Re:Life After Eroom’s Law
by refoundable on Monday December 28, 2020 @12:31PM
(#25561095)
I agree that discovery is far from a straight-forward process, and that many of the cavalier ways in which it happened in decades past would be criminally prosecuted and beyond the pale today.
But I don't think that there was ever much low-hanging fruit to begin with. Discoveries always look "obvious" ex post and almost never ex ante.
I find it best to think of "low-hanging fruit" as a euphemism that polite people have agreed to use in order to excuse the massive failure that has taken place and stop curious people like Scannell from asking too many questions.
My understanding is that MAOIs are still more effective for mood disorders than SSRIs, but have the unfortunate issue of being toxic when taken with just about anything (and not just drugs, e.g. most fermented foods have fairly negative interaction effects with MAOIs).
>In the course of his own later investigations, Scannell began to focus on a serious technical problem called “model validity,”
(Sentences adjacent in original.)
The problem, here, I think, is backwards. I have a little window into the world of drug discovery by a decade-long interest in nootropics (tl;dr: I don't take them anymore).
These statements aren't wrong, but they obscure the history. The way we discovered antidepressants was mostly like this:
- tuberculosis patients treated with hydrazines become "inappropriately" happy
- modified hydrazines first used to treat depression
- hydrazines shown to increase serotonin levels by monoamine oxidase inhibition, now called MAOIs
- serotonin hypothesis of depression developed
- serotonin-targeting drugs developed for depression
There are a few more steps here, but the point is: the low-hanging fruit was the hydrazine, then we climbed the tree and got SSRIs.
The problem is that we started climbing all the trees with low fruit but never developed a way to find new trees. We also stopped looking for low-hanging fruit, in that we don't test random drugs on people the way we used to in the '40s. Everything has to be justified by an existing theory even though the existing theories are nowhere near comprehensive or complete.
I don't think we should start testing random drugs on people, but I do think we need to avoid this framing of the problem. We don't understand human biology that well. Our primary focus needs to be not on developing drugs per se but on understanding the extremely intricate hydrogen-bonded nanomachines they're supposed to modify.
But relative to that, our efforts to find drugs should be tempered by a certain humility about the validity of even our most "well-understood" theories. Most of them were developed by accident. The whole risk-averse investing and management apparatus wants reliable theories, so researchers need to be louder about the fact that in many biology problems we don't have reliable theories. We have moderately supported working models.