I'd like to briefly contrast two self-published books I have reviewed, one of which went down very well, and the other that didn't to show what is needed for a self-pub to succeed in this field.
The high flyer was a little number entitled The Rocketbelt Caper. We are happy to cover technology if it's interesting - and this was interesting in spades. It had a topic that anyone brought up on Tomorrow's World and James Bond would love - the development of the jetpack or rocketbelt - it was professionally told, and kept the reader interested throughout. It was no surprise to me that this book was picked up by a mainstream publisher, so is no longer self-published... but it was when we first reviewed it.
Less successful was the recently received Unraveling the Universe's Mysteries. This couldn't be faulted on the professionalism of its publicity machine, but left something to be desired as a book. I had let it through the sieve because the author had an appropriate background (physics degree/worked for IBM) and seemed from the blurb to be going for a straightforward explanation of science rather than pushing his own theories, but it wasn't a great read. It wasn't well edited (string theory, for example, became both sting theory and spring theory in the space of a couple of pages), the author didn't have that professional writing touch and there was at least one hiccup with the science. Now don't get me wrong, practically every popular science book I've read (including mine) has had more than one error - but it's a matter of proportion.
However the thing I was most uncomfortable with, and hence probably shouldn't have accepted the book for review in the first place, is that after a fair amount on current theories, the author started to give us some theories of his own. This is something that is usually an instant turn-off in self published books. That might seem unfair, but if, say, a leading theoretical physicist at a well-known university comes out with a speculative theory they have two advantages. First they have the knowledge to come up with a theory that could be meaningful, and secondly other people with similar expertise will have looked at that theory and commented on it. And both of those are necessary.
The fact is I'm not a physicist. I am a science writer with a rusty physics degree - not at all the same thing. And that means I really can't judge if a theory has any merit (I can tell if some are total fruit-loopery, but that's a different thing). I therefore feel uncomfortable recommending a self-published book by someone who isn't an academic working in the subject (the author of Unraveling isn't either) that comes up with original theories. The ideas in this book are certainly not totally off the wall - but they still make me feel uncomfortable.
So there we have it. It might seem unfair, but if it's self-published it mostly doesn't make the grade, and one real turn-off is going to be if the author has his or her own little theory...