True, but Santana also threw 30 more innings than in 2016, with 5 complete games and 3 shutouts. He carried a big portion of the load. If not for his impact the staff would've ranked near the bottom of the league again.
I agree with your point that without Ervin, they probably would have been near the bottom in 2017. My point is that the improvement from 30th to 23rd didn't primarily come from Ervin pitching better in 2017. It came from the improvement in other places, primarily from two spots in particular.
Ervin was huge last year. But in 2016, the starters overall ERA was 5.39. If you take out Ervin's 2016 starts and replace them with his 2017 starts (including the 30 extra innings), the starters ERA would have only dropped by 0.10 to 5.29, and they still would have been close to last.
However, going from 30th to 23rd was from the starters ERA dropping from 5.39 in 2016 to 4.73 in 2017, a net drop of 0.66. So my point was that if Ervin's improvement dropped the ERA by about 0.10, the remaining 0.56 came from improvements elsewhere.
So looking elsewhere, 2016 Gibson vs. 2017 Gibson was virtually a wash, as was 2016 Nolasco vs. 2017 Colon (really, I was surprised too!). 2017 Santiago was a tiny bit better than 2016 Millone and 2017 Hughes was a smidge better than 2016 Hughes, in each case in similar numbers of starts.
I didn't go to the bother of totaling the riffraff of small-volume starters, other than to note that the leash was short on Turley, et al., in 2017, both in numbers of starts individuals got and in how long they lasted in a game. By contrast 2016 gave us 9 starts of Pat Dean (6.85).
If those were mostly a wash, where did the big improvement come from? Ultimately, it wasn't primarily from Ervin's slight improvement in 30 additional innings. Rather, it was from having...
- 25 starts from Berrios (3.89) and 21 from Mejia (4.50) in 2017 (46 starts total)
- vs. 26 starts from Duffey (6.43) and 14 starts from Berrios (8.02) in 2016 (40 starts total)