I don't think so. A/B testing is just not a cure-all and alternative for good design, that is all. I mean, if the pattern is 'don't bother thinking too much about ux, just A/B test everything in PROD to death', then yeah, you are right.

However, there are many products which allow users to enter into a beta or test group, where they are the willing subjects to their experiments (in exchange for the latest shiny new stuff). This has the aspect of consent, leaving the 'stable ring' free of such variability. The fact that google and many startups are not using such consent and offering stability, doesn't mean it can't be done or isn't done.

> "You're doing it, but in some subtle but very important way that's not at all obvious to you, you're doing it wrong."

Doesn't that exact statement describe how Agile (or many other concepts, really) is used in a lot of companies? There's nothing wrong with it in principle, but practice is all over the place.

I think of it as validating and improving imperfect user assumptions.

This reads exactly like parody corporatespeak jargon.

Clearly it's the users who are wrong with their stupid assumptions and we will correct them with our mandatory weekly update.

Assumptions about the users, not assumptions made by the users.