using word middling hand doesn't really make sense. just bet whatever doesn't fold out too many worse hand combos.
if shove gets called byy 1 combo vs betting 2/3 pot getting called by 9999999 combo clearly bet 2/3

Your original post is neither true nor is it totally false, Dan.
The matter of fact is: we can't simplify things like that to an overall rule that always applies. We have to think what we want to achieve with our betting. What does betting want to achive in general? It wants to maximize the expectation of our strategy. So we also need to find a betsize that maximizes the expectation of our strategy. If there was a simple formula or given betsize that does that (maximize EV) in every situation every truly good poker player would probably lose very much of his edge since i think betsizing is one of the things even some good mid or even hs regs do very suboptimally.
In a nuts/air scenario on the river our choice of betsize is always very simple: all in (if that isnt clear to you i would recommend you to watch the toygames series of sauce here on rio, the best series discussing GT out there imo), but the nature of poker is usually very much NOT like nuts/air vs bluffcatcher only. Usually ranges are continuous and distributed very differently amongst players ranges. A decent start for picking a betsize would be to figure out range assimilarities and then sizing such that it favors the hands in your range that he is lacking off. Just try to think about every situation in a vacuum and what strategy he will/should use to play against every one of your sizings and then try to figure out the sizing that you think should maximize the expectation of your strategy with either the power of your brain or the power of programs like CREV. Often a good indicator of a good betsize might be that you feel like it will be very difficult for him (or for yourself if you were in his shoes, maybe because one of your opponents keeps putting you in that spot and you always struggle making the right decisions) to find a good counterstrategy.
Lets make a quick example of a single raised hu pot. The flop comes QQ2r. You decide to bet 80% pot on every street. Now I would call that a pretty bad sizing. By the River your opponents strategy will probably be to only call with trips+ making your bluffs very unsuccessfull and also getting very little value even from your good trips (AQ KQ) due to card removal. Betting either way bigger than that (targetting trips with a very narrow value range) or betting smaller on every street (which should usually be the preferred option), forcing our opp to call down with hands way weaker than trips should both be strategies that are supperior to betting in such fashion that it will be correct to defend the river with all of his trips but no hands weaker than that.
While this spot is a bit easier to navigate thru since the board is very static and ranges are easier to understand in comparison to most other situations, the capability of playing those situations that are more difficult optimally aswell (or at least as close to as we can get) is what will take you so much work and what makes a really good player be really good.
That being said i wish you gl on your journey, I hope i could help a little bit.

@omgimsexy
I was trying to say this " So we also need to find a betsize that maximizes the expectation of our strategy. "