But in doing so they ran Rice into the ground, which doesn’t seem particularly sustainable, hence them buying Guimarães this summer. I suspect they’ll now rotate a bit more this season, which may well be why they want to keep MLS as well.
For us, I also think it makes more sense to rotate our starting pair between league, CL and cup games, with the exact combination changing depending on the type of game. Certain combinations will make more sense for different opponents.
As for the fourth midfielder only being used rarely, that doesn’t really stack up. Sure, it applied to Arsenal and Chelsea last season - for Arsenal because Arteta didn't trust Norgaard and for Chelsea because Lavia is always out injured (although James featured in midfield quite a bit).
But for the other teams you listed the number of appearances were fairly spread out - as you’d expect with injuries, rotation, sub anppearances etc. I’ve added their league appearances from last season below:
City - Rodri (21), Bernardo Silva (38), Nico (25), Reijnders (28)
Liverpool - Gravenberch (36) Mac Allister (37) Szoboszlai (36), Jones (34)
Villa - Kamara (18), Tielemans (25), Onana (25), Barkley (21)
(You also missed McGinn who had 30 appearances)
I think appearances alone make the rotation look far more evenly spread than it actually was, because they include late substitute appearances and players covering different positions. I think starts and minutes are a much better comparison.
Liverpool are probably the clearest example. Jones made 34 league appearances, but he only started 18 (6 at right back) games and played around 1,900 league minutes. Compare that to Gravenberch, Mac Allister and Szoboszlai (played a lot at right back and attacking mid) , who were all close to or around the 3,000-minute mark. So despite making plenty of appearances, Jones was still very much the fourth-choice midfielder.
The same applies at Villa. Barkley made 21 appearances, but only started 7 league games and played roughly half the minutes of Tielemans and Onana. McGinn is relevant to Villa's squad, but he often played further forward or from a wider position, so I wouldn't class him as one of their main central midfield options in the same way.
Enzo started 35 league games and Caicedo 32, compared to Santos' 13 and Lavia's 4. Santos' 27 appearances sound a lot until you realise 14 were off the bench. There was clearly a first-choice pair.
As for Arsenal, running Rice into the ground was Arteta's choice. He had other options but clearly didn't trust them enough to rotate him. That's on him. And yes, that's probably why they've bought Bruno G, but they've effectively upgraded Norgaard. They now have Rice, Bruno, Zubimendi and Merino, plus MLS, who barely played last season. Merino was also used a lot up front before his injury.
As for City, I don't think they're the best example because Rodri's season was heavily disrupted by injury. That naturally forced others into the team, so it's difficult to use those numbers as evidence of their preferred rotation.
I'm not saying the fourth midfielder never plays. Injuries, suspensions, cup games and substitute appearances mean they'll always get minutes. My point is simply that most top teams still have a fairly clear pecking order, with one midfielder generally receiving noticeably fewer starts and minutes than the main three.
I also agree that different midfield combinations suit different opponents. But even then, once a manager settles on his strongest partnership, someone is inevitably going to play less. That's just how most top squads are built.
So going back to my original point, I wouldn't be surprised if we go with Mainoo, Tielemans and Santos as our main three, with two backs (Mount and other).