Speed of the servo has nothing to do with 'accuracy' from an electronic viewpoint. The servo may resolve the receiver signal into x number of steps which might affect the rotational degrees of a 'step', however, the accuracy is determined by the 'dead band' allowed by the servo. The 'dead band' is how close does the servo make the signal from the pot match the signal from the receiver. The broader the dead band, the bigger the window of rotation that the servo considers to be 'correct'. This accuracy is for the servo only, it ignores and linkage or mechnical slop.

Speed is a matter of gear ratio, voltage and motor strength combinations in the servo. You can take a servo at 4.8V and it will have a certain time frame for x degrees of rotation and have a certain amount of torque. Up the voltage and the time to rotate that same x degrees decreases, the torque probably increases, but the accuracy remains the same.

The tighter the dead band, the more the servo works to find 'center', and the closer to center it will return each time. When you hear a servo 'buzzing', that is the servo trying to find center. The weight of a control surface, drag in the linkage, etc, may inhibit the servo ability to land on 'zero' and stay there. It is easier for a servo manufacturer to build a servo with broader deadband, the components can have a little more tolerance, etc.

Woodie