Then it makes absolutely no sense and the data is flawed considerably by assumptions. Based on this data, it sounds like you should just bench all rookies in year 1 and throw them into the fire in year 2
They note exceptions, such as Khalil Mack who had no jump between years 1 and 2, but yes in general you do not want rookie starters. That would be an accurate statement or if you are going to have rookie starters be prepared for them to be the weak link. It is not as though throwing them into the fire year 2 though has any negative consequences though. We have multitudes of examples of players who thrive after having their rookie years be spent getting coached up.
Easiest example of this is QBs (the position where you would think live reps would be most important). The best QB of the past 20 years, Rodgers, sat for 3 years and did fine his rookie. The QB most likely to take that title from him, Mahomes, won the MVP in his first year starting after a year learning. History suggests you do not need to force reps upon a player for them to learn in the NFL. They just need time to be coached up and generally that coaching takes about a year and sometimes 2, but it rarely is immediate.
I wish the numbers did not suggest that, but reality does not change just because I wish it were so.