That is, until I installed Team Fortress 2 for office LAN party. Having played it on Windows, it was a world of difference - OS X version seemed so inaccurate, even after getting rid of acceleration.
When you're playing a game like that you become highly attuned to the characteristics of your input device.
I find the same thing happens when playing console games and the "look" stick acceleration curves are completely different between games. It is disorienting until you get your bearings and adjust to the speed and acceleration of each input system.
Just as switching between Windows and OS X means switching from CTRL to CMD, you kind of have to go with the flow on mouse speed, acceleration, and so on.
I play TF2 almost hours every day and I'm often in the top 5 players in the server when I play sniper or soldier.
I play in Mac OS X and I use a Razer Deathadder 3G Spawn. No extra drivers.
However, I tried wiggling the mouse pointer and I can't see a lag on my iMac, but if I grab a Finder-window and wiggle it left-right very fast I think there's actually some lag.
I can't really compare TF2 to Windows since in Windows I use the Razer drivers. But yeah, the mouse in TF2 in Windows feels a bit more "glued on".
Edit: Just tried the Explorer-wiggling in Windows and it too felt a bit laggy.
I really wonder if the "input lag" problem is as simple as a design difference. For instance, if on Windows the mouse position is polled immediately prior to rendering the screen but on OS X it's polled immediately after you would get a different feel.
That would introduce about 8ms of "lag" if you're really dialled in and able to perceive it.
That is, until I installed Team Fortress 2 for office LAN party. Having played it on Windows, it was a world of difference - OS X version seemed so inaccurate, even after getting rid of acceleration.