That was the deceptive thing about residency, especially 3rd year. You'd bust your ass, send a half dozen patients to the unit, not have time to document, eat, or pee, and felt like you were seeing 3 patients/hr. Then you'd actually count them up and realize that you were seeing ~1.4 pts/hr. You'd just had the same 8 pts that came in within a half hour of each other for the last 6 hrs. You'd maybe move a fast-track patient through quickly, but they'd be replaced by someone with another 6 hour work-up. Add time waiting for your trauma and surgical consults to make their way up the chain of command (they usually sent someone to see the patient quickly, but actually getting a decision took hours in the not critically ill), and most of us probably never had a month in residency were we averaged 2 pts/hr.
I agree with you that this isn't necessarily a 3 vs. 4 thing, but it's something that most programs don't have time or the practice environment to teach. I'd say about a quarter of my attendings wouldn't hack it outside of academics, but that was mainly due to most of them pulling some absolutely brutal shifts (25 in the lobby and being single coverage on overnights being not uncommon) at our two community shops.
42 months sounds about right in my mind as an optimum training length, with mandatory moonlighting the last 6 months. Also, the RRC should mandate some kind of metric/pt sat education in the last year of residency. I know I made it through residency having exactly 1 month were I had an accurate count of how many patients I saw. I never saw any info on LOS, door-to-doc, or patient sat scores despite these being the numbers that determine whether I get a bonus (or get to stay working at my current location).