I'm not sold on either QoT or QoC being the more important. Until I see some good statistical analyses, I remain ambivalent. The argument that, over the course of season, QoC evens out seems specious.
Using them feels very eyeballish. I don't get why no one has developed an adjustment variable that includes things like TOI, cf, scf. It seems like such an obvious thing to do (though a ton of work, no doubt).
With and without are reasonably good for individual/line impacts (but still have big blind spots).
never understimate the ego angle. they don't actually want to find it, they would prefer that it didn't matter.
1. cf% was a huge breakthrough, and was very powerful, and seemed like the better way to describe player quality - so it was only natural to then try and measure QOC using cf% as well. this naturally ended up indicating that qoc had little impact....but only because it was circular - I.e. if we are suggesting cf% can be warped by qoc, then measuring qoc by cf% is self defeating and cancels itself out, as it will view players with high cf% in easy usage as "better competiton" than players with lower cf% in much tougher usage.
2.half the fun in analytics was outsmarting the hockey people. pointing out when they were stupidly playing bad players in top roles and good players in minor roles. most of the great success in early analytics was based on finding all the overrated overplayed guys (hello big slow slugs) and underrated underplayed guys (hi there small speedy guys). after doing this over and over with great success, it's naturally very hard to then go back and want to believe that coaching/management decisions like TOI could be a useful indicator of player value. it went against everything they had figured out to that point. so they naturally resisted it.
3.And now, even though they have started to appreciate indirect measures like TOI (partly because the league has started to appreciate the analytics in their TOI distributions), they still naturally lean more to dismissing it than to endorsing it...or even fully exploring it. So even though they have found some correlation with TOIqoc, they are more eager to endorse ways of dismissing it - so when their first stab at using straight TOIqoc shows them a small enough correlation that they don't have to worry about it much, then they are happy to leave it there without further testing. So at this point, we're at a stage when TOIqoc has been looked at without though thinking about what it actually means. We know TOI can only ever be a proxy for quality, yet the experiments so far have taken it as an actual measure or quality - I.e. that if your opponent's average 16mpg they are literally 33% better than opponents that average 12mpg. Myself, I think that's obviously not the case - McDavids are more than 33% better than Browns....more like 333% better. But for obvious reasons ice time cannot be distributed by actual player quality. So I'm pretty sure when the some analytics guy actually stops trying to dismiss TOIqoc and starts really trying to work through it, I'm pretty sure they'll get there.
but as for the qot v qoc - again, I think the relative stats do a decent job of giving us something to at least eyeball adjust the raw numbers for team/teammate effects. in my mind I am always balancing the % and rel stats to try and get a better idea of what they're actually doing out there, and then I try to add in the qoc in equal share too.