Some things in Linux have gotten better in the interim.

. . . and then there's GNOME.

That's not a very good point. Sure, it's true, but it has nothing to do with what they originally said.

There's a difference between making a statement about the odds of something being true and claiming something is always true. I do believe it's more likely that the father is religious than not. This can be true, and yet not imply anything about the set of all religious people (as indeed it does not).

I'd never get used to trying to figure out how to dress if it was 20C out.....

Yes, you would. I decided one day I'd use Celsius all the time, even though I live in California. My goal was to try to get an intuitive feel for what different temperatures mean, without mentally converting. It took me a little while, but it didn't really take any effort.

I'm not saying you need to switch, but it's kind of fun, and I'm sure you're capable of it.

Actually, the opposite is more likely to be true. Pop/rock is often more "difficult" to encode, as MP3 performs worse when you have greater spectral density. See e.g. Subjective Evaluation of MP3 Compression for Different Musical Genres (AES Oct 2009).

You can't actually reproduce a sine at 16kHz with a 32kHz sampling rate. A wave that looks like [ -1 1 -1 1 ... ] is either incorrectly sampled or infinite in length.

Anyway, you need a transition band, so in practice you shouldn't count on properly representing frequencies above 20kHz with a 44.1kHz sampling rate. With 32kHz, you're looking at something probably close to 14kHz.

You're conflating dynamic range compression with perceptual coding.

You have a fundamental misunderstanding of how MP3 works. You can certainly degrade something further that already sounds terrible. MP3 doesn't "squish" audio. In fact, MP3 degrades audio like this *more* than it would degrade non-squished audio, because music that's "compressed-to-shit" almost certainly has plenty of clipping, greater spectral density, and so requires more bits to preserve.

I assure you, even if something already sounds terrible, it still can sound worse, and MP3 at 128kbit is a great tool to make it that way.

Good post. To be fair, your post discusses an idealization of perceptual coding. In practice, old formats with known problems are in common use, and both old and new formats are often used at bitrates where a significant portion of noise added is above masking thresholds.

People also listen in sub-optimal environments and on (very) colored speakers or with lots of EQ, which makes artifacting more noticeable by defeating the assumptions of the psychoacoustic models in question.

This doesn't change your point that we don't need Pono, of course.