First, at my R1 math dept in the US, we no longer pretend to linearly rank grad applicants. That's ridiculous and meaningless.
I'd wager that the faculty person who made such remarks has some sort of resentment of grudge about some demographic of students... and tells his/her own students that [some demographic] of students is inferior to them...
(I'll not go into details about my observations of such behaviors... tho' one stereotypical riff is non-US-origin faculty to disparage lazy americans...)
In any case, although some of my colleagues here do definitely like algorithms and numerical rankings, all my (multi-decade) experience indicates that these numbers are not merely not good indicators, but just worthless. They test many things, but not what we want for a PhD program in mathematics, of whatever sort.
So, yes, there are people who have poor enough judgement to say such things, but it's basically disconnected from reality. Still, even deluded people with enough power can make their delusions be operationally real, so be alert... sigh...
EDIT: among other things, in my R1 US math dept, by now we do not pretend to linearly order applicants. It's more like a first-pass that tries to assess whether people would succeed in our program, or not. Obvs don't want to admit people (for their benefit and ours) who are not at all prepared, or maybe don't have the interest or ability. But among those who could meet our expectations ... how to "rank". There's really no sane system. GRE subject test scores have shown their own irrelevance, ... GPA? Whah? Predicting what 22-year olds may accomplish based on the previous few years of their life is ridiculous.
In my 35-year involvement with grad admissions, I find that the only reliable assessment method requires "reading between the lines" in personal statements and letters of recommendation. How in the world could I describe this in algorithmic terms? (Ok, I could make up something, especially if litigated-against, but, c'mon...)
Yes, I am all too well aware that many programs do behave as though some algorithm is sufficient. In fact, our program plays against that by looking at candidates whose "typical algorithmic" status would be not-so-high, but whom we can anticipate would be great students. :)