I have some substantial issues with this study, primarily related to the fact that he almost completely ignores the impact of injuries on his findings. I'd have to do a similar scale analysis to find the average number of major injuries (n starts missed, say possibly 10, as quantified by the average number of starts by the top 2 pitchers between the injured pitchers' starts, for example) to pitchers over time, but I'd imagine that for a long time it's been at least more than 1/2 per team per season. Given that, the basic conclusions he keeps coming back to - that teams don't want to use their 4th and 5th starters as heavily, and 3rd starters get less work - are almost entirely invalid. The guys coming up as #4, 5, etc. in his analysis aren't necessarily the last guy the team wanted to turn to. Many of them were just injured for big chunks of seasons. It's not like Clayton Kershaw was the Dodgers 3rd choice pitcher last season and they didn't want him to pitch as often as their top 2 guys. He starts to address this kind of factor in his days/games of rest analysis, but he shows far too little of the actual data and far too much of his "largest numbers of days of rest" type stuff for me to find this analysis particularly compelling.