Re: So, we're now at 2nm
You actually can't even blame the chipmakers, it was the industry organization that generates (or at least used to generate) industry roadmaps that kept the old process naming in place when the technology changed. They could have talked about processes in terms of density in transistors per mm^2 since that's what they effectively meant, but there are no standard ways to measure that. Do you mean theoretical density or maximum actual density. For logic or cache? Do you include dummy cells and other non functional transistors or only active ones, etc. etc. So continuing with the nanometer based names was probably the LEAST confusing alternative for the industry (they don't really care about the general public, foundries, equipment makers, chip designers etc. are their "customers", not consumers.
The chipmakers themselves have abandoned those names, TSMC doesn't call the process making Apple's M1 chips "5nm", it calls it N5. Samsung gives their processes names like 7LPP. Intel was the laggard but finally went along with it with names like "Intel 4".
That doesn't stop everyone else from still calling them "5nm" chips, of course, but that's not the chipmaker's fault.