Is it not just simpler to say it's providing a little positive feedback to raise the input impedance?
Indeed, and since the gain of a follower (of any type, be it BJT, FET or op-amp) is always slightly less than one, the positive feedback can never be enough to generate negative input resistance, let alone oscillation.
Indeed, the advantage is inversely proportional to the difference from ideal (i.e., Av = 1 - epsilon, for epsilon ~= 0), so that a gain of 0.5 is able to double its input resistance, a gain of 0.99 improves it by 100x, and a gain of 1.0 improves it infinitely. Assuming, of course, the coupling capacitor is large enough by the same factor. The input resistance starts rising at F1 = 1 / (2*pi*Rin*C) and levels off around F2 = 1 / (2*pi*Rin*C/epsilon).
Note also, an impedance which rises with frequency looks like a certain component...which one?
Indeed, the input impedance has an
inductive characteristic between F1 and F2, and a circuit designed specifically for this (called a gyrator) can be used to synthesize inductors from capacitors, among other things.
Positive feedback is neat stuff; if you add just a little bit of gain to the circuit, say, suppose you place a noninverting gain-of-2 amp after the follower and bootstrap its output to the first input: now the output goes the same direction as the input,
and then some, so that the output wants to drag the input along for a ride! If the source is able to keep control over it and it remains stable, this is simply negative resistance in action! If you add a potentiometer before that gain stage, so the output is variable from 0 to 2 gain, you can clearly observe the input resistance rising, from its initial value, up towards infinity (at exactly Av = 1), and wrapping around to negative numbers, until settling at -R_initial at Av = 2 (or, even smaller resistances if you keep adding gain, of course).
I like this example, on a fundamental level, because it demonstrates the continuity of the real numbers, from zero, to positive finite, to infinity, to negative infinity, to negative finite, and back to zero. There are indeed many physical and mathematical situations where this becomes apparent; this example plays merely one humble testimony to that deeper truth.
Negative resistance is hardly a useless, unstable gimmick; it can be used for a number of (stable) purposes, like compensating for voltage drop along transmission lines. A direct example regarding the bootstrapped emitter follower would be, overcompensating to account for other loading and loss and leakage at the input, so that when the whole circuit (source, follower, and other loads attached) is considered, it indeed achieves a near-infinite impedance.
Tim