by Ben Goldacre
Rating: ★★★★
A collection of Goldacre's Bad Science columns, plus a few more odds and ends he's written over the years on a cluster of related topics. Almost all of it is over a decade old, so the issues aren't usually pressing ones any more, and indeed it's a little nostalgic reading about things like the Daily Mail's campaign categorising everything in the world as either causing or curing cancer, or the MMR vaccine fiasco, or magical bomb detection scams. In some ways things have improved a lot -- I don't read the papers often, but my impression is that journalists at least aren't allowed to get away with blatant distortions of scientific publications in the same way any more (or at least, not as much), and a little scientific literacy seems to be developing, as journalists respond sluggishly to the pressures applied by Goldacre and his affiliated army of nerd bloggers. In other ways there are serious backslides at the moment -- Goldacre was able at the time to scoff disbelievingly at early publications raging about the 'fascism' of science and promoting 'other ways of knowing', but that kind of talk has exploded recently, and is being uncritically rebroadcast (along with other postmodernist word salad) from prestigious medical journals that absolutely should know better.
This collection isn't as big or hard-hitting as his previous works. Bad Science and Bad Pharma have a focus that is missing here, with just occasional runs of articles on a topic, as more sleuthing is done and cranks or shills are exposed. The one item that most interested me was Goldacre's undergraduate(!) essay on heroin vs methadone prescription for heroin addicts. If the results he presents are reliable (and I have good reason to expect he would've corrected it if not), the choice to start prescribing methadone instead of heroin is one of the most bizarrely harmful public health decisions doctors have ever made, founded on apparently nothing other than a gut suspicion of heroin maintenance coupled with price disparities caused by a legal monopoly. Methadone is, on this now old review of evidence, significantly more dangerous to the patient, patients are (understandably) less likely to continue methadone treatment, and methadone is much harder to wean addicts off than heroin. I've never paid much attention to drugs policy but unless Goldacre is missing something significant out this is obviously crazy.
Goldacre's gone comparatively quiet recently, after his push for clinical trial preregistrations (possibly one of the most important things anyone has ever worked on) he has shifted into a general health data focus, and has been feeding into government policy and building tools for safe health data sharing with his own group. Predictably, he's been busy with the pandemic effort recently. With some luck, he'll start writing more columns in his old style again soon. (In the meantime, I'd suggest Unherd's Tom Chivers -- he has a similar statistical literacy focus and bullshit-cutting style, plus he's a Warhammer nerd.)