I don’t think Jeremy was questioning the legality, his point seemed to be exploring the social implications of the technical decision.
My understanding of “caching” is that I’m getting the content last known to be at a URL, based on information provided by the publisher, and my browser believes it is using that URL. Having a different URL breaks that assumption.
Tim Kadlec’s piece on the session explores the same ideas — in particular the impact of AMP being used for SEO on how the underlying technical project impacts the Web: will it lead to the work on performance improving the Web as a whole or does it just centralise the Web… or more likely to what extent does it do each of these things and is that a “Good Thing™”…
And as both acknowledge, there are legitimate technical questions here, and the answers chosen have a significant impact on what this project might do to and for the Web.