Consider the following tube transmitter with an adjustable pi-net output. It will be easier to understand if we assume class-A operation. Assume the tube has a preferred load-line, the one for which it was designed. Let's call that load-line a Zg-match. Let's call any other load-line a Zg-mismatch. For the sake of simplicity, let's assume ideal conditions. At key-down, the tube sends a signal from point 'X' to point 'Y' which takes one nanosecond to get there. If the impedance at point 'Y' equals Zg-match, there will be no reflections from point 'Y' back to point 'X' and the tube continues with the Zg-match load-line. No feedback from point 'Y' causes the tube to continue with the same Zg-match load-line. If the impedance at point 'Y' is not equal Zg-match, how does the tube know that fact? It can only know what the impedance at point 'Y' is if the tube gets some sort of feedback from point 'Y' and that feedback cannot reach point 'X' for another one nanosecond. What is the nature of the feedback that tells the tube that point 'Y' is not equal to Zg-match and is instead equal to Zg-mismatch so that the load-line can be changed from Zg-match to Zg-mismatch? What else could it possibly be except reflected voltage/current? Please be specific. I am surprised that the lumped-circuit model has caused so many otherwise intelligent people to completely miss this simple concept.