I thnínk the more interesting action is going out by the rim rather than by the hub.
Ideally(from a propelling perspective) one would like the spokes to be attached tangentially to the rim, in order to have all the tension increase in the spokes immediately and completely translated into a force perfectly aligned to move the rim along.
This isn't possible of course, so let's look at the options.
With a radial spoke(assuming "zero" displacement) there's NO component in that direction(the spoke being in a sense perpendicular to the rim and being unable to transfer a bending moment), all tension increase from spoke wind-up goes straight into trying to collapse the rim w/o doing anything "useful" from the rider's perspective.
With a spoke leaving the hub tangentially the spoke is at an angle out by the rim, causing at least a fraction of the increased spoke tension trying to pull the rim forward rather than only trying to collapse the rim.
So with a radial lace a minute part of the tension increase will go into moving the rim along, with a huge amount simply "disappearing" into trying to collapse the rim, while for a tangential lace apparently a sufficient amount gets translated into a propelling force.( not that the tangential doesn't do a fair job of trying to collapse the rim too...)
So why aren't more people touting the superior acceleration properties of high-flange, high cross hubs?