Thread by @Meaningness

“A science of science” seems like it would be enormously valuable, but so far attempts have been disappointing. Michael’s analysis leads me to think we’re skipping a step. We need “the natural history of science” first.

https://twitter.com/michael_nielsen/status/1481732031046307840

Scientists often denigrate natural history as unscientific and uninteresting and amateurish and antiquated, but each specific science couldn’t have gotten off the ground without the extensive and careful natural history work on the same topic that preceded it.

We actually have nearly no idea what scientists do. We know what they say they do, but those of us who have done it and watched many other people do it, who have also wondered “what are we doing,” realize the concrete work is dissimilar to theory and mostly unnoticed.

There are only a handful of studies of what scientists do. Anthropologists or sociologists went and spent a few months in a lab watching them and trying to figure it out. Only a couple dozen of those have been done in the past half century. They’re fascinating but barely a start.

Here’s an example of me doing amateur natural history of science, on the PCR method used in covid tests. My analysis may not be good, but my write-up is more accessible than professional ones because I’ve left out the anthropological jargon. metarationality.com/rational-pcr

“Science of science” is almost entirely about how many citations journal papers get. That’s extremely distant from the actual doing of science. It’s cheap and easy to run citation analyses, so “science of science” does that, but it doesn’t seem promising.

We can’t yet do science of sciencing itself, because we haven’t even looked to see what sorts of things sciencing consists of. We should start with what biologists derisively dismiss as “butterfly collecting.” Observe carefully, collect examples, work up preliminary taxonomies.