That doesn't really fulfill the "quality signal" aspect of journals, though -- the replacement would be some sort of open journal that still had peer review.

There was an Elsevier math journal where the entire editorial staff got fed up, quit, and then simply founded a new journal on the same subject. That's the sort of thing most likely to bring the big journals down, not self publication by individuals. (Well, that or direct government intervention.)

You're probably right that there's a good value add to come from journals. However the internet works well, and that's a pile of individual pages/web sites, so I think it'll all work out.

I do. But I also publish them in expensive journals, because senior people have told me this is good for my career.

Once I get tenure (knock on wood), I intend to be more selective about which journals I publish in.