Correct me if I'm wrong, but the RPO had nothing to do with NE giving up 143 yards rushing to the Steelers or 113 rushing to the Dolphins (in December).
I agree the pats defense was not an average defense last year. Rather, it was an above average one. But I fail to see significant improvement in the D last year from the beginning to the end. A lot of it was obviously personnel. And I agree the Pats D will be a lot better this year, especially the run D.
To say that the Pats' run-D was "weak" in 2017 is something of an understatement. It was vulnerable to all teams that had anything resembling a decent ground game and that could execute play-action effectively (e.g., the Fins and the Steelers - who also had the dual-threat, Bell, to help complicate matters).
The intent of all play-action schemes is to use the threat of a run to make it easier for the offense to be successful passing the ball - using the run to set up the pass, not a new concept.
When the threat of the run is real, and especially when the opponent's ground game is good enough to regularly produce chunk plays with the run, the edge-rushers (who are also edge-defenders against the run) are forced to hesitate for an extra beat to diagnose run-or-pass on any play that looks like it might be a run. This attenuates pass-rush pressure.
When the front six is virtually incapable of containing a decent ground game on its own (like the 2017 Pats), play-action may also have a significant effect on coverage. DBs hesitate an extra beat before getting into their coverage assignments in order to decide whether or not they need to come up in run-support. The boundary corners shouldn't really do this - they should stick to their coverage assignments regardless - but they often do so, nevertheless.
It appeared to me that MattyP attempted to compensate for the weak front six by playing the Big Nickel a lot more than just situationally (virtually every play all season long presented a "situation"), and they were in the 4-safety dime frequently (with Richards, unfortunately). The concept, I believe, was to rely on two boundary corners and a deep safety (usually Harmon) to try to contain deep-passing as much as possible, and then to use the other two safeties (McCourty and Chung) as "coverage LBs", in the box or on its periphery, in zone/pattern-matching coverage against short and intermediate routes, RBs and TEs. It appeared to me that MattyP tried Rowe in a sort of 4th-safety role a couple times (against the Steelers on a couple plays, I think), but it didn't go particularly well.
-----
SIDEBAR: I strongly believe this hesitation wrt run-support/coverage was a major contributing factor to Malcolm Butler's coverage performance decline in 2017. I believe that he was too eager to help out on run-D, trying to do too much, instead of focusing on HIS job. I believe that he was told NOT to do this on more than one occasion over the course of the season, but that it became a chronic problem anyway.
Gilmore had this same issue early on, I think, but fixed it. Rowe wasn't "hesitating" like Butler; he was simply misdiagnosing and then going all in on run-support (he wasn't doing this in the SB, though). As far as I could tell, Jon Jones never had the problem.
-----
The RPO is an extreme play-action technique/scheme. When well-executed, it can cause the DL and DBs to hesitate even longer than more common play-action. When the RPO is backed by a very good ground game, and an accurate QB, and playing against a weak run-D that requires assistance from the DBs, it's like taking candy from a baby. In the SB, the Eagles had perhaps the best OL in the league, a strong ground game with two RBs who could break runs for big chunks, and Foles played like he was snorting lines all game.
