His early work was very interesting, drawing attention to the fact that sometimes what "looks good" or "works well" according to someone's intuition doesn't actually produce the best measured results. He was a pioneer, at least in Web terms, of promoting the kind of data-based decision making that many successful Internet businesses take for granted today. Anyone who designs web sites but hasn't read early Nielsen is missing out, IMHO.
However, as time passed, his articles became increasingly self-referential and self-promotional, and seemed to include less and less substantial, original content. By his own arguments his material was becoming worse.
His old site famously wasn't a shining beacon of graphic design, and he always seemed rather proud of that -- ironically, really, with the body of research we now have that shows how much influence a good visual design can have on readability. With the redesign used since the move to the NN Group site, he seems to have completely jumped the shark. Even famous Nielsen material on the F-shaped reading pattern and the use of underlining for hyperlinks seems to have been discarded. The text is set in a small size and with a font stack that will render unpleasantly on a significant number of visitors' computers (because Helvetica is often installed on Windows PCs, but only as a cheap and nasty screen version of a printer font). And what happened to not making real content look like an advert?!
It really is a shame, because every now and then he does still write a great article. I wonder what the Nielsen of 15 years ago would make of his site today.