Simple answer is that structrures 1st Nodal point in a vibrating structure will be 1/2 its length.

So, your aileron is fluttering. Actually everyones ailerons and all moving structures are fluttering.

What is Flutter?

The ends of the aileron, tip/root, are moving up and down, up and down, up and down due to slack in hinges, slack in servo linkage, slack in servo gears, vortex bubble shedding due to high angle of attack, vortex shedding due to drag which increases by the velocity squared and with the difference in drag between the bottom and top of the airfoil at the trailing edge creates a pressure up/down depending on flight profile. Your tip/root of your aileron will always be flexing as they are a structure under said load. The only question is really how much.

Your servo is trying to dampen out all of these vibrations. If you place your servo linkage at the most common node point the structure at this point is actually NOT moving until a third harmonic or greater is introduced. Therefore your servo is not able to dampen the structural response and the flutter, while not noticeable under most conditions is free to either get LARGER, or SMALLER. Well, obviously you never notice when it gets smaller, but when the vibration/flutter becomes large enough, the dynamic force of structure is large enough to force the servo to move. Generally this strips the gears and your plane just lost aileron authority and probably went into a spin unless you are VERY quick on the sticks and know to apply opposite aileron throw, which may or may not help depending on how much you use.

Placement of linkage.
1) Higher pressure air under the wing is pusing aileron up
2) You do not want to tip stall
3) Pushed up aileron decreases angle of attack
4) Linkage placemtn along lenght of aileron will increase/decrease this flexing of the tip of the aileron
5) If one is trying for VERY snappy barrel rolls on a lightly wing loaded plane where stalling is not an issue, then place linkage at some prime number integer of the length of the aileron towards the tip.

FOREX: 18" Aileron. Ribs every 2" 5th rib is centered on center of aileron. Prime integers of 18/2 will be 1/9, 2/9, 4/9, 5/9, 7,9, 8/9 just to be VERY thorough. Obviously 1/9 and 8/9 are essentially useless as the flexing of the aileron will mean one end is fully functional while the other essentially sits in place. 3/9 and 6/9 are NOT good as the next most common harmonic ossilation of a structure is its length divided by 3. 2/9 and 7/9 length essentially fall under the same problem of 1/9 and 8/9. This leave 4/9 and 5/9 length. This position along the aileron will mean that any flutter of the most common harmonic 1/2, will be trying to move the servo arm back and forth as well. Servos do not like this as their "brain" told them to HOLD @ 'X'. The flutter is trying to move said servo OFF its HOLD spot. Therefore said servo PUSHES back against this flutter canceling the flutter and keeping your airplane in the air.

There is another component to this in that the servo in question has to be able to ACT quick enough to PUSH BACK. This gets into latency, torque curves. Why digital servos are superior essentially. A standard pot(resitor) servo will not Push back with its maximum current(POWER) until said resitor(position sensor) moves by a LARGE margin. So, yes, said servo will be able to bring 'y' torque to bear, it only does so when its position sensor tells it that it has a LARGE movment to under go in a simplifed explanation. A digital servo does not have this limitation as it can bring its FULL current to bear on even a minute needed change. Well, nearly its full current anyways.

Hope that is sorta clear. Others on here can probably do so in a clearer fashion. Gotta go. Questions can get to them tomorrow maybe.