positive feedback
Positive feedback is when a circuit feeds part of its output back to its input in a way that reinforces the change, instead of opposing it. Negative feedback is a thermostat that keeps nudging the output back to target, calming things down; positive feedback is the opposite, a microphone held too close to its own speaker. A small sound gets amplified, comes out the speaker, re-enters the microphone louder, gets amplified again, and within a fraction of a second it howls. The output runs away from where it started rather than back toward a setpoint.
Concretely, suppose an amplifier has gain A and you feed a fraction B of its output back and add it to the input. With negative feedback the closed-loop gain is A divided by (1 + A times B), which is smaller and more stable. With positive feedback the sign flips: the gain becomes A divided by (1 minus A times B). As the loop gain A times B approaches 1, that denominator approaches zero and the gain blows up toward infinity. At exactly A times B equal to 1, the circuit can sustain an output with no input at all. That is the doorway to oscillation.
Positive feedback has two faces. Pushed hard, it gives latching, snap-action behaviour, exactly what a Schmitt trigger uses for its clean, chatter-free switching. Held right at the edge, loop gain of one, it gives steady oscillation. The honest caveat is that uncontrolled positive feedback either latches solidly to a rail or grows until the output clips. A useful oscillator deliberately keeps the loop gain hovering near one and adds amplitude stabilization so the swing settles instead of slamming the rails.
If an amplifier has gain 10 and you feed back a fraction B of its output positively, the loop gain is 10 times B. Set B to 0.1 and the loop gain is exactly 1: the circuit will hold a self-sustaining oscillation. Make B even slightly larger and the amplitude grows until something limits it.
Positive feedback reinforces; at loop gain one it sustains oscillation, beyond it the output runs away.
Positive feedback is not automatically bad; it is the deliberate engine of oscillators and Schmitt triggers. The danger is accidental positive feedback in an amplifier that should be stable, which turns it into an unwanted oscillator.