There are important papers and then there are nicely written important papers. For the field of AI, ML and DL which papers are of the second kind?

Important works that are terribly written have to be filtered by a textbook author to take them to the masses. A nicely written paper doesn't require that.