New here, and inexperienced, but I was under the impression comps were required so agents could calculate the intended audience and market appeal of the book, and, ( most importantly) their potential profits if they accept a manuscript for publication.
Yes, they use them. But how they claim (including to themselves) that they use them, and the real inner psuychological process are unrelated. They claims are pseufo-scientific business, whereas the reality is a gut thing that is - like all gut things - more wrong than random chance.
Of course, the implied connection given by the comp titles creates a perceptive connection (even when one doesn't truly exist) with the manuscript. And so the snowball rolls on.