Take a look at the comments on the blog post.
From Scott (the author of the blog post):
NER: All I claim is that empirically, it’s a very successful heuristic.
As for why this is so, one might speculate as follows: those who can’t be bothered to learn the tools, notation, conventions, etc. of whatever subject they’re writing about, are highly correlated (though not identical) with those who can’t be bothered to learn the subject itself.
From a comment by asdf:
What can I say; if a paper is full of spelling errors, that doesn’t necessarily mean the mathematical content is wrong, but it’s not a good sign. TeX is part of mathematical culture and using anything else might be similar to using an unconventional notation for a familiar math concept. Again, not conclusive, but not a good sign. Nothing short of actually reading a paper (or at least reading enough to find a definite error) is enough to form a conclusive judgement of its content. But lots of things, like the presence or absence of spelling errors and weird notation and typography, can raise or lower expectations at the outset. That I think is what Scott was trying to get at.
From an anonymous comment:
Nobody’s saying the use of TeX is a fully reliable test for whether a paper is good, but here’s an analogy:
Suppose someone handwrites a paper in crayon. Logically, this tells us nothing about the content of the paper, but in practice, it tells us a lot about the author. Very few people who write in crayon have anything interesting or worthwhile to say. It’s not a logical guarantee, but if you see a research paper written in crayon, it’s perfectly reasonable not to take it seriously (since it almost certainly isn’t serious).
Not using TeX in a math/CS paper is not nearly as extreme as writing in crayon, but it is similar in spirit. There are a handful of older or eccentric researchers who have never learned TeX. Other than that, anybody in math or theoretical CS who doesn’t use TeX looks like a rube. This may not be fair, but it’s a true statement about appearances within the community.