It does, but it probably doesn't need them to make bold, contrarian, under-evidenced, extremely high risk public health policy interventions on the basis of untested modelling methods, in the middle of a pandemic, in the Guardian. https://twitter.com/globalhlthtwit/status/1267443033303367681

It's not been a 'good war' for science in the UK public sphere.

To spell this out: let's say the value of Friston's intervention is in a medium-term transformation of epidemiological modelling approaches, at the cost of some short term forecasting errors. Why would this be a *justification* for high profile and high risk policy interventions?

Or to put it differently: *science* benefits from off-the-wall iconoclastic alternative approaches, even if they're badly flawed, because *science* can test, evaluate, modify and build upon those approaches over the medium to long term. *Policy-makers* need to be *right*.

Or - if "right" is too hard - policy-makers need to have a clear sense of the likely space of risks and probabilities, as best the current science can give it. Iconoclastic and untested novel approaches are themselves a risk, and should be used very cautiously to *assess* risk.

Moreover, for the reasons outlined here, Friston is simply unreliable. That may not be a problem for iconoclastic speculative science (which can be moderated and modified by other researchers) but it is very bad news w/r/t policy advice. https://twitter.com/AdamJKucharski/status/1267143250797830149