alea, legacy,
(as memory serves the membranous wings of bugs as thermostats versus the feathering of highly articulated limbs in birds represent alternate paths convergent upon the same value point - flight).
While the SED gambit seems all of incremental and comparatively tractable, even there legacy persists. In this case the improvements over the legacy and the ability to cooperate with legacy implementations are the points that hopefully bag sales.
It seems to me that legacy is a comparatively small hurdle in SED implementations compared to NAC implementations which by my reckoning require complete conversion. Presumably Wave has jumped on SEDs for this reason, and of course the obvious object oriented vectoral understanding that losing a laptop with liability-laden data presents.
I've wandered into a valley of less optimism lately, and am beginning to subscribe to the notion that molasses will equal failure.
BEWML would/will do wonders for my optimeter.
The above content is my opinion.