I read the Myers-Briggs link down to "we'd expect ... bimodal", i.e. "I'm criticizing a binary version of Myers-Briggs that doesn't match the continuously-scored tests roystgnr took decades ago", and then successfully predicted what most of the subsequent criticisms would be and why they would have been similarly inapplicable.

I had a similar reaction to that argument. Overall the article counts strongly against the credibility of 80,000 hours. I don't especially advocate Myers-Briggs, but the reasoning in this particular essay is terrible.