It is easy to discount the amyloid hypothesis and amyloid research based on the facts that 1) amyloid is not required for neurodegenerative disease pathology, 2) many people who do harbor amyloid fibrils/protein plaques do not exhibit neurodegenerative disease pathology, and 3) amyloid-reducing drugs do not seem to affect neurodegenerative disease progression. It is also true that amyloid is an enticing target for a lot of people with vested interests in this area and those interest may have put too much emphasis on amyloid itself. Like most things in biology though, there is more to the story.

Consideration 1: Amyloid itself is a structural characterization of protein assemblies. Amyloid structures generally form in intrinsically disordered (ie unfolded under normal conditions) proteins as a low-energy conformation that is self-templating and derives its stability from a large number of interactions between many copies of the same protein. Amyloid is not really related to protein function in a direct way. The biological roles of disease-associated amyloid-prone proteins (some famous ones are Aβ, Tau, and TDP-43) can be quite varied in the non-amyloid state. These proteins are also associated with neurodegenerative disease through orthogonal means, namely through genetic analysis of disease-related mutations, so it isn't just the fact that these protiens wind up in amyloid fibers that leads us to believe that they're associated with neurodegenerative disease. So, we know because of genetic analysis that these proteins are associated with disease, they can all form amyloid, but we aren't losing any consistent biological function due to their dysregulation–this strongly suggests that the amyloid or otherwise dysregualted state of these proteins is problematic on its own.

Consideration 2: Recent structural work has tied different amyloid isoforms to different disease pathologies. I work in an adjacent field to this and these results were really what changed my mind about the validity of the amyloid hypothesis despite years of doubting/discounting it (often in a public way, often to the detriment of my own career–turning the corner on this was a bit challenging). See this paper and the first 5 citations in it (https://www.nature.com/articles/s41586-021-03911-7) for a good example of direct structural links between amyloid structure and disease pathology, and the associated comment (https://www.nature.com/articles/d41586-021-02611-6) for a good overview.

Consideration 3: The presence of amyloid plaques in patients in the real world is generally assessed by histological analysis or other methods that don't capture information at the molecular level. This, combined with the two points above that seem to point to some shared structural features of amyloid-prone proteins being disease causative, suggests to me that studying amyloid (or maybe more specifically the behavior of amyloid-prone proteins) is still a worthwhile endeavor.