It really is the last resort; even if you were able to find a psychiatrist willing to administer a course of it without extensive trials of multiple classes of psychotropic medication as an adjunct to psychotherapy, your insurer is definitely not going to spring for it. The NHS’ NICE guidelines outline what’s to be considered before ECT is even proposed (https://www.nice.org.uk/guidance/cg90/chapter/1-Guidance#seq...), and criteria are at least as stringent in the US.

A lot of medicine is rather more grey than black and white than we'd prefer, since we either find things out through long tails of researching lots of things and seeing what works better than placebo, or by verifying unexpected reported outcomes (several entire classes of medications have come from unexpected side effects - Rogaine and Viagra were blood pressure meds, the entire SS*I class of drugs was found when someone noticed by accident that a type of antihistamine seemed to improve malaise in some patients, ketamine is rather well known at this juncture as an anesthetic and party drug that got signoff as a novel antidepressant...), and observational evidence isn't necessarily predictive of how it happens on its own.

You're not wrong that a lot of psychiatric medications are "this seems to work, who knows why" and experiments around testing mutations on those things to see if they're more or less effective, because we have so little visibility into the brain and cognition that we can't really gauge these things objectively v. well.

But for people with pathological brain problems, big unpredictable hammers can be a reasonable choice to try if you have an urgent need for mitigation (e.g. acutely self-destructive problems) and/or less invasive things like directed behavioral therapies have not proven sufficiently effective.