Silly me, forty years of engineering we call them "negative feedback" loops. When you write out the differential equations defining a servo loop, the feedback term should be negative if the loop is to be stable.
Positive feedback can be at best "metastable", balance a pencil on it's point, it will stand straight up until the slightest disturbance causes it's center of gravity to move slightly out of vertical alignment with the contact point. This applies a small torque increasing the magnitude of the disturbance which continues to increase as the pencil falls over. (positive feedback)
Negative feedback may be illustrated by dropping a marble into a spherical bowl. The marble will roll "downhill" into the bottom of the bowl. It will probably over shoot and roll past the bottom and up the side some but then it reverses course and rolls back down until it finally comes to a stop at the lowest point. This oscillation about the rest position is typical of a negative feedback system w/o a damping term. If you tried the experiment with various liquids filling the bowl, you would find that water has little effect, light cooking oil has more of an effect, and honey will prevent the marble from overshooting the rest position completely. The viscous drag of the liquids provides a "damping" effect the will slow the marbels's travel and cause it to stop at a stable position. Critical damping is achieved when the moving marble reaches it's rest position in the least time with out overshooting. A little less damping will allow a small overshoot and a few oscillations (under damped). A little more damping will increase the time it takes for the marble to arrive at it's rest position (over damped).
Most positioning servos system are intentionally designed to be slightly under damped as the minor oscillations near setpoint improve repeatability by reducing hysteresis.
You're right, of course!