You need more spokes to provide overall wheel reliability, at least in conjunction with choice of rim. You don't need 36 spokes with a high-profile carbon (or alloy) rim, but you might with a light 300 gram low-profile alloy rim.

The place where high spoke counts become a particular issue is on low flange hubs because there you may have less metal between spoke holes than you've taken out by drilling the holes. In those cases, if you tear two spokes out by breaking the metal bridge between the spoke holes, you can simply zipper around part of the rim. To my comment above, this really only gets this bad in a crash or abusive impact situation, but it does happen. This is where Chris King, who only makes low flange hubs, is quite picky about radial lacing. The real issue about radial lacing is only that if you hit a point on the rim, you are applying loads to a couple spokes that happen to show up right next to each other at the hub (where in most other lacings, adjacent spokes at the rim terminate in very different places on the hub). So if you overload two adjacent spokes on a radial-spoked wheel and the spokes lead down to adjacent, closely spaced holes on a low flange hub, they have a slightly greater chance of tearing loose. Now the incidence of this is very very rare, and it's more of a liability issue for Chris King than anything (the rider who gets injured and sues Chris King because of their own abusive riding). So I don't want to make a lot out of it. As I said above, riding-related flange failures appear to be almost nonexistent -- it takes major abuse or a crash to cause a failure. And by the way, with a large flange hub, the holes in a 36-spoke drilling are spaced farther apart so this kind of issue tends to go away.