Comment by mandevil

As to the uncertainty and mysteries, you are 100% correct. One of the big failure modes for engineers in dealing with human health is the assumption that things are as simple and logical as the stuff we build, when it's simply not at all like that. There are (1) big arguments over basic things like "why do SSRI's work?" Outside of LLM's I can't think of a thing in software where we are still arguing about why things work in production. We never say "Why does Postgres work?" in the same way. (2)

And yes, this is true for many other areas of discussion at HN. It's just that it is most obvious to me in the area that my wife specializes in, because I pick up enough via osmosis from her to know when other people don't even have my limited level of understanding.

1: Or at least were 15 years ago when my wife told me about it- the argument might have been largely concluded and she just never updated me since I don't keep up with the medical literature the way she does.

2: Two decades ago there was a huge push for the "human genome project" under the basis that this would be "reading the blueprints for human life" and that would give us all of these medical breakthroughs. Basically none of those breakthroughs happened because we've spent the past 20 years learning all of the different ways that it is NOT a blueprint and that cells do things very differently from human engineers.