He is referring to the range above the transition frequency. There is more context in the book as this quote is taken from a full two or so chapters on the topic. It is just a snippet.
If you have not yet read it, read that book. It is not the only good source of information out there nor is it the only smart approach but it sure is a worthwhile book to read if you really want to understand the science that is available and not the marketing.
Toole and Olive and friends conducted many types of blind & double blind tests which almost nobody else has done(at least publicly, manufactures should promote and publish their research as way of confirming their marketing claims)
They also did a lot of study of other previous testing done by other groups.
It is pretty hard to argue with his claims in my opinion.
Really much of this is discussed through out ASR but it is hard to assemble and there is a lot of noise. I think starting with the book is best and then diving in deeper after that, not before. Totally fine to disagree with him but hard to disagree authentically without fully knowing what he has to say.
I knew, and had read those parts in his books. My comments stand, not that I disagree with anything they said, or you said, but I just don't know enough about how factual that is, as supported by quantifiable measurements or counter measurements (if there is such a thing lol..) If I remember right, for me there were two (at least) takeaway:
1) He, or they in Dr. Olive's included.., never actually said RC/REQ should not be done in the above Schroeder frequency range, but did say at those higher frequencies, the way those software collect the room information with the kind of mics used and their placements would not be reliable and could therefore result in corrections that may not work well. Not in such words, just my own takeaway from reading it, and listening to him if it was on a video too. Cautionary note: Again, just going by memory now, so if I may be misinterpreting something he said unintentionally.
2) He said those things a few years ago and it did not appear they had followed up on even more years ago when those studies were done, at the time most likely before Audyssey XT32, the very early versions of Dirac, Anthem ARC, let alone RoomPerfect and Trinnov's gaining more popularity, and the much improve UI, such as Audyssey's apps, ARC Genesis, and Dirac's much improved UI, and DLBC etc.
So I believe it is time Harman follow up on this topic of "automatic" RC/REQ based on the theoretically much improved versions of those products, and not fixated on the concept that manual ways such as using PEQ to focus on improving in room respond based on the speaker's anechoic measurements. That's all I am saying, basically, being open minded about possible technological changes (hopefully for the better) over the year.