I think he was hopeful that the same benefits could be achieved with a different server-side runtime. It's a long-shot, but with good socket-sync and something like the client-side MongoDB adapter, Rails could come pretty close.
We got pretty comfortable with the commodities the Rails environment provides, so it's understandable some people would rather just add another layer (Backbone/Spine) and avoid the paradigm shift.
The advantages of attacking new problems with a new toolkit can sometimes appear less beneficial, because swapping out the current mega-stack for something slimmer inevitably leads to reduction in features (ie.: You end up making another `view` in place of a collection partial. Mailer, gems/middle-wares, etc..)
That said, despite getting pretty comfortable with Rails/Backbone|Spine, I'm in the pragmatic camp, looking forward to tackling a few side-projects with Meteor.