There's a difference between "recruiting rankings are a perfect indication of the future success of players" and "recruiting rankings are a directional indication of the relative strength of teams." ESPN has likely found that there is some predictive power in these rankings. This makes intuitive sense to me. When we sign a recruit out of high school all of us immediately want to know how many stars. And our excitement level is often correlated with what the ranking services are saying.

The model can give partial weight to recruiting rankings and get some predictive power from it without relying on it too heavily. At the beginning of the season if I showed you a previously unknown football team and you to predict their future record, would you not want to know about their recruits?

Also important to note that there's just not a lot of numerical measures of incoming high school players that are reliable other than recruiting services. High school stats are so unreliable and location/context dependent. The recruiting services quantify in a numerical way a whole bunch of information that just wouldn't be available otherwise.