To me, the unmixed version evokes feelings of attending a live concert, whereas the mixed version sounds like a generic radio song.

I also find the unmixed version more interesting. You can hear the singer’s naked voice and really savour every little nuance, with the occasional saxophone accent and silent passage where the drum+bass is audible.
In the mixed version, attention is completely robbed from the singer’s voice, because everything is there all the time. The saxophone accents just as much have to helplessly blurt atop the ever-present bass line.

In general, there is no dynamic range in the mixed version. The whole composition is always around 95% of max volume. There is no silent passage, which could help the following louder passage stand out.
But it’s also each individual instrument. For example, the saxophone player puts in a lot of effort, playing more pronounced during their solos, which is practically inaudible in the mixed version, because the minimum volume of the saxophones got pushed up massively.

Well, that’s all my opinions on why I like the unmixed version a lot more. Certainly not perfect, it could use a tiny amount of mixing, but yeah, that particular mixed version I like less.

As for why mixing is done like that, I’ve usually heard it explained as that people generally don’t actively listen to music anymore.
To hear all these nuances, you can’t have much background noise, like you might have on your commute.
And even if you can hear the nuances, you wouldn’t be able to really appreciate them, if you’re currently distracted doing sports, dancing, partying or even working.

Well, and yeah, some hardware just falls off a cliff when the music goes silent, too. The dynamic range is a lot smaller in that moment, so for properly portraying all the details happening, you actually need rather precise hardware.