As someone with a physics PhD from one of those ‘great universities’ he’s banging on about, may I say that this ‘recruitment ad’ from Dominic Cummings is dangerous a cargo-cult-data-science load of cock. dominiccummings.com/2020/01/02/two…
His central thesis, which seems to be that some kind of machine learning can be deployed on geopolitics, would be a fascinating decade+ research programme, similar to Tetlock’s superforecasting work he cites. It is absolutely not ready to be deployed in the heart of government.
Hiring some data scientists to apply findings from ‘Early warning signals for critical transitions in a thermoacoustic system’ to predict the economy should absolutely not be being used in policymaking any time soon. 
TELL ME MORE ABOUT CRITICAL DYNAMICS IN A THERMOACOUSTIC SYSTEM, DR PHYSICIST
This is science as hero-worship.
This is science as hero-worship.
The use of the word ‘finance’ in the speculation in the introduction to a Nature paper (a journal where you need to show why your work would be of interest to a broad readership of different scientists) doesn’t mean there will be a breakthrough any time soon.
He says that there are ‘profound problems’ in the civil service, caused by ‘a confluence of: a) Brexit…’ Er, so you literally created the main headwind you cite, and are now using it as an excuse to tear up our institutions in the name of the ‘Conservative’ party?
For all this hero-worship of experts and superforecasters, he doesn’t seem to have bothered to consult any—even regular forecasters could tell him that Brexit is a terrible idea, and its dizzying scale together with civil service cuts is why it’s causing policy problems.
Another problem is that many of these super-talented weirdos will be anti-Brexit, if the consistent results of surveys of academics are anything to go by. You just ruled out 80–90% of your target workforce, Dom…
Finally, ‘I don’t want confident public school bluffers. I want people who are much brighter than me who can work in an extreme environment.’ As literally everyone has pointed out…isn’t that just a description of his boss?
It is absolutely the case that government should be more evidence- and data-driven in policymaking, and it seems prima facie likely that more scientists in the civil service could help with that.
This is not how you do it. There is no evidence or data behind it. And, if moving fast does indeed break things, I’m worried it will tarnish the reputation of evidence-based policymaking and the role of scientists in government for longer than Dom will be around to blog about it.
