The thing that seems to be missing from these discussions (except for the tow from the CG) is the fact that the problem is the up force of the towline exceeds the ability of the tow plane elevator to resist. I do not know how much downforce the elevator can create, but it is a LOONG way from the breaking strength of the rope or even any weak link you can reasonably tow with. Near the breaking strength of the rope, I think you will find the angle that will provide enough lift on the tail to "win" over the elevator, will be quite shallow. And of course, that angle at the tail is the same change in angle that might happen from the tow pilot reacting to turbulence. All the thought of angle detection, make me think the inventors have never towed in rotor, where we see lock to lock control movement fairly regularly.

My favorites in this thread, are pushbutton release, no effort, fast, accessible, and yes, in bad cases, too slow. And perhaps the integrating load cell to look for a sustained high load, but we have no data to know if that is a valid condition to look for. A few of our upsets (thankfully at altitude) we from a glider release, in a climbing turn, that resulted in an instantaneous nose down of the tow plane, not the sustained load that is thought to happen in a kiting situation.

I am afraid, without a means to tow from a point on the tow craft that does not influence pitch, we are stuck with better, and recurring training.

RR