Too little too late. To make that work, it must work right from the start.
IMO the problem was that they didn't think it was necessary to have a system (that works), exactly because the original idea was more like "we came up with the concept of PrCl, but it's probably too awkward and few people will use it, so we present it as an optional DM-DIY tool". They did something similar with custom classes (see the Witch example) but people largely ignored that option, and designers also used it only a few years later and not in the same magnitude as PrCls.
Note that 5e is taking only a slightly better approach, with "guidelines" about building subclasses. It has to be noted however, that it looks indeed easier to create custom subclasses in 5e than it was to create custom PrCl in 3e, at least for the following reasons:
- all classes have the same proficiency progression (instead in 3e every PrCl should define its own BAB and ST progression)
- all spellcasters are either "full" or "half" casters (instead in 3e every PrCl either advanced spellcasting of a previous class or had its own)
- classes and subclasses grant something at every level (instead in 3e there were empty levels also in PrCls)
The problem is (as I mentioned before) that not every single concept will easily fit in a feat chain or a subclass, so there is still room in 5e for prestige classes, at least in the form of custom material (as I said, I don't think they will ever be published).