A lot of people view Philip K Dick as a hack writer; and yes, there is a lot of merit to the argument. NF himself has been quoted a few times saying something to this effect. Mostly, the type of people who hold this view take writing as a craft very seriously and on a lower, sentence-by-sentence level, PKD is very hackish - no doubt about it. (A wonderful, perhaps apocryphal incident involves someone looking at a remarkably low quality pulp fiction journal and wondering who'd ever read it, to which PKD retorted that he was one of those who wrote it.) In fact, it's a common complaint against science fiction and fantasy writing in general, held presumably by those in the literary fiction camp, who are very self conscious about what they're writing - sometimes so much so that they go to great lengths to make their prose rather ugly and/or purposely un-self conscious.
However, there are times when PKD breaks through his self limitations of uninspiring writing, vapid dialogue and plot-twists-of-varying-degrees-of-efficacy. One may wonder how much of it is a mere law of large numbers (he wrote 44 novels - at least once every year) and how much is truly great writing.
As someone who's read more than his fair share of PKD, NF was taken aback at rereading a classic - The Man in the High Castle - an alternate history so well realized that it's hard to believe that it's by the same writer who wrote The Man Who Japed or Galactic Pot-Healer or Game Players of Titan or several other really forgettable works, whose plotlines NF has dutifully forgotten, with only murky, vague, not-so-agreeable remnant impressions.
However, when PKD does manage to make magic happen, the impact is enormous. NF has powerful memories of reading Do Androids Dream of Electric Sheep?, Ubik and A Scanner Darkly - incredibly high quality fiction - all of which NF enjoyed thoroughly. He's more confused about his later writing - The Three Stigmata of Palmer Eldritch and The Transmigration of Timothy Archer - he didn't enjoy them so much (the latter particularly), despite their undeniable literary merit.
All this is also curious because it shows that a description on a lower level (sentence-by-sentence level hack writing, which PKD is guilty of) can be totally overturned or may be wholly inadequate to describe the same system at a higher level (the big, complex philosophical themes that PKD often writes about). Indeed PKD's oeuvre, when viewed from this higher vantage point has much more merit than a lower level description based on mere literary qualities will suggest.
This "phase change", so to say, is common in physics and is by now a firmly established part of modern science where it goes by the name of emergence. Lower level descriptions don't segue continuously into higher level descriptions. That's why knowing physics doesn't automatically make you a good biologist. That's why phenomenology is not reducible to morphology. That's why David Foster Wallace said "Fiction's about what it is to be a fucking human being."
NF would like to see some emergent literary criticism take shape, where the critic pays close attention to which level in the overall hierarchy the writer is being critiqued at. Perhaps an $n$-tuple of scores may be assigned to each writer, ranging from lower to higher levels. PKD will get really low scores for lower levels but high scores at higher levels to compensate for mere bad writing. A similar fate awaits Dostoyevsky, though he will outscore PKD at all levels (a uniformly better writer in a mathematical sense). This however, is not the same as multidimensional scoring, though the operation will be mathematically similar.
Will this ever be a part of mainstream literary criticism? Will you want lit crit to be done this way? Is this a regressive way to look at literature? Is it even desirable? Won't the ghost of Robin Williams haunt your dreams yelling Carpe fuckin' Diem?
Not sure, though PKD could definitely tell an engaging story where it was being done all along until some misfit stumbled upon this alternate reality on the very last page.