That's not what he's saying though. Points per game is a bad metric because it's often used to measure an individual unit when it is far more complicated. Is your defense **** if your team gives up a ton of points because your QB is an INT machine and you constantly are giving up the ball and then your special teams sucks and gives up some horrible field position and even some run backs for a score? Is your defense good if your offense controls the ball for 75% of the game and gets out to a hot start and the opposing offense has to abandon their game plan to play catch up? Because points per game will say the answer is yes.
Points per game is based off total team points, if we're breaking down an individual unit I prefer
"per drive" which isolates offense or defense in a vacuum. But you're acting like we don't know how good Brady and the offense were for the 21 years the graph represents... we do. You act like we don't know how good the Patriots special teams were for two decades, we do... they were the best example of special team's excellence for the better part of two decades.
We can also hold team points allowed up against points allowed per drive to get a better representation of how much we can attribute to the defense alone. In this case the Patriots the offense was consistently great, and the points allowed defensively and points allowed per drive were similar... which means the defense was consistently good/great also.
Points allowed at a league leading pace among his peers shows team competency. It shows that in relation to his peers Tom consistently had better teams around him over the course of his career and it's not debatable... it's a 21 year sample size. And yes, some years the defense wasn't as strong as others, or the special teams in other years... it doesn't change the fact his teams were better on average.
For example points per game for Super Bowl 51 will say the defense was the weakest unit in the game for giving up 28 points even though they realistically had nothing to do with the pick six Brady threw, and then there was the one Blount fumble in bad field position that led to another score. In all truth, they gave up one bad score that the offense didn't contribute to in each half. But if you look at points per game it will make it look like the offense bailed out a poor defensive performance.
A one game sample size is laughable in relation to a 21 year sample size.
We can also look at points per drive if we want to isolate defense alone, we can look at simple stats from the box score.
DVOA is straight up beter if you want to purely measure a unit. Points per game is more of a total team metric that you really have to analyze and find the "whys" to get real data out of it.
DVOA is made up nonsense that introduces human error, personal bias and has been constantly changing since football outsiders created it. It's junk science, the conspiracy theorist in me says it was likely concocted by Vegas to convince a bunch of rubes that bad is good and vice versa so they could steal your money.
If you want to measure an individual unit like defense; points per drive, forced turnovers, 3rd/4th down percentage, red zone percentage are much better, more accurate stats that even a 12 year old can understand. DVOA is junk science, it's better to wipe my butt with.