Defensive play is a tough thing to quantify. If you're going to make a claim like a Norris Trophy nominee being worse defensively than they were earlier in their career, you'll have to have pretty cogent reasoning. Because this is a complicated breakdown, the first thing I want to focus on is defining what makes good defense.
Traditionally, hits and blocked shots are most often associated with defensive play, as both attempt to quantify a player's ability of willingness to reduce goals against. However both of these statistics focus solely on decisions made without the puck, and large totals of both are indicative that the players in question have trouble getting or keeping the puck in general.
Similarly, turnovers are seen as a sign of poor defense, however the more often you have the puck, the more likely you are to turn the puck over. Raw counting stats for hits, blocked shots, and turnovers can actually mean the complete opposite of what general fan consensus believes them to mean, rewarding players for not having control of the puck.
So what does make good defense?
The most logical measure of good defense would be the frequency of which opponents are putting shot attempts at the player's net while they're on the ice, which you could then whittle down to things like shots on goal, scoring chances, and goals against.
You can make an even stronger statistic if you compare how a player's shot prevention stacks up against their teammates, using relative statistics. To give even more context, you can break down the role a player is used in, using time on ice, strength of teammates, and zone start numbers. At this point, you have a fairly robust system for analyzing defensive play.
So how does P.K. stack up?
Here's the thing about P.K. Subban, before we get too far. Over the last few years, I don't think that anyone would argue that Subban isn't a better player. However, while a player can get better overall, parts of their games can still drop off, and that's the case here. In order to illustrate it, I've divided Subban's career into two sections, his pre-Therrien years and Norris year (2010-2013), and his "buy-in" years, where he very faithfully adhered to Therrien's system (2013-2015).
Using shot attempts in a "Hero chart" style breakdown of the 210 top defensemen in the NHL, here's how Subban looks in a league-wide breakdown.
The two right-most sections signify top-pairing level performance, the next two would equal second-pairing, then third pairing, then fill-in or replacement level players behind the y-axis.
While Subban's ice time has jumped, and his quality of teammates has plummeted beyond what a player of his calibre should ever have to deal with, his overall performance is up, and his offensive production is up as well. His zone starts are a bit more favourable than before, but not significantly so. If you were to just look at this chart, you'd think that the drop in defense is negligible, but shot attempts don't tell you everything. Let's take a look a bit deeper, at shots on goal.
Here you can see that the results are quite different. While shot attempts are in general still being reduced, more shots are getting through than ever before. In fact, Subban's impact on reducing shots against has gone from eighth best in the entire NHL, to 106th best, a gargantuan drop in effectiveness. Essentially, Subban has gone from an elite, first pairing defenseman defensively, to a low range second pairing defenseman.
Because of that drop, his overall shot differential has also dropped from third best in the entire NHL, to 36th best. This drop in effectiveness has led to the team giving up more goals against while he's on the ice, going from just 1.95 goals against per 60 minutes played from 2010-2013, to 2.35 goals against per 60 from 2013-2015. Just to bring that home for the non-stats-inclined readers, that means Subban today is on the ice for an average of 10 more goals per season than Subban from 2010 to 2013.
What has changed?
Just a couple of years ago, there wouldn't be much we could say conclusively about what's different about the way Subban plays now in comparison to his old self, but thanks to Christopher Boucher's microstat tracking, and similar work on zone entries and exits by Olivier Bouchard, we have some idea.
From Bouchard's tracking, we know that the success rate of dumping the puck out of your defensive zone hovers around 50%, meaning half the time the opponent recovers the puck and immediately re-enters the zone. We also know that a clean zone exit with possession, either by passing or skating it out, has a about a two thirds success rate of not just clearing the zone, but entering the opponent's zone.
Unfortunately, Boucher doesn't have as robust of a data set going all the way back to Subban's rookie year, but he does have hundreds of tracked games, and what he has seen is that the last two seasons, Subban has begun to dump the puck out of his own end at nearly double the rate he had before.
Typically, Subban's defensive zone tendencies pre-Therrien were in the range of an 80% chance to pass the puck when he had control of it, either to his defense partner, an open teammate to create a zone exit, or a stretch pass. Only about 10% of the time would Subban dump the puck out of his own zone, yet in 2013-14 he dumped the puck out 16% of the time. That dropped to 11.5% in 14-15 during the regular season, however in the playoffs that number skyrocketed to 30.3% of his puck plays in the defensive zone.
We don't know right now what the ideal breakdowns are for defensive tendencies, but we can make some logical assumptions about Subban specifically, especially when you look at previous results.
For starters, Subban is at his best when he has the puck. Whether that's to create space with his body or his skating, or whether it's to make a pass almost no one else in the league could make, the last thing you want Subban doing habitually is just dumping the puck to empty space, he isn't Douglas Murray, where that's the only option.
Yet, Michel Therrien's system demands this approach, forcing Subban to play not only inefficiently, but against his own instincts. The nature of Therrien's system is to habitually clear the puck as quickly as possible, and unless a player is given a surprising amount of time and space by their opponent, this will lead to dumping the puck out more often than not.
For an unskilled player, this approach may lead to fewer high end mistakes, but for a player like Subban, it limits his greatness to merely good, and it has to change.