Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • NEWS

Government call for science ‘weirdos’ prompts caution from researchers

Dominic Cummings.

Dominic Cummings, a political strategist with a keen interest in science, is one of UK Prime Minister Boris Johnson’s chief advisers.Credit: Daniel Leal-Olivas/AFP/Getty

Researchers have reacted with surprise and caution to a bizarrely worded job advertisement posted by a senior UK government adviser. The notice calls for scientists, mathematicians and “super-talented weirdos” to work for the prime minister and to help make “rapid progress with long-term problems”, and cites several scientific papers.

Researchers, including the authors of some of the papers, welcome a focus on data-driven techniques and scientific skills from the top level of government, but caution against oversimplifying how science is applied to policymaking.

Dominic Cummings, a political strategist who is chief special adviser to Prime Minister Boris Johnson, posted the advertisement on his personal website on 2 January. Cummings is known for his strong and sometimes controversial views on science, as well as a ‘move fast and break things’ approach that aims to disrupt the status quo. Johnson’s Conservative Party held on to power — and increased its majority in Parliament — in a general election last month.

In the 2,900-word advertisement, Cummings says that he is seeking “data scientists, project managers, policy experts, assorted weirdos”, as well as “unusual” mathematicians, physicists and economists, to work in the prime minister’s Downing Street office, which will oversee major policy changes as part of the United Kingdom’s departure from the European Union. The post attracted attention for its brash language and odd requirements, such as that applicants show excitement about a slew of specific scientific ideas.

Cummings describes his plan to take a data-driven, computational approach to tackling the science of prediction and decision-making in government. The ad calls for candidates to familiarize themselves with research papers1,2,3,4 in fields such as data science and artificial intelligence (see ‘Reading list’) — although it does not mention the overarching policy goals or societal problems that the government wants to solve.

“What Cummings wants seems to be a classic Hollywood scientist, coming in wearing a white coat and telling him what the answer is and saving the day, and it doesn’t work like that,” says Jack Stilgoe, a researcher in science and technology studies at University College London. “The idea you can come up with technological fixes to social problems — we have a century of experience showing how flawed that theory is.”

Reading list

Four of the papers that applicants to the prime minister’s office should consider.

‘Early warning signals for critical transitions in a thermoacoustic system’ This engineering paper looks at the early-warning signs that show that a model engine is tipping into a destructive state. The authors suggest that their observations can be applied to spot similar tipping points in other domains, such as finance.

‘Computational rationality: A converging paradigm for intelligence in brains, minds, and machines’ This review paper explores the concept of computational rationality — the idea that intelligent systems should optimize their performance while taking into account constraints on their knowledge, memory, speed and energy. It notes that parallels exist in such optimization across psychology, neuroscience and artificial intelligence.

‘Scale-free networks are rare’ This paper applies statistical methods to 1,000 networks used to represent and study complex systems in areas from biology to transport. The work debunks the popular scientific idea that most networks are ‘scale-free’, a mathematical description of their structure that depicts them as having a small number of highly connected hubs.

‘Statistical and machine learning forecasting methods: Concerns and ways forward’ This paper investigates whether certain machine-learning methods are more accurate than conventional, statistical ones in forecasting data that occur infrequently, such as a country’s gross domestic product or company sales figures. The authors find that although machine-learning methods can be very accurate for some data, in this case statistical approaches beat pure machine learning.

Complex problems

Many authors of the papers, contacted by Nature’s news team, expressed surprise — and in some cases delight — at a senior government figure paying attention to the scientific literature. “It gives us immense pleasure to observe that our results can pave the way for the young talent to solve issues of highest importance,” says R. I. Sujith, an aerospace engineer at the Indian Institute of Technology Madras in Chennai, whose paper1 on catastrophic changes in physical systems was among those cited.

Policymakers should indeed exploit the huge scale and granularity of data now available, adds Aaron Clauset, a computer scientist at the University of Colorado Boulder, who authored two of the cited papers3,5. But he warns against blindly using these tools on complex social and economic problems, the solutions to which often involve understanding humans rather than technical issues. “In many cases, we don’t understand causality well enough to formulate a policy that will not do more damage than good,” he says. For a lot of questions, science is just one of many factors that should determine policy, he adds.

This opinion was echoed by Sujith and four other authors contacted by Nature. Sam Gershman, a cognitive neuroscientist at Harvard University in Cambridge, Massachusetts, whose work Cummings also cited, says that some elements of his research could be relevant to policy projects. “However, we must be extremely cautious in how we apply these ideas, because there are many historical examples of social engineering gone awry due to overly simplistic assumptions about human cognition,” he says.

Advice networks

Researchers familiar with how science advice is already used in the UK government said that new and unusual talent is welcome — but that Cummings’s post is unfairly critical of the government’s civil service. He should focus on bringing new scientists in “rather than dismissing the whole service as it is at the moment”, says geneticist Paul Nurse, who leads the Francis Crick Institute in London.

Natalie Perera, executive director of the Education Policy Institute in London and a former senior civil servant, added on Twitter that Cummings’s call — which explicitly referenced applicants having attended a “great university” and forgoing a work–life balance, and shrugged off the need for what it called “gender identity diversity blah blah” — would prevent the advert from attracting the “genuine cognitive diversity” it purports to target.

Gershman adds that Cummings’s approach to research encapsulates a dilemma facing scientists in current political environments. “There is a danger that science could be used to serve programmes that many scientists might not personally wish to support, given the divergence in political attitudes between most scientists and the current government.”

Cummings’s desire to model and predict real-world events might face an additional hurdle, writes Jeni Tennison, who leads a London-based non-profit company called the the Open Data Institute, on her blog. “I think Cummings’ team will pretty rapidly come up against a lack of usable, accessible data for the things they want to look at,” she says.



  1. Gopalakrishnan, E. A., Sharma, Y., John, T., Dutta, P. S. & Sujith, R. I. Sci. Rep. 6, 35310 (2016).

    Article  Google Scholar 

  2. Gershman, S. J., Horvitz, E. J. & Tenenbaum, J. B. Science 349, 273–278 (2015).

    Article  Google Scholar 

  3. Broido, A. D. & Clauset, A. Nature Commun. 10, 1017 (2019).

    Article  Google Scholar 

  4. Makridakis, S., Spiliotis, E. & Assimakopoulos, V. PLoS ONE 13, e0194889 (2018).

    Article  Google Scholar 

  5. Clauset, A. Preprint on arXiv (2019).

Download references


Nature Careers


Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing


Quick links