I'm not Don, but... I agree that names are necessary and more reasonable (i.e., with both your points). I just don't think making them part of the standard name scales *in an observing framework*. The key point here is that I know what an instrument (think 1000s of instruments in an observing system) can measure, and I'm going to move the instrument to different places, and I don't want to change the name of the measurement -- even the standard name -- as the instruments move around. The post-processors and modellers can do that if they want, but for operational concerns it is an inappropriate linkage.
I understand, but I think in practice you would not have to change the stdname for this reason often, would you? I'm guessing you're not likely to use the same sensor and/or measurement technique for a quantity at sea level and at the tropopause, for instance. Best wishes Jonathan