Hector Martin

This is what I mean by loop gain. These days a lot of discussions seem to start right off the bat with effectively "Look at how evil this person is". People are immediately boxed as the "other side". This is not conducive to productive dialogue. It just brews more disagreement.

There’s a serious and real problem with this idea, though: When we force people to constrain their tone— to minimize ‘loop gain’— we’re also effectively forcing victims to expend mental energy to massage their arguments into a form that shows no upset. It’s inherently silencing.

Tfw the tweet youre replying to said runaway _negative_ feedback and I can't overlook it XD. More seriously, well said and I totally agree with you :).

Closed loop gain has to be > 1 when phase of feedback is > 180 relative to input for oscillation to occur. Phase > 180 means multiply gain by -1, so negative becomes positive feedback.

That's a frequency-dependent analysis, yes. Usually the steady-state construction has negative feedback, which then becomes a positive gain of >1 at a given frequency due to loop delay putting the feedback out of phase.

I don't understand what you're talking about unfortunately/how steady state plays into it. I'm thinking mainly in terms of phase margin (a freq-dependent analysis?)

I'm talking about the problem of control system stability. A closed loop control system has negative feedback (take the error signal, invert it, apply it to the input somehow), but that can yield positive gain at certain frequencies because there is loop delay.