Disrupting editors does not mean eliminating them. Successful self-publishers pay for, among other things, their own editors. Readers are not stupid. Between descriptions, reviews, free samples and so on assessing quality (both abstractly and personally) isn't hard.
Amazon doesn't want to push prices down for the sake of pushing prices down. They want to push prices to where their data says they'll make more money (and, assuming they get a cut of the total, to where everyone will make more). People make a lot out of Amazon not paying 70% for books priced above $9.99. Far fewer notice that they also don't pay 70% for books priced below $2.99.
Amazon wants to push prices down for the sake of their own bottom line. What happens when what's good for their bottom line becomes even more divergent from that of the authors working with them? In a monopsony, where do they go?
Which are all very small relative to the size of the Amazon market. You can't practically compete and not sell through Amazon, hence the problem with a monopsony.
You don't even need to not sell through Amazon to compete with them, just be willing to take a smaller cut with competitors that give you a more favorable terms (allowing them to undercut Amazon).
Three publishing houses offered term sheets to me. Not one offered a decent editing package. And even then, paying $10,000 out of pocket was cheaper than absorbing their ludicrous royalty costs, ignoring all the other externalities that come with going with a publisher...
> an editor is really helpful. Revision, pagesetting, etc, as well.
At what point could this function be largely automated? At what point fully?
Algorithmic editing brings up novel possibilities. You could have footnotes added as information is corroborated or countered. Editions could be instantly localised for almost every language and cultures. One may even get to on-the-go re-editing for individual readers (for example, I may prefer a punchier writing style with non-referential footnotes inserted in-line). This changes what a "book" is, but so did paperback publishing and the Kindle.
>At what point could this function be largely automated? At what point fully?
Once we have strong AI capable of analyzing semantic content, determining which passages most strongly support the intended message, determining which unwritten passages could be written to more strongly support the intended message, determining the semantic linkings between different sections as the message is built up and optimizing the order of those sections.
I have the impression the Kindle is heading in that direction. For example, sections other people have highlighted show up, practically as suggestions to do it yourself.
For some reason this creeps me out, as reading is quite a personal thing, but it seems the attempt to turn it social and crowdsource data from it is happening.
Editors
Having a "everybody can publish anything" model is great. But it doesn't fit all cases.
Yes, an editor is really helpful. Revision, pagesetting, etc, as well.