FWIW, I re-ran the RSCI without Rivals. The Duke related players rankings change to:
Cam Boozer #3 (with Rivals also #3)
Nik Khamenia #15 (w/ Rivals #16)
Cayden Boozer #20 (w/ Rivals #19)
Sebastian Wilkins #40 (w/ Rivals #48)
Dame Sarr #67 (w/ Rivals #81, but meaningless either way)
Shelton Henderson #27 (w/ Rivals #22)
So I guess it's not that big a difference, but it is a difference and I still I think it's a mistake to include Rivals.
Both 247 and On3 also publish "composite" rankings, that each seem to basically do the same thing as RSCI -- combine the rankings from 247 (in-house), On 3 (in-house), ESPN and Rivals.
Each claims that it does so through an undescribed "proprietary algorithm":
247: "
The 247Sports Composite is a proprietary algorithm that compiles rankings and ratings listed in the public domain by the major media recruiting services….”
https://247sports.com/season/2026-basketball/compositerecruitrankings/?InstitutionGroup=highschool
[Note: while the 247 site doesn't identify which "major media recruiting services" it is referring to, and Wikipedia says it is 247, ESPN and Rivals, according to this Reddit thread, after On3 started in 2021, 247 added On3's rankings to its "composite" -- while I can't find documentary evidence of that, logically it makes sense.
On3: "
The On3 Industry Ranking is a proprietary algorithm that compiles ratings and rankings from all four primary recruiting media services [On3, Rivals, 247, ESPN].”
https://www.on3.com/news/on3-industry-ratings-rankings/
Neither discloses much about their purported proprietary algorithms, but this piece from Georgia's 247 site in 2020 does offer some further explanation, namely that 247's composite: (i) "converts average industry ranks and ratings into a linear composite index capping at 1.0000," (ii) gives "all major media services ... an equal percentage in the Composite Rating," and (iii) "equally weights this percentage
among all services that participate in a ranking for that specific prospect." https://247sports.com/college/georg...t-247sports-rating-process-143324123/#1365356
Wikipedia also claims (again, I didn't see this as actually supported by the cite Wikipedia linked to but it strikes me as reasonable) that 247 "does not include [in its Composite] any athletes that do not yet have a recruiting profile on 247Sports in order to prevent spoofs from other sites."
So, at least 247's methodology may weed out some of the ways in which RSCI ends up being obviously "wrong." For instance, last year, 247's in-house ranking of Maluach was #4, and On 3's was #9, but ESPN and Rivals didn't rank him b/c he was an International. RSCI had him as #51 because it counted the non-ratings by ESPN and Rivals as 0, whereas the 247 Composite had him as #6.
The Reddit thread suggests that one other factor contributing to differences between the 247 and On3 Composites was due to how they converted Rivals ratings to the 1.000 scale, but there's no way exactly to know how they do that.
On Kedsey's point about inclusion of Rivals ratings, I would suspect that, with the recent acquisition of Rivals by On3 and the announced restructuring of the On3 business, soon enough both On3 and 247 will phase out inclusion of Rivals ratings and that going forward each of them (and RSCI) will just be averages (obscured in some form of purported "proprietary algorithms") of the three of ESPN, 247 and On3/Rivals.