I can see why some people may come to this theory but logistically it falls apart. I am with dreamkd.
Imagine 100 people applied for round three. They sent out 10 interview invites. Let’s assume with this famous Reddit chat that somebody read, that means they would have to read thoroughly 100 applications before they issue an invitation to interview. They will have to read the essays and all the other details. That’s a lot of time.
Now, imagine they don’t use this novel approach and simply browse through some applications and quickly pick 20 that they think are interesting candidates and they spent half an hour interviewing them. In the interview they can potentially see that half of them are not interested in the program and have no clue about it. So it probably wasn’t worth reading their applications to start.
I think the reason some people may think that’s what happens is that the quality of applications in round three becomes very inconsistent in terms of quality. You get a lot of random people who are applying last minute, you get a lot of strong people who are gambling and it’s easy to see winners in a sea of lower quality apps. And with time short, Adcom has to move things quickly esp if they see someone they like hoping to snag them.
Sorry for a rambling reply, I think this has to do with a certain standard that has already been set by R1, R2, and those waitlists so it is easy perhaps to draw a line and eliminate people based on basic parameters since many likely do not measure up. Seems it really depends on the applicant pool and acceptances.