While I get what you are saying, I mostly disagree. You have to differentiate between if the problem exists and whether the solution is any good. When computing became really mainstream we were standing on the shoulder of giants. A lot things had trickled down from things like research and were really good for the time. But they also lacked in a lot of things we care about today like encryption, automation, design, cross-platform and security. Today these things aren't coming out of research, nor necessarily large companies, but some random person who can see the problem. It would be a lot better if something like authentication was included in HTTP, but that isn't happening for various reasons, so it is going to be a job for a framework. That isn't necessarily a good solution, but it a solution to a problem that exists.

It really seems like things like PARC was an anomaly. Companies eventually figured that it was just better to "be part of the problem" and make a lot of money like in many other industries.