2019. Early on he ordered analytics software to be installed across government websites to give the Government Digital Service (GDS) – part of the Cabinet Office which works across departments – a unified view of Britons’ interactions with its online services. Even though many commercial websites gather users’ personal data in just such a way – and GDS itself asked for the software – this largely innocuous move caused uproar among privacy campaigners, mainly because of Cummings’ involvement. “Dominic Cummings may claim to be a pioneering advocate of data science, but when it comes to its ethical use, he is woefully behind,” Pascal Crowe of the Open Rights Group, a campaign body, said at the time.

An early order from Cummings to gather analytics on government websites raised hackles among privacy campaigners Leon Neal/Getty Images

Cummings is the kind of contrarian who delights in criticism as confirmation of his own genius: “great ideas”, he later wrote on his blog, will “almost inevitably seem bad to most” when they are first conceived; “media noise” was not a problem. He began hiring people to work on digitisation.

One new recruit was Ben Warner, a data scientist who had worked with Cummings on the “Leave” campaign. Warner was made a special adviser in December 2019, charged with reorganising how Britain uses the statistics it collects. The British government prides itself on collecting an extraordinary range of data (number of peg makers in 1841, anyone?), but despite the formation of the Office for National Statistics (ONS) in 1996, many streams of information siphoned off to different departments are never linked together.

Warner wanted to create an integrated data infrastructure that pulled together data on households covering everything from education, income and benefits to migration status, crime records, health and so on. New Zealand had built just such a database, a feat that proved its worth during the pandemic when public health officials were able to combine tax, transport and household data to predict the spread of Covid-19. That came later.

In the first instance, though, this scheme was to be the first step in Cummings’ plan to increase central oversight of government departments. According to documents obtained under the Freedom of Information Act, he thought improved data flows would also affect specific policies such as the “levelling-up agenda”, the government’s plan to provide investment to poor communities largely in the north of England.

It may seem strange that a man characterised as Britain’s digital Machiavelli and his cabal of quants took such interest in the dull grind of statistics distribution. But as Cummings saw it, such digital plumbing was necessary to allow more exciting projects to flourish. Unless the signalling is correct, it’s impossible for a train network to ferry goods and passengers around.

The revolution might have become institutionalised. Instead, as February turned to March, a new coronavirus started infiltrating populations across the world, with devastating effect. Everything changed. And the pandemic became a test of state capacity. Citizens had to trust not just that governments had their best interests at heart but were capable of protecting them from a deadly virus.

The provision of healthcare was central to each government’s response: were there enough beds, medical staff, ventilators, protective equipment? But to vanquish a pandemic you have to treat it as an information problem too. You need to find the humans who are hosting it and prevent its transmission to others. Speed is essential, both in learning about how the disease moves through a population and in communicating this information through a system designed to isolate positive cases. This was Cummings’ moment. And he fluffed it.

The pandemic offered an unrivalled opportunity for government to consolidate and analyse data to lead policy Tolga Akmen/Getty Images

On March 11 the full horror of the Covid-19 pandemic set in across the British government. The Bank of England cut interest rates to their lowest levels. The following day Boris Johnson announced that Britain had failed to contain the spread of the virus and that the best it could do was delay its spread. Cases of Covid-19 were growing exponentially.

Cummings held a meeting at No 10 with representatives from 22 technology companies to discuss how they could support the government’s response to the coronavirus. Representatives from the world’s five most valuable companies – Apple, Google, Microsoft, Facebook and Amazon – were present. Their powers of data collection and processing were exactly what the country needed.

To fight the pandemic the government had to establish a system that would identify people who had tested positive for the virus, contact them and then contact those who they might have infected. At that time Britain’s network of government laboratories could handle only 1,500 or so tests a day, and it took the best part of a month for the government to come up with a plan to set up a number of new laboratories. These were called Lighthouse Labs, after the fluorescent dye they used in the tests.

The construction of these new labs to process tests was only part of the solution. They had to inform infected individuals – as quickly as possible – to prevent further transmission. Armies of call-centre staff, summoned by outsourcing firms such as Serco and Sitel, proved inadequate at getting in touch with enough people who had tested positive (over the summer they failed to reach 20% of cases) or their recent contacts (40% of them were not contacted).

The government was not entirely responsible for the lack of testing capacity. Over the previous 20 years both Labour and Conservative administrations had shrunk the government lab network, taking the view that excess state capacity was wasteful. Yet the likelihood of a respiratory pandemic was well known. A three-day government-run exercise in 2016 called Cygnus had focused on the country’s preparedness for an influenza pandemic. The final report stated that a pandemic would require the capability to “surge resources” into key areas – a capability, the report noted, which was “currently lacking”. During the exercise it also found that participants were confused about which sets of epidemiological data were the most significant.

Many of the failings exposed in the Cygnus exercise were not acted on. To some extent this was a vindication of Cummings’ view that the civil service was a place where reports get written but nothing gets done. So, when the pandemic hit, the government lacked both the necessary capacity to deal with it and, more profoundly, it lacked the knowhow to build that capacity quickly. According to multiple sources involved in the government’s test-and-trace efforts, management consultants were hired not just to conduct outsourced work but devise a plan of action.

They were only too eager to help. “Consultancies were running amok,” says the boss of one technology company, who attended Cummings’ meeting on March 12. “A lot of them are washing up at the pandemic. Where else were they going to be working?” For all Cummings’ aspiration to run a government on the principles of data science, it is noticeable that as the crisis hit, he reverted to the playbook of the Blair and Cameron years: when in trouble, call in the consultants.

The ineptitude of this scrambled response became evident in October, when NHS Test and Trace, which ran the national scheme, temporarily lost more than 10,000 cases as a result of using Microsoft Excel spreadsheets. No 10 was explicitly warned about its reliance on Excel months before the actual foul up. Yet Cummings seemingly either considered it unimportant or was incapable of solving a relatively simple database issue. Whatever the reason, this failure was a poor advert for Cummings’ ambitions to re-engineer the state.

An early glitch by Test and Trace proved a poor advert for Cummings' data-led vision of government Daniel Leal-Olivas/Getty Images

Cummings recognised the lack of technical skills in government as a problem and had set out to fix it. But he was as much part of the problem as the solution: many people simply didn’t want to work for him. The only tangible result of his provocative job advert for “weirdos and misfits” to come and work at No 10 was the hiring and then rapid firing of two men whose views were weird enough to put them beyond the pale of acceptability.

The post of chief digital officer of the government, which had first been advertised in September 2019, remained empty (it still is). Several technical leaders and entrepreneurs told 1843 that Johnson’s government had asked them to work on digital projects, including test-and-trace, but that they had declined (a number of them had no interest in helping the government deliver Brexit.) As a result Britain stumbled through the summer, struggling to keep up as infections rose in a second wave.

In early September, Cummings established a new “control centre” in Whitehall to improve the government’s response to the pandemic, which he himself described as a “shitshow”. But he never made clear how it would help. It turned out that the state still needed talented people working for it – faith in data alone would not save it.

The NHS, not Cummings, was responsible for the government’s few successes. Treatment for Covid-19 improved steadily. In late March 58% of people admitted to intensive care in England with the disease survived more than 30 days; by the end of June that had increased to 80%. A technology incubator called NHSX, founded in February 2019 before Cummings joined the government, had attracted a coterie of talented techies dedicated to improving the technological capacity of the NHS. They devised a system to distribute scarce ventilators to places in greatest need. Bureaucratic politics meant that NHSX did not work on the test-and-trace system. And the second most powerful person in British politics couldn’t break down that barrier.

Cummings also encountered opposition from branches of government he had hoped to dominate. Johnson wanted a dashboard of NHS performance data piped directly into Downing Street. Sir Simon Stevens, head of the NHS, whom one senior government official described as “the most astute politician of the lot”, is said to have resisted, asking what use Johnson would have for such detailed medical data. No 10 ended up with a simplified version of the dashboard, though it’s unclear to what use they have put this data (a government spokesperson did not comment on the dispute).

Britain’s shambolic response to the pandemic cannot be laid exclusively at Cummings’ door. The machinery of government is too complex for that. But it is undeniable that Cummings failed to impose his vision of the data-driven state when handed more power than most can dream of.

“I think Cummings is a bit of a fraud,” says a former government adviser. “I’m kind of annoyed because I drank the Kool Aid early. He’s a charlatan who talks a big game over dinner, but you’re left with the sense that he doesn’t actually have it.” The man who walked out of Downing Street with his cardboard box of belongings on November 13 was a techno-enthusiast whose grand plans had come to naught.

Dominic Cummings leaving No 10 for the last time Hollie Adams/Bloomberg via Getty

Perhaps Cummings never stood a chance of constructing a digital Leviathan. Half the country saw him as a villain who’d tricked Britain into voting for Brexit. Some were horrified by the thought that this ideologue would replace civil servants with data-crunching algorithms to consolidate power in the executive. “This government is churning out increasingly dystopian plans at a huge cost to the public purse and civil liberties,” said Silkie Carlo of Big Brother Watch, a civil liberties campaign group. What some people would present as tedious proposals to digitise and network anachronistic systems, others saw as a stake to the heart of British democracy.

The British civil service occupies a special place in the public imagination. TV comedies skewer its unique combination of bumbling ineptitude and icy careerism. It may be parochial and cumbersome at times, but it also serves as a check against the worst impulses of government. That civil servants do not change when governments do is both the institution’s greatest strength, and an enduring weakness. And it was their careers that Cummings’ power grab put at risk – a white-collar version of automation that had already rendered many blue-collar jobs obsolete. It’s hard to imagine that they would not have put up a fight, as Stevens did over the NHS dashboard.

The resistance would have grown stronger as it reached deeper into the power structures of Whitehall. Under the current system the Treasury holds the reins of state: it analyses the spending of each department and directs its activities by changing its budget. Flows of money serve as proxies for flows of information: government accounts provide a crude version of the data-gathering and decision-making algorithms of which Cummings dreamed.

It is undoubtedly true that data has advantages over money. It can be analysed in real-time, not just after periodic reviews. It offers a more granular view of performance, which could theoretically help the state serve its citizens more effectively. The British government used to have a public dashboard that tracked a wide range of government services, but this was shut down in 2017 as the most recent band of digitisers lost power to reactionary elements in the civil service. Cummings faced a bitter struggle to hack through the networks of patronage that coiled within the budget-oriented system.

Got a Story?

We welcome tip-offs from the public and we always protect our sources

Find out how to work with us

Cummings left Downing Street a failure. The digital revolution, however, still looms. Whether it likes it or not, the British government is already in a digital competition, both with other states and with the large technology companies that increasingly resemble them. At some point it will have to confront this.

The tech companies have been accruing power for decades and already know more about a given country’s population than that country’s rulers do. Unlike tech giants, democratic governments across the world are accountable to their citizens. If the gap between the capabilities of these entities continues to grow, and states become more detached from their citizens, democracy itself may be in crisis.

One policy director at a company that researches artificial intelligence points out that tech firms were able to use their superior data-gathering and processing capacities to see the dangers of the new coronavirus long before governments did. Apple and Google concluded as early as February that a pandemic was coming, based on trends in the aggregated data flowing out of their networks. Most Western governments were left in the dust. “Government doesn’t have the machinery to glom onto anything effectively,” says the policy director.

Given the asymmetry of power, it is little wonder that governments have proved to be ineffective at regulating tech companies – a feat that has nonetheless become more necessary as these giants have become more state-like. “Cloud platforms in the US and China are absorbing many of the formal and de facto functions of states: currency, cartography, identity,” said Benjamin Bratton, a social theorist at the University of California at San Diego. Governments will have to bulk up their digital capacities in order to monitor these companies effectively and keep up with the pace of the change in the digital world.

There is, of course, the Chinese option: brute force and total control. The Chinese government subjects any business, even ostensibly private technology giants, to state control: only recently it prevented Ant Financial, a fintech company, from floating its shares on public stock exchanges. During the pandemic Alipay and WeChat Pay, China’s two largest payment apps, hosted the government’s public health software. In Britain, by contrast, the government had to implore citizens to download its app.

The digitisation of the state cuts to the heart of statecraft. Bratton notes that sensing, modelling, forecasting, then taking action and repeating the cycle is an effective definition of governance. For centuries, this loop has run through the familiar structures of government. Now we live in a world where private companies have far surpassed the state in their ability to sense, model and forecast.

Western democracies face a choice. They can build up their own digital capacity, enabling them to regulate tech companies within a free market. Or, like China, they can force these companies to serve the public good. If they do neither they will become less powerful and less relevant, both geopolitically, and in the lives of their citizens.

It is unclear how much Dominic Cummings pondered the future of democracy. He did not respond to multiple requests for an interview. But he was one of the few people in Britain who had the necessary vision to reshape the state for the 21st century, proximity to power and unflinching confidence in his own abilities.

Despite his intellect, Cummings was precisely the wrong person for the job. He was enamoured with data but scattergun and haphazard in his own management style. His abrasiveness deterred talented experts from joining the government. His contempt for the existing system raised the hackles of civil servants who might otherwise have been persuaded to change. More importantly, he never articulated a vision of how a data-driven government would change the lives of British citizens for the better. Cummings may be, once again, a prophet in the wilderness. The urgency of his fixations remains.

Header image: Dominic Cummings after leaving Downing Street on November 13. Credit: David Cliff/NurPhoto via Getty

Our reporting on Decision Machines is funded by Open Society Foundations. None of our funders have any influence over the Bureau’s editorial decisions or output.