I am opposed to evasion, in that I believe that it reduces the skill involved in the game, and increases the presence of luck, which I do not believe would be in any way beneficial to the metagame.
~ Obviously, like any other stat-ups, evasion can be phazed. This is, in a way, easier than phazing the likes of Swords Dancers and Nasty Plotters, simply because a Double Teamer doesn't pack the insane power that offensive stat-uppers do, and, as far as I know, all the common phazing moves go straight round evasion. The biggest problem, in my opinion, with this argument, is the fact that quite simply anything (Well, nearly) can learn Double Team or Minimise. It throws phazing out the window. Sure, if you start Double Teaming with a Tyranitar, I'll just send in Hippowdon and Roar you out. But if you start Double Teaming with a Vaporeon, what can I do? I'd then have to stick Roar on my Vaporeon counter. The list goes on; it's just overcentralising. The same thing can arguably be said about Gravity; so few viable Pokes learn it (To my memory, Blissey and Dusknoir) that it's very situational to suggest it as a way of getting past evasion.
~ Aerial Ace, Faint Attack, Aura Sphere, any others I've missed, are another commonly used argument as a way of bypassing evasion. They're perfectly effective against anything fragile, but two of them will hardly scratch anything that's got any kind of defenses, which most Double Teamers will have, whilst the other is only available to two Pokes in OU, only one of which gets STAB on it.
~ The most common argument I've heard is that it's no different to moves just not having 100% accuracy, or the 'natural' critical hit chance. Thing is, you can't feasibly ban all moves without 100% accuracy, no side effects that aren't guaranteed to occur, as you're left with such a small range of moves that the metagame is just going to become overly limited. Same goes for crits; keeping Lucky Chant up the whole game isn't practical. And even then, there's the fact that moves do variable damage anyway. However, in an attempt to make the game as skilled an affair as possible, minimising the presence of luck is, in my opinion, the best way to go about it.
~ As for the difference between Double Teaming and letting the likes of Gyarados stat up; the difference is simple. I know that if I let Gyarados get a Substitute, he's going to be sitting on about 500 Attack and 350 speed (Not sure how accurate that is) and is in a position to do damage to my team. However, I can switch in my Starmie/Gyarados counter, knowing that I can survive an attack unless they run bite, and OHKO back with Thunderbolt.
However, if the same Gyarados decides to use Double Team instead, I don't know what to do. I know he can 2HKO my Starmie, but I can OHKO it. I switch in on the DT, miss with Thunderbolt twice, and lose my Starmie, opening my team up to a Gyarados sweep. Alternatively, I can hit first turn, and I've got nothing to worry about. The difference is that the game becomes about predicting the RNG, as opposed to predicting your opponent.
~ As for the 'if you were more skilled, your opponent wouldn't be able to use it anyway' argument, I'm not sure :s The simple fact of the matter is, unless you have one Pokemon that is faster than your opponents entire team, which is also in a position to OHKO them, then they will have a chance to use [move]. If [move] happens to be Double Team, then your opponent will be able to get one off, however much skill you have.
All in all, whether you think evasion should be banned depends entirely which viewpoint you take; that luck is a given in the games and so we may as well just accept it, or that in order to make it as skilled an affair as possible, the presence of luck should be minimised. Personally, I would be inclined to agree with the latter, as my overly one-sided, opinionated post probably shows :p However, don't get me wrong, I do believe that testing would be a good idea; I just don't think that testing is going to change anything.