Utterly dumb, complicated and/or pointless. I can only see them doing 14nm-based PS5 in 2017, because they are unsure about AMD's future, whether or not it's gonna go 10nm for potential 2019(-20) release of PS5 or skip it and go straight to 7nm for 2021(-22) potential release, stay alive, etc.
Releasing PS5 in 2017, taking full advantage of 14nm node shrink(Zen, Vega), would allow them not to worry about such things and sit on it at least until mature 7nm arrives, but that won't be an orthodox console launch nor would it be an orthodox console, it won't have any serious exclusives in launch window and even though for a first few years it will be sort of compatible with the PS4(because all of the development is centered around it), eventually PS4 would have to go(in a way that next ND game, or, say, GT7, won't come out on it), it must die in that scenario to not drag PS5 back, which is the whole point of this topic, right?
Is it worth to kill ps4, a console that potentially could end up reaching ps2 results(well over 40m units sold through now), over worry?
I think they should just stick to conservative way, introduce PS4 "Ready for Everything" Edition with PS VR and 4K(BD UHD) built-in support, maybe even make it standart version and not offer anything else with 14nm revision(btw, didn't AMD say it will offer node shrink first to Sony and MS? That means we should get 14nm PS4 in a few months, but since there is no noise about it yet, we probably won't)