I recently listened to Napoleon the Great, by Andrew Roberts. Roberts is a fan, not just of his military genius, but of his general moral magnificence (his fawning in the BBC series is embarrassing) so it’s necessary for him to do some defending of Napoleon’s rigged plebiscites. The rigging wasn’t terribly subtle with 99.94%, 99.76% and 99.93% “voting” in Napoleon’s favour. Roberts spends quite a lot of time explaining the various ways the rigging was organised, but ends up concluding that he had impressive popular support anyway. It was as if he was imagining subtracting a few % from the ridiculous figures and thinking that 90% is still pretty great. Napoleon obviously was very popular for a lot of his reign, but I don’t remember hearing any reason to believe those figures were anything other than completely worthless. Evidence of popularity has to come from elsewhere and even then, putting a % on it is probably hopeless.
This is what came to mind when I was reading Scott Alexander’s review of Manufacturing Consent by Noam Chomsky and Ed Herman. He chastises C&H for their obvious distortions and describes pretty accurately their way of deceiving. He mentions a bunch of prominent examples where exactly the opposite of C&H’s thesis happens. He brings up Chomsky’s Cambodian genocide denial and how he is unconvinced by Chomsky’s later pathetic squirming in his own defence. He thinks the book inexcusably biased in places.
But Scott is still convinced by the overall argument of the book that the American government is as evil as Hitler’s, Stalin’s or Belgium’s from back in the day (including the current US govt. The book was published ages ago but Chomsky’s views haven’t softened). Why? Because, “Chomsky and Herman are both academics, and they’re both relentless. When they try to prove something, by golly, it stays proved. This is a good thing, in that the book deals with very controversial topics and anything less would be unconvincing.” I think this means lots of references. Scott talks about the trade-off between successfully shifting people’s opinions and maintaining balance, but “I think C&H handle this impossible balancing act better than most.”
But when he actually checks their claims independently he finds “This is a really different story than C&H’s version. C&H never lie per se, but they leave out things as significant as a giant foreign invasion happening during the middle of the events they’re describing.” And he seems to realise that this might be true in other places.
My point is, if you know that the way they marshal evidence and argument is sometimes bogus, and it is easy to find more examples, why are these references worth anything? Why take the authors word for it that the sources really support what they say it does? It sounds an awful lot like saying, “sure Napoleon and Saddam Hussein didn’t really win 99.94% or 100% of the vote but their election results are still super impressive.” The fact that they felt the need to rig in the first place is evidence that there is something wrong with their case.
Anyway, Scott Alexander is still awesome, but it is a useful reminder that even your heroes are fallible. I just wish that he hadn’t used his own status increase the status of a liar like Chomsky.