Some simple math should show that one must rationally expect to lose money on the transaction. As I remarked a couple of months ago, the probability of even breaking even is pretty poor. (Believe it or not, it's actually easier to make money on long books than short ones under the current pricing structure.) The breakeven point for copies sold is quite high, unless the author spends even more money on marketing. Bumping up to a higher "service level" only complicates the math and lowers the probability of reaching the financial breakeven point. Keep in mind, too, that this does not compensate the author for his/her time spent in producing the text!
None of this is to say that one's nonfinancial satisfaction cannot outweigh the financial burden. That is precisely the point. The problem is that self-publishing/vanity publishing boosters gloss over this with visions of dollar signs that are about as achievable as the promised $50,000 in a chain letter. A very, very few people actually do make a profit from chain letters (and other pyramid schemes)but the profits almost always flow to the people who operate the scheme, not those who try to benefit from it. Hmm. That sounds a lot like the publishing and entertainment industries in general…