For listening, anything higher than 16 bit at 48kHz simply does not make sense and will never make sense, for reasons stated in the article.
24 bit could make sense for archival. Higher sample rates can make sense when processing some effects (it is immediately downsampled again afterwards).
I'm sure the other statement was made with the same confidence. Downloading doesn't have to be for listening today. In the future we may not have the same hearing limitations.
In short the overconfidence is a result of not imagining a future that's different than today.
Or, without needing a far flung sci-fi future, the concrete impact of frequencies under 50hz on frequencies above 50hz. Particularly as it applies to the resonance of a listening environment. Maybe you won’t notice in your headphones, but on a PA system in a physical space, e that indoor or outdoor, there’s a measurable and discernible change to the audible frequencies based on the existence or non existence of the inaudible ones. The same applies on the high end, but it’s less practical of an example, and I’m not sure half of people would care to listen long enough to notice.
This is a slightly broke take because your ears do not need to hear a particular frequency for it to impact what you do hear. Harmonic resonance creates overtones which can be audible through modifying those frequencies which are heard.
There is a website which AB tests different bitrate audio. I can’t remember it. Not a humble brag, but even on shit consumer headphones I got a statistically significant amount correct (~80%). So fornsome people, yes it does make a difference.
But unless you are hearing side by side, I very much doubt you’d care. Even my audiophile friend I tricked by having the TV speakers (one of those OLED “screen as a speaker” ones) turned on with my surround off, and he thought it was amazing.
For listening by people, they don't.
You'd really do not want that much accuracy in your hearing, it would drive you totally mad.