开云体育

Hi Keith,

Don't conflate my being didactic with being angry. Email is a lousy medium because the reader will attach an emotion track based on scant clues. That is a noisy, error-prone process.

That said, I would refer you to my initial response to the OP. Not surprisingly, I prefer my example to yours. I have taught feedback to a large collection of quite diverse audiences, and have settled on a particular set of explanations as a result of many pedagogical experiments.? I craft the particulars to match the background of the intended listener. It is very tricky to provide simple answers that also do not introduce fundamental errors.

The reason I began my response with the quote, "You cannot control what you do not measure," is that many folks have heard it before, although likely in other contexts. Even if they are unfamiliar with it, it makes intuitive sense, and it captures the essence of what negative feedback is about: You need to measure the variable to be controlled. That's necessary (but not sufficient). So in that likely familiar quote is the notion of control based on some sort of measurement of the thing to be controlled, which also then implies the existence of some sort of reference that conveys when that control is successfully achieved.

The example I offered was the thermostat. Everyone has one in the home. Everyone knows what it's supposed to do, even if the particulars of how it does what it does may be mysterious. It's a commonly encountered negative feedback system. All you have to do is point out a few of its features, and how they map to features of the quote. This simple answer to the OP's first question has all of the attributes that you enumerate, and has the added distinction of not introducing fundamental errors in the process. I teach a freshman class that has many non-EE/non-STEM students. They all grasp negative feedback's essence from the thermostat example. I had an occasion to test that explanation again this past term. And it triggers the right set of follow-on questions.

The discussion about noise wasn't a digression unrelated to the first. A proper description of negative feedback, simple or not, should set one up for answering more sophisticated questions. A poor simple explanation will align neurons in a way that actively militates against an intuitive understanding of the OP's second question (or is it the first?). Once you've got the right block diagram implicitly or explicitly implanted in their crania, adding a couple more inputs (noise or signal) to the system poses no cognitive problem. If you haven't given them the right block diagram, answering the noise question becomes nigh impossible.

So, mere simplicity is not a virtue. There's a lot of engineering that should go into crafting an answer that is both simple and correct. That kind of simple answer is scalable to address more sophisticated questions. I would go further and argue that a good simple answer stimulates precisely those types of question. A random simple answer is often a "lie that we tell to children" to get them to stop asking questions. That's bad enough, but even worse is that answering the more sophisticated questions that do get asked requires undoing the simple explanation. Why do this when there are demonstrably better alternatives?

-- Cheers
Tom

--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070