One thing that publishers provide to readers (in the economic sense) is a brand. This is the expectation that someone has vetted the manuscript to a level of quality similar to that of other books from the same publisher. This works to the advantage of a newly published author -- the reader gets the imprimatur of the publishing house (or line of books within a publishing house) as an indicator of likely quality. Depending on the genre, it may also indicate something about the likely attributes of the book.
As an example, if you buy a book from Baen you can be reasonably sure that it is:
a) either science fiction or fantasy (with the occasional "airport novel thriller")
b) generally optimistic in overall outlook
c) focused more on story telling than on "literary merit" (as defined by the "litrachoor elite." This doesn't
exclude literary merit by any means, but it does say that strong-literature-with-weak-story won't be coming from them.
d) not a nihilistic plotless mood-piece full of allusion and metaphor, bemoaning the pointlessness of existance (from the PoV of an SF-lover looking for entertainment), or
d-alternate) full of exploding space-ships (from the PoV of a member of the afore-mentioned 'lit-crit' crowd.
I've spent some time as a volunteer reading through the slush-pile at Baen. You wouldn't
believe how terrible the average submission is. I sometimes think that my (deceased) dog could do better. From beyond the grave, no less! Time spent with the slush-pile is almost guaranteed to induce the need for mental-floss.
So, my challenge to those who think that publishers aren't needed in the electronic world is this:
How is a reader to discover books they might like to read? And how might an author bring their book to the attention of readers who may like it? And who is it who'll work with the author to get their manuscript from "almost good enough" to "knock-'em-dead wonderful?"