Jeremy Snippet has contributed a lengthy comment on my post of 2 December 2005, which was written in response to an earlier comment of his. As 2 December is now far back in the mists of time, blogwise, you are unlikely to see this comment unless I point it out, so that I now do.
Jeremy (a pseudonym) is evidently a person well connected in the media/book world and he gives some real-life examples of the advances paid by publishers to real-life writers. These advances he describes as fees, a terminology which I think is appropriate. He also maintains that 'known names', or favoured parties, get paid a hell of a lot more than unknowns from the sticks. I wouldn't disagree with that either.
Where I do differ from Jeremy slightly is in this respect: I do not believe that publishers wholly disregard sales calculations when deciding what advance to offer an author, whether a person of standing or otherwise. But their sales calculations are made on a basis different to that which most writers might expect, given what is stated (or used to be) in the typical bog-standard contract.
Writers' remuneration used to be based on a royalty rate, typically (for hardback) 10% of the recommended retail price. That used to be sensible for books with sales of a thousand copies or two (and maybe still is sensible). But when you get a book which is likely to sell tens of thousands -- or which they hope will sell tens of thousands -- the arithmetic becomes completely different.
If you order a print run of 100,000 copies, then the unit cost of each book falls to an amazingly low figure. Which means that a publisher can afford to pay to a writer a sum of money for each book which is far in excess of traditional royalty rates and still make a substantial profit.
That is the main reason why huge advances can be 'justified' -- at least in the eyes of an optimistic editor/publisher.
Other factors also enter into the calculation. A bidding war may 'force' a publisher to pay more than she would wish. After all, you've got to win an auction every now and then or you cease to be a credible player, and you don't get offered 'big' books at all.
Also, there may be other perceived (if not readily demonstrable) benefits from publishing a famous name for well-publicised huge advance. Prestige. Getting talked about as a big-time publisher, which encourages other big names to write for you. And some of those big names may be dumb enough to be so dazzled by your big name that they will work for less than their market value. That kind of thing.
Of course, it isn't an exact science, and, yes, many hard-working writers do end up getting a pittance and little thanks. But, as they used to say in the trenches in World War I, if you can't take a joke you shouldn't have joined the army.