You'd be surprised how much of the fundamental target discovery literature relies on tool compounds with terrible molecular properties… poor specificity, awful bioavailability, so on.
It makes you wonder whether some of these studies are even hitting the intended target at all