My initial reaction was, Duh! Did you also know that the Internet's not just a fad?
I mean, obviously, you have to be ignoring an awful lot of evidence that self-publishing works. There are the successful newbies, and the traditionally-published authors who are selling their backlists themselves. There is even increasing evidence that if you want a traditional contract, the best way to get one may very well be to self-publish.
But I think a larger part of the problem is that this is all anecdotal evidence. There's no good data, and what data there is, will likely get worse as self-publishing accounts for an ever-bigger piece of the pie.
For example, Bowker, the company that issues ISBN numbers, found a sizeable uptick in paper book titles attributable to self-publishing. But it's actually a fairly major undertaking to produce a paper book--it's easier and cheaper to just do an e-book. I would guess that a fair percentage of self-published titles have never been released as paper books.
I would guess, because I have to. Why don't I know for sure? Because there's no data. What percentage of self-published books are released only as e-books? I don't know! And neither does anyone else!
Think about what that means: The number of paper titles could level off or even shrink as more and more self-published writers decide it's not worth their while. Bowker could release report after report noting the decline in paper titles. And that might mean absolutely nothing about either self-publishing or e-books.
But what about e-books? Well, I keep making fun of Barnes & Noble's claims to be controlling X percentage of the e-book market for the simple reason that no one--no one--knows how large the e-book market actually is. Amazon does not share that information. Perhaps they could be made to, but what about the new retail sites that keep cropping up? What about authors who sell books on their own Web sites?
Whenever someone says, The e-book market is exactly THIS big! It has grown precisely THIS much since THIS date! they are leaving indie authors out. What they are doing is counting only those e-books released by certain (larger) publishers.
Think about what that means: The more e-books are self-published, the fewer e-books will be counted. If Random House and Simon & Schuster lose e-book sales because all their writers have gone indie, the data will indicate that e-book sales have fallen, even if those newly-indie writers are selling e-books like gangbusters on their own.
There have been some efforts to generate decent data on self-publishing, including a survey that used some pretty questionable methodology, but they haven't been great. Some places, like Smashwords, like to share data, but plenty of places either don't (Amazon) or can't really be expected to in a meaningful way (author Web sites). I don't see this changing any time soon.
What does this mean for writers? Well, unfortunately it means that you really can't trust any data-based generalizations about the industry. Not even the ones from fancy-sounding analysts and consultants--they're all getting their data from the same source, which is the larger publishers. It's annoying, but there you have it--it's better to acknowledge that you don't know something than to swallow someone's snake oil because it has a bunch of impressive-looking numbers attached to it.
I would also try to avoid getting hung up on questions like, How many e-books are being sold? In addition to being impossible to know, it's kind of irrelevant. After all, the exciting thing about self-publishing e-books is that writers can make good money off sales that traditional publishers would consider laughably small.