I think some of it is whether you're used to listening to the type of music. All those samples are types of music I don't normally listen to and thus scored 0%. They're also mostly low-information songs that are vocal-heavy. Vocal-heavy music is very easy to compress compared to say something like electronic or EDM type music which will easily create distortion with lower bitrates.

I got 2/6 right, the 4 wrong were 320kbps. Using the cheap-but-decent shp9500s and no dac, but quite good hearing. I'd chalk it up to having been listening to almost exclusively streamed 320kbps, it was interesting to note that the 4 I got wrong were because I discarded the more 'neutral' toned one which was actually lossless.

The 128kbps was of course very easy to pick out.

I'm of the opinion that 320 vs lossless is meaningless enough to not really bother, but I know for some tracks it was trivial for me to tell the different. Mostly it was the very complex/atmospheric tracks, Mind In A Box being a good example, where 320 lost a notable amount of quality. Probably others like Ne Obliviscaris as well.

The biggest problem with Spotify is that not all of their library is available at 320kbps.

I got 3/6 right using the iPhone XR built-in speakers (other three were 320k). I then tried it again with someone else clicking the buttons (in case they weren’t randomized order) using my DAC/amp and headphones and got 5/6.

The thing that gives it away for me is the extreme highs and lows, both are clipped when compressed usually. I should note, I wasn’t picking which “sounded better”, but rather which “sounded uncompressed”. Several of the tracks sounded better at 320k than they did uncompressed to me. Despite being rather close to 40, I am still able to easily hear sounds well over 20kHz.

I'd say it's nearly impossible to compare that way, because the volume hast to be precisely the same (louder always sounds better).