In its original form it was simply an aid to get static html from human readable text (<- that there is the basis for md).
For better or worse, it's been forked a few times, adding new abilities or even in efforts to make it stricter. The main idea of readability still applies. I could hand over a markdown formatted txt file to someone and they'd probably read it just fine, whether they could code or not. Html in plain-text form, though? Not sure that would work.
It has quirks, but the whole I really like it. I've used it for writing guides, taking notes... Maybe even for web content at some point.