Seems like American English has been slowly collapsing the meanings of 'political' and 'partisan' for a while. Which is certainly annoying for choosing words to mute.
I suspect that my Twitter feed was better curated. Since it was quiet here, I followed some interesting but a bit sketchy people. But of course, there’s also the fact that important people in science have been going bonkers in an attempt to curry favor with the new administration.
Yep, because the claim that science has 'value' or is 'political' — which is correct — is often used to discredit science as a biased or subjective opinion: THIS is also a huge social problem. Therefore, the claim that science has 'value' or is 'political' cannot be defended uncritically.
My naive impression is that the way science is criticized academically is somewhat different from criticisms in popular culture. This gap might have become critical in recent years.
I should read the paper. While I share the sentiment that truth value (or the equivalent) should be judged independently from external factors as much as possible, I’m not sure it is always so modular. This is partly why with @zvihb.bsky.social we argued for being vigilant to external manipulation.
I think it is wrong to think that anything here turns on NHST vs Bayesianism, as Dan Steel has shown in a couple of papers. Ampliative inference -> inductive risk. Also FYI, @ewinsberg.bsky.social has argued convincingly that the standard interpretation of Jeffrey is mistaken.
The bad idea here isn’t significance testing per se (though that’s indeed a terrible idea—see below) but the idea that ampliative inference hinges on some arbitrary threshold (be that Lockean or significance level). An even worse idea. https://www.davidpapineau.co.uk/uploads/1/8/5/5/18551740/bayestlspapineau.pdf
Despite the common framing, I don't think arbitrary thresholds are essential either. What matters is that "Believe truth" & "shun error" can pull us in different directions, plus the need to settle issues at some point in order to act. Bayesians can't avoid either (& still be discussing science).
Correct. Jeffrey suggests the possibility that moving to probabilities could save the VFI, but says there are serious difficulties with this. My own view is that credences are just things we have at the sub personal level. Probabilities we announce are decisions, and decisions involve values
There are two different problems. Of course, science as a human and social activity is full of 'value,' but this does not mean that science is merely a biased or subjective opinion. Unfortunately, the claim that science has 'value' is often used to discredit science, which is a huge social problem.
Comments
BlueSky will probably have a higher proportion of university educated people.
What do we even mean by not political? Choosing which experiment to do by drawing lots according to some of collective Bayesian probability?