It's extremely paraphrased and slightly exaggerated, missing some of the context (intentionally, because I don't want to break confidentiality).

It was an attempt to provide an example of why working long hours is not a good idea overall. More often than not, work done in a rush, while tired and sleep-deprived and not thinking straight, is of poor quality, and I don't think it's in our interest to lionise "heroic" performance when the end result is not acceptable. More often than not, they would have done a better job if they slept on it and picked it back up in the morning. Or, at the very least, carefully reviewed their work in the morning and fixed all the defects they introduced during the night. Ultimately, every code review needs to be held to the same high standard, and the circumstances of poor quality work should not be used as a justification for submitting or accepting bad work.

That's probably not the best way to frame it, but it's important to make people aware that they can't work 60 hours at the same quality of the first 30.

Even if they could, they'll either burnout, get promoted or leave within 24 months and then you'll need to backfill with multiple people.

Even as an evil manager, you're still better off stopping people from overworking, as you'll get more work overall from them at a sustainable pace.