Completely disagree.
In fact, taking it to the team level, I saw a great (dorky & exhaustive) study that showed the greatest predictor of future team success was simple goal differential. They correlated it to everyone's various favorite "advanced statistics du jour", including Corsi, and found that a more reliable predictor of future wins/losses wasn't Corsi or other stats, etc...., but simple goal differential.
This makes complete sense to me for various reasons. There's a LOT going on on the ice beyond what is captured in a few given statistics, and everything taken collectively results in goals scored (or not). In essence, the goals are resultant from the largest data set of variables combined.
For instance, after roughly 25% of the season the best and worst goal differential teams are:
BEST:
Toronto +24
Colorado +22
Nashville +22
Tampa +22
These are 4 of the top 5 teams in the NHL. Buffalo has the best record, but a GD of only +11, suggestive that the Sabres should drop from here.
WORST:
L.A. -22
Chicago -22
Ottawa -18
Vancouver -18
These teams are pretty bad, but the Blues have the 2nd worst record in the NHL, with only a - 6 DG, suggestive that St. Louis should climb higher.
If you're taking it to the team level then it's not GA/60, it's goal differential, which is a completely different stat. I don't think anyone is questioning that goal differential is a great way to evaluate teams at a glance, I certainly use it.
But team goal differential is COMPLETELY different than skater GA/60. Goal differential gives you important context to define a player's role, style, and performance. With GA/60 all you're getting is what happens on one side of the ice when the player is on.
Taylor Hall, for example, is leading our forwards in GA/60, but he's also pushing play really hard at the other end of the ice. But if you didn't know that you'd think he was just a bad defensive player, when in reality he's just a high event player that gets a ton of ice time and ends up as a net positive.
Presenting rates against without quality against, rates for, quality for, scoring effects, etc is just regurgitating meaningless data for the sake of having data.
By the way, whats the best predictor for goals for and goals against? Could it be...shots for and shot against? Shots certainly give you much more resolution because you have many more data points to evaluate.
Most current models integrate both goal differential and possession together. I'm not sure why you're still harping on this all-or-nothing approach to Corsi, no one has used Corsi in that way for years. We have so much data now that Corsi is just one tool that can be used to evaluate players.