I completely agree that analytics are in their infancy - the problem is that very real decisions with very real ramifications are being made by these infants. I have no doubt whatsoever that there are employees of this organization that have some level of influence that are basing their suggestions on these findings.
I equate analytics in hockey to ghost hunting. You have a hunch something is there, a feeling, so you use any device at hand to quantify something you already believe and are trying to confirm - NOT learn.
An EMF meter shows fluctuations in electro magnetic energy. It doesn't equate that to a "ghost".
I agree with this in general. If we're talking about working backwards, where a conclusion is made, and people look for evidence to back up their statement, I agree that's misleading.
A registered failed zone exit that doesn't lead to a scoring chance against doesn't mean that the player made an error on par with one that does lead to a goal.
In this case, a Dubois shot on goal that leaves a rebound that could potentially lead to something - but doesn't - isn't necessarily a better thing than holding the puck and making a play. In the end, its a "so what?" stat.
In regards to Dubois, I agree. Not only because of what I observe, but there aren't a lot of analytics that go along with what was presented to demonstrate he's performing at an acceptable level. If there were more analytics, or if it was presented on a different player, I'd be interested in seeing it, if anything, just to see it and how it compares to others.
Ultimately the goal of analytics here is to create a blueprint of how and what players should do to create something positive. Its to positively influence the game.
But ask yourself if you really want to watch that kind of game in the first place. Do we want to see robotic sequences of players thinking in terms of percentages or do we want to see players exhibiting read and react skill and the passion to challenge them? Do we want more Matt Roys than Quinn Hughes'?
Besides, the second somebody does something that works we know that there is a cottage industry of coaches working on methods to nullify it. Its a cyclical game, read and react at all levels. So the whole point of "perfection" is moot as the game will just keep evolving.
Haven't we already been seeing this though? The Devils implemented a trap system that, coupled with Brodeur, brought them cups well before analytics was used. The left wing lock was used until there was more diversity in handedness. Players play a system that coaches design where they believe it will yield more goals for than against. And players who don't adhere to that system (or can't outscore their deficiencies) get reduced ice time or benched.
As you said, things change and evolve. They have been since before analytics were a thing. They just give us the opportunity to look at the same events from a different angle. And I think just like anything else, the organizations that utilize their resources more efficiently (including analytics) will end up prevailing more often than those who don't.