But even if we assume that an advance was necessary - A couple of years of no starvation wages ought to be negligible compared to the sort of amounts we're talking about for a successful book. You don't pay fixed amounts back in terms of percentage of take - that's foolish. At worst you might pay it back with a risk premium, like a bond.

Publishers would presumably have you believe that they're providing a service, at risk to themselves, and that risk is sufficient to justify their take. That they have to make as much money as they can on the few successful books to pay for the masses of dross that they fund. But this just makes it an exceptionally bad deal for people who are consistently good. If you have a track record, then you should be in a different risk pool.

The questions are really ones of risk management and information asymmetry: how many successful books are you going to get and how well can you predict your risk? Does the author even need an advance? Are the editors even honestly trying to predict the success of a book or are they just going with their gut or their political bias or whatever? It's not a very competitive market after all.

If people had the option to not take an advance, and to take a much larger cut of the sales, I think many would.