My colleague art medlar sent along an interesting point on the way PeerJ is accomplishing its disruptive pricing.
ETA: I am editing this article to correct errors of fact. As always, any mistakes here are my own fault.
In a word, they're socializing the cost. The price PeerJ charges you is, in effect, a fee to submit. There's no guarantee that your article will be published, only that by paying the $99 fee you'll get it reviewed.
ETA: Peter Binfield, the publisher of PeerJ and Jayson Hoyt, the CEO, have written comments you can read below kindly letting me know that I've misunderstood the model. There is indeed a $99 up-front option, but it's not required. If you choose not to pay that and do get published you pay a $129 cost.
This compares with a PLOS publication like PLOS1 that might not take your article for review but if it does so it pays the cost at that time for you and only on publication do you pay your $1350.
Well, a little math will show you that if they publish 1 out of every 14 submitted articles then PeerJ has a better revenue model. The publication rate for submitted articles at most high-quality journals is probably less than that. Many journals don't publicize these rates (presumably to avoid discouraging potential authors) but a rate of 1 in 20 or 1 in 25 would not surprise me. In fact, one of the things that makes certain journals so prestigious is that they are hard to get into. Often an author will find that a very good paper isn't published not because of anything to do with the paper but just because the journal only has resources (pages, editors, reviewers) to handle a fixed number of submissions. A common academic strategy is to submit a potential publication first to the best-possible venue for it and if it's not accepted there then turn around and re-submit it to other places.
ETA: Binfield wrote in his comment that they "...expect to end up accepting ~70% of all submissions". That's a fantastically high rate and assumes a level of quality in submitted papers that was not present when I was doing academic peer reviewing. It will be interesting to see what the numbers end up being.
The review-reject-resubmit cycle can lead to long delays for publication, which in turn relates to another pointer art sent: a notice in Nature's blog about the White House's incremental move toward open access. The new policy is that "taxpayer-funded research should be made free to read after a year’s delay" which seems like a lot on the face of it but is still shorter than the often multi-year delays involved with pay-subscription journals.
It's progress. Slow, but it's progress.