I happen to prefer the effects of curves to linear probability in results, but for that in particular, I'd rather go to 2d10 replacing the d20. I'm sure that would go over like a lead brick, though.
Opposed rolls (or anything else that makes the DM a potential bottleneck in handling time) do not scale well as you add players. I've heard plenty of people that think that 6+ players is Something Man Was Not Meant To Do, but we like it, and when we really like it in a game, our large groups buy twice as many books as your smaller groups. So at least throw us a bone every now and then.
However, in general, if we are to have opposed rolls with d20s, then the opposition roll needs to mean considerably more than, "sorry, you failed after all." Lost Soul layed it out exactly. You've got four possible outcomes. All of those four should be different, at least most of the time, or opposed rolls are then merely a sop to players to make them feel better about ability they don't actually have--AKA "illusionism"--which should be put into a rocket and sent into the center of a sun.
One way that opposed rolls would be worthwhile is to shed a bunch of the other things that simulate more intricate action (e.g. some or all of multiple attacks, more complicated spell attacks, 4E secondary effects, opportunity attacks, etc.) Then build such things in the opposed rolls, so that if you succeed in an opposed defense roll, you get to adversely affect the attacker, not merely stop them. A real "riposte" from warrior types would be fine in this environment. Now, you are trading slower resolution of a given exchange for faster combat resolution overall (because win or lose,
something is likely to happen that will move the action towards the finish line).
This makes trying risky things against superior opponents often counter-productive, but a certain amount of that can be seen as a good thing, and the rest can be regulated by limits on how much a character can so react. (That is, the wizard out of spells rushed in to try and smack the orc with a quarterstaff, even though it probably will earn him a riposte, because this sets up the fighter and cleric to finish off said orc.) Perhaps each defense roll after the first gets a cumulative -2 penalty per round.
However, if we have such opposed mechanics with meaning, then there should also be some limited ability to do something similar in magic versus magic effects. An enemy cleric tries to "hold person" on a fighter from across the room, the best the fighter can do is try to toss a dagger or javelin at him. But try the same thing on a wizard, and he's got counter abilities to make you wish you had not. Of course, the opposite would be true for trying to cast a spell on a warrior in melee.