So everyone is saying, and I don't deny the theory. But given the extreme rarity of people actually building and riding radials all we actually can say is "bad in theory". Unless tested or properly calculated we don't really know if current materials would actually let us get away with it.
What we do know, because it happens with some regularity, is that it's possible to rip a hub flange apart with too high spoke tensions in a radial.

But that's true regardless if it's a radial or a tangential spoke.

Does it really? I'm not so sure. What'll happen with a radial wheel exposed to torque is that as the hub begins to twist in relation to the rim the spokes will begin to increase in tension, just as they would if they were laced tangentially.
The only difference I can see is that with a tangential lace a smaller angular displacement will cause a more immediate increase in spoke tension - i.e. as soon as the hub twists with relation to the rim this will immediately mean that the spoke's attachment point at the hub is displaced along the spoke's axis.
With the radial lace the first degrees of displacement would primarily move the spoke's attatchment point at the hub perpendicularly to the spoke's axis, thus delaying the increase of spoke tension required to move the rim somewhat.

Now, if we were dealing with a system with some slack in it this would definitely be significant. But bike wheels have their whole survival based around the fact that they are heavily pre-stressed.
It'd be fairly easy to test actually.
a)Label your rear wheel spokes. Measure and register their tension.
b)Select your lowest gear, but front wheel against immovable object. Have Friend #1 support you as you get on the bike and put pressure on the pedals.
b)While you're doing your static honking, have Friend#2 measure your spoke tensions.
c) Compare spoke tensions. Unless pedalling causes significant increase in spoke tension radial wheels should do OK in torque transfer as long as the flanges are up to it.