Comments on: Philip M. Parker writes and publishes over 85,000 books on Amazon

Philip M. Parker writes and publishes over 85,000 books on Amazon

stbalbach — Fri, 08 Feb 2008 10:11:53 -0800

Philip M. Parker ^[1]^[2] has written and published over 85,000 books on Amazon in the past few years, although by his own count the total published is over 200,000. He is like a writing machine - in fact, he has created a machine that churns out an original book about every 20 minutes. A few sample titles:

* The 2007 Report on Wood Toilet Seats: World Market Segmentation by City (330pp., $795) * The 2007-2012 Outlook for Lemon-Flavored Bottled Water in Japan (140pp., $495) * Avocados: A Medical Dictionary, Bibliography, and Annotated Research Guide (108pp., $28.95) * Brain Injuries - A Medical Dictionary, Bibliography, and Annotated Research Guide to Internet References (244pp., $28.95) (Via)

By: wendell

wendell — Fri, 08 Feb 2008 10:26:09 -0800

So we don't really need the web for research, because he's printing it all out for us? The forests are doomed.

By: blahblahblah

blahblahblah — Fri, 08 Feb 2008 10:27:05 -0800

This is really odd. The article in the Guardian (which was supposed to be continued last week and actually talk about Prof. Parker's technique) is written by the editor of the Annals of Improbable Research, which would lead me to think the whole thing is a joke. But there are all those books on Amazon. Perhaps this is an economic experiment conducted by Parker using random pricing and titles? Fascinating stuff!

By: neustile

neustile — Fri, 08 Feb 2008 10:28:17 -0800

So weird that no explanation of this "machine" exists either in the guardian article or wikipedia or the blog post. Did the guardian actually talk to this guy and took his "machine" explanation at face value? Are these just P-o-D views of a large data set?

By: swordfishtrombones

swordfishtrombones — Fri, 08 Feb 2008 10:28:36 -0800

The follow up. Fascinating, indeed.

By: blahblahblah

blahblahblah — Fri, 08 Feb 2008 10:33:42 -0800

The follow up in the Guardian is remarkably information-free:

The book-writing machine works simply, at least in principle. First, one feeds it a recipe for writing a particular genre of book - a tome about crossword puzzles, say, or a market outlook for products. Then hook the computer up to a big database full of info about crossword puzzles or market information. The computer uses the recipe to select data from the database and write and format it into book form. Parker estimates that it costs him about 12p to write a book, with, perhaps, not much difference in quality from what a competent wordsmith or an MBA might produce. Nothing but the title need actually exist until somebody orders a copy. At that point, a computer assembles the book's content and prints up a single copy.

Either the books are pretty low-content or something is fishy with this explanation.

By: pyramid termite

pyramid termite — Fri, 08 Feb 2008 10:36:23 -0800

i've invented a program that can review each one of these books in mere milliseconds 10 PRINT "THIS SUCKS" 20 GO TO 10 my reviews will be done in about an hour and a half

By: Horace Rumpole

Horace Rumpole — Fri, 08 Feb 2008 10:38:37 -0800

my reviews will be done in about an hour and a half Yeah, but it still won't be as fast as Harriet Klausner.

By: dersins

dersins — Fri, 08 Feb 2008 10:41:27 -0800

Are these just P-o-D views of a large data set? The followup makes it seem like it:

"...one feeds it a recipe for writing a particular genre of book...Then hook the computer up to a big database full of info about crossword puzzles or market information. The computer uses the recipe to select data from the database and write and format it into book form....Nothing but the title need actually exist until somebody orders a copy. At that point, a computer assembles the book's content and prints up a single copy."

So, did he write all of the text in his database, or copy-and-paste it from somewhere?

By: seawallrunner

seawallrunner — Fri, 08 Feb 2008 10:43:38 -0800

when you think about this, it's not very hard to replicate. Any of us could do this. Just because he allegedly has published well over 200,000 books does not mean that they do - all - sell. I could also publish thousands of photography books (by featuring a random set of 100 photos per book) using my Flickr stream, and a service like Blurb. I can't imagine that the process of doing this would be so hard to automate. Do Parker's books sell? Do they sell a lot? That's the deeper question.

By: blahblahblah

blahblahblah — Fri, 08 Feb 2008 10:44:10 -0800

I also find it a little sketchy that many of his best-selling automatic books are titled The Official Patient's Sourcebook on [Obscure Disease Here], which would prey on those who really need info. Strangely, many patients seem to like them, though this review on Amazon of the Official Patient's Sourcebook on Interstitial Cystitis was telling: "I was very disappointed when I reviewed this book. It was almost as if the author(s) went to a search engine, and the NIH's Medline, and the National Library of Medicine (PubMed) did a search for IC then made a book out of the results. Of course, the problem with this is that that many of these listings will become (if they aren't already) outdated and "old" information. The book was apparently not reviewed by someone who has knowledge of the IC community, as it listed old and outdated addresses as well. In conclusion: In my opinion, just a few hours on the web "today" will yield more current and useful information than that provided by this book. For those seeking information on IC, I suggest a search on "google.com" instead."

By: cortex

cortex — Fri, 08 Feb 2008 10:44:55 -0800

Parker estimates that it costs him about 12p to write a book, with, perhaps, not much difference in quality from what a competent wordsmith or an MBA might produce. That's a hell of a "perhaps".

By: tapeguy

tapeguy — Fri, 08 Feb 2008 10:51:14 -0800

I also find it a little sketchy that many of his best-selling automatic books are titled The Official Patient's Sourcebook on [Obscure Disease Here], which would prey on those who really need info. Sounds like he's trying to cover the long tail...

By: echo target

echo target — Fri, 08 Feb 2008 11:26:19 -0800

You know, print-on-demand has a lot of great applications, but this is nasty. A souped-up Markov generator hooked up to a Google search would be kind of cool as a proof-of-concept widget on someone's site, but to then print the results and charge for it is dishonest. Especially in the case of all those 'Official Patient's Sourcebook' ones. It seems like there could be a lot of liability issues there. Can he be held legally responsible for anything that might turn up in these books? I mean, he doesn't even really know what's in there.

By: not_on_display

not_on_display — Fri, 08 Feb 2008 11:31:26 -0800

Parker is also enthusiastic about books authored the old-fashioned way. He has written three of them. That's three more than me. I don't feel so little now.

By: ZaneJ.

ZaneJ. — Fri, 08 Feb 2008 11:37:18 -0800

Until it can churn out ravenous smut stories, cheesy romance novels, and unabashed nationalist war stories for the proletariats, I shall not be impressed.

By: koeselitz

koeselitz — Fri, 08 Feb 2008 11:38:16 -0800

From The Guardian: The book-writing machine works simply, at least in principle. First, one feeds it a recipe for writing a particular genre of book - a tome about crossword puzzles, say, or a market outlook for products. Then hook the computer up to a big database full of info about crossword puzzles or market information. The computer uses the recipe to select data from the database and write and format it into book form. This is indeed a very thin and vague description of his 'machine.' seawallrunner: when you think about this, it's not very hard to replicate. Any of us could do this. Just because he allegedly has published well over 200,000 books does not mean that they do - all - sell. I could also publish thousands of photography books (by featuring a random set of 100 photos per book) using my Flickr stream, and a service like Blurb. I can't imagine that the process of doing this would be so hard to automate. It'd be easy to make photography books. You'd just make it put together the pictures, print them, and bind them. Of course, you'd get in a hell of a lot of trouble when the copyright people came knocking, but you'd be fine so long as you made sure you were using public domain photos. It's really not so simple with writing. It'll be a phenomenal feat if Mssr. Parker has pulled off the trick of managing to write a program that can construct wholly new graceful and readable sentences from inputted information. I sincerely doubt he has done this. It's more likely that he's figured out some way to make his "machine" focus only on public-domain information (on, say, Avocado consumption) and copy and paste it into book form. However, I'm assuming he went to the trouble of filtering out copyrighted text. And even if he is, I wonder how long this silly charade can go on.

By: steef

steef — Fri, 08 Feb 2008 11:49:29 -0800

It doesn't sound as if the "machine" is actually writing anything: all it does is format data compiled and written by someone else. Probably in the public domain. Judging from the titles, it looks like pages and pages of statistics, graphs, maps, lists, and charts. Rows and columns, numbers, dictionary entries, and bibliographies. That's all. (The crossword books are a little puzzling.)

By: i less than three nsima

i less than three nsima — Fri, 08 Feb 2008 12:00:49 -0800

Heh, the reviews for his opus "The 2007-2012 Outlook for Consumer Non-Riding Dual-Stage Snow Throwers and Snow Blowers Excluding Attachment Type in India" are funny.

By: Bugg

Bugg — Fri, 08 Feb 2008 12:09:45 -0800

"This book seems like it was written from a template. In other words all the sections are written generically for "Fill in the blank" disease. The authors are not experts on Wislon's disease and don't claim to be. It's basically a reference book that refers you to other sources. There's no real advice regarding Wilson's disease. I looked at this book for 5 minutes before deciding to return it. Buy George Brewer's book instead. He's been studying and treating Wilson's disease for more than 20 years." http://www.amazon.com/review/R2ZUSVH3BCC535/ref=cm_cr_rdp_perm

By: crapmatic

crapmatic — Fri, 08 Feb 2008 12:15:21 -0800

I haven't really looked at any of these links, but for years I've been amused by some of the glowing, flowery fake reviews of the astronomically overpriced industrial books on Amazon (which I guess are his).

By: Bugg

Bugg — Fri, 08 Feb 2008 12:16:59 -0800

It looks like his machine also has a 'generate book review'-setting: http://www.amazon.com/review/R1C4ZMYA7TCSRC/ref=cm_cr_rdp_perm http://www.amazon.com/review/R1S7805ZUBLLFV/ref=cm_cr_rdp_perm http://www.amazon.com/review/R22MUH1I8TKAU7/ref=cm_cr_rdp_perm http://www.amazon.com/review/R3U6RPJSJ6JUCM/ref=cm_cr_rdp_perm http://www.amazon.com/review/R3U6RPJSJ6JUCM/ref=cm_cr_rdp_perm ... I sure could have used one of these machines in grade school.

By: Pater Aletheias

Pater Aletheias — Fri, 08 Feb 2008 12:18:47 -0800

It's like "A Room With A View" but with corn starch. I've just made a solemn vow to say that exact sentence eight times this week.

By: cortex

cortex — Fri, 08 Feb 2008 12:18:54 -0800

It'll be a phenomenal feat if Mssr. Parker has pulled off the trick of managing to write a program that can construct wholly new graceful and readable sentences from inputted information. I sincerely doubt he has done this. It'd be less phenomenal if he just pulled off the trick of managing to write a program that can slot vocab and phrases into previously templated sentences and paragraphs, not bothering to prevent repetition from title to title on account of any individual customer being unlikely to be aware of said repetition. I'd be interested to see how, exactly, he's doing it; but short of a peek at his source code or a decent survey of some portion of his "work", I'm inclined neither to doubt nor to praise the idea of churning this stuff out, as an algorithmic writing achievement. It's not new, and if you aren't striving for really groundbreaking generative results, it's not even hard.

By: seawallrunner

seawallrunner — Fri, 08 Feb 2008 12:23:41 -0800

@koeselitz: It'd be easy to make photography books. You'd just make it put together the pictures, print them, and bind them. Of course, you'd get in a hell of a lot of trouble when the copyright people came knocking, but you'd be fine so long as you made sure you were using public domain photos. In the example above, I would not be in any trouble with the copyright folks as I would only be using the photos that I took by myself, using my own cameras. Full disclosure - I sell my work already, and I just published one book of my photographs. I would never contemplate doing what Parker is doing simply because his expertise does not have any credibility given the size and range of his oeuvre. Similarly, in works of art, if the work is not curated, it also does not have credibility.

By: koeselitz

koeselitz — Fri, 08 Feb 2008 12:29:28 -0800

Yeah, the Markov thing had occurred to me. But there's a difference. It would be hard to make a Markov string generator that wrote sentences that are about real things and were readable. Markov works almost in an opposite fashion from the machine that Parker describes; you input the words you want used, and you get those used randomly in grammatically correct ways. But a program to rewrite inputted information in its own words? Seems unlikely, though not impossible. I guess you could just make it cull a thesaurus to change out words for similar words. But would that be constructing sentences? To create new sentences, you'd have to have the thing actually work out "subject-predicate-object" on its own, and do it in a way that humans would find smoothly readable. Which would be a very interesting project. Not impossible, but very involved, and very interesting. I'd like to see it tried; I wish I had the resources.

By: cortex

cortex — Fri, 08 Feb 2008 12:41:58 -0800

Yeah, I don't mean the Markov thing as an example of how to handle low-level sentence generation in an elegant fashion; part of the charm of the naive Markov model is in the way it consistently fails to be slick. It's almost certainly not part of Parker's method, because the results would be nuts. But it stands out as an example of a very simple technique to synthesize text. And there are others, also relatively simple—synonym/antonym replacements, sentence-creation through simplified generative grammars, stocking key terms into pre-written template sentences/paragraphs, and so forth—and creating a program that combines several techniques as a series of passes would be, if not trivial, at least no great challenge. I've done it myself in bit and pieces of experiments over the years, though it never occurred to me either to do it en masse or to try and charge for the resulting dreck. Getting it to generate sparkling, good prose? Hard. Getting it to generate insightful prose? Hard. But those aren't a problem if your goal is to just synthesize a superficially coherent if crappy bit of writing. And it's clear that neither of those hard problems are on Parker's to-do list.

By: Kadin2048

Kadin2048 — Fri, 08 Feb 2008 12:48:15 -0800

This really sucks; it's the sort of reason why the "long tail" and authors who write meaningful-but-niche content and want to use PoD to publish aren't taken seriously. This guy is creating the book equivalent of domain-squatting pages. In fact I think his whole operation is really just "title-squatting"; he creates a bunch of 'book' titles in Amazon's database, waits for someone to order one, and then prints up a bunch of crap and mails it out. It's the exact same business model that drives people to buy up domain names that are confusingly similar to real ones, toss a bunch of ads up, and hope somebody's stupid enough to click on them. If I were Amazon, I'd be taking a look at this guy and trying to figure out a way to get rid of him ASAP. He's bringing the overall quality of the catalog down -- every time I search for a niche-topic book, I'm going to wonder 'is this a real book, written by an expert, or just 250 pages of Google results written by a machine?' It wouldn't surprise me if he's spamming or forging his own positive reviews, too.

By: melissam

melissam — Fri, 08 Feb 2008 13:46:17 -0800

Sometimes I really hate Amazon. From the whole top reviewers scam to deleting negative reviews to this whole crap fake book thing. Any respectable company would boot this scammer off. In fact, I think the school that employs him should take a look at his publications. There is definitely a plagiarism/copyright charge to be filed here unless he is paying for all this information and citing it properly. It's so typical that he's " the Eli Lilly chaired professor of innovation, business and society." Industry funded professors = poor quality.

By: Kadin2048

Kadin2048 — Fri, 08 Feb 2008 14:16:19 -0800

Take a look at where the 5-star reviews on his "books" are coming from. You don't have to be an econ professor at INSEAD to spot a trend. For example, the The Official Patient's Sourcebook on Interstitial Cystitis has 5 reviews; one is a 1-star, written by someone who looks like a regular customer (10 other reviews on various topics, including a few other Cystitis books plus random fiction). The other 4 are all 5-star reviews. Of those: - Two are duplicates of the same review, by the same "Briana" person, who has reviewed nothing else. - One is by "Health Book Abstracts" and seems to be basically a copy/paste of the table of contents with some filler text. They have 7 reviews total, all of them are of Parker's faux-books, all are 5-star ratings, and all just have the same boilerplate information for each book. - The last is by "I. International "Science"" (purportedly their 'Real Name™' too), who has only one other review to their name ... another Philip Parker book. Ah, the smell of well-groomed astroturf.

By: dersins

dersins — Fri, 08 Feb 2008 14:29:59 -0800

You could always click the "report this" link next to those astroturf-y reviews and report them as inappropriate.

By: fishfucker

fishfucker — Fri, 08 Feb 2008 15:46:39 -0800

Parker estimates that it costs him about 12p to write a book, with, perhaps, not much difference in quality from what a competent wordsmith or an MBA might produce. That's a hell of a "perhaps". Too true. I'm pretty sure that there's no way an MBA could ever write a better book than a computer.

By: bad grammar

bad grammar — Fri, 08 Feb 2008 16:03:20 -0800

I work in a school library that accepts donations. We're obliged to reject (or later weed out) many of these, including self-published authors of zero merit.* The worst book along these lines (I'm not claiming that its writing was completely automated) was a "dictionary" of medical vocabulary for medical transcribers that was just a list, without definitions. I don't know what's sadder, the books that appear to have been automated, or self-published authors under the delusion that their pathetic life stories and skimpy fiction plots will win them fame and fortune. Some worse huckster will probably get the idea of combining them and sell would-be authors software that will automatically write novels for them, saving them the effort. There is already "novel-writing software" but it is mainly an organizational tool; it keeps you from forgetting plot threads, but it doesn't do the actual writing. * I'm not claiming that all self-published (print on demand) authors are of zero merit; it's an acceptable way for beginning poets and for niche hobbyists to publish their work.

By: flotson

flotson — Fri, 08 Feb 2008 16:51:02 -0800

Trying to make a buck by enerating thousands of crap books on diseases culled from internet sources is evil. This guy can scam the corn starch and wood toilet seat markets all he wants, but preying on sick people is low, low, low.

By: flotson

flotson — Fri, 08 Feb 2008 16:51:30 -0800

generating. not venerating.

By: dersins

dersins — Fri, 08 Feb 2008 17:00:08 -0800

preying on sick people is low, low, low. Yes, you are correct. Philip M. Parker, "author" of tens of thousands of "books," appears to be a sleazy huckster trying to make a buck or two by taking advantage of sick people. For that reason, it is my opinion that Philip M. Parker appears to be a scumbag.

By: lupus_yonderboy

lupus_yonderboy — Fri, 08 Feb 2008 18:17:33 -0800

Hah, I scrolled down also to call this man a scumbag. If the books were obviously nonsensical it'd be funny but the medical themes are quite evil, and even the commercial ones are pretty evil.

By: lupus_yonderboy

lupus_yonderboy — Fri, 08 Feb 2008 18:38:19 -0800

I'm about to send a letter -- any comments on the phraseology? I'm not quite sure I have the right boss: their top level's here, and his page is here. [my name and address here] To: Isabel Assureira CC: Philip Parker Good day. I've been reading about your Professor Philip Parker. His idea of producing automatically generated books sounds "cute" until you realize that he's generating fake medical books this way. His extremely expensive automatically generated books aimed at businesses are at the very least morally distasteful -- if the books were correctly labeled as to their manner of production, you can bet that no one would purchase them. But selling fake medical books, that's morally disgusting. You're profiting off sick people by selling them garbage. You can see real reviews on Amazon by people, clearly not doctors, who bought one of your books and were disappointed. I'm also contacting Amazon, a company I've done business with happily for very many years. You should be deeply ashamed of yourself, Professor Parker. You bring shame upon yourself and your institution by making money through fraudulent means and profiting from the sick.

By: flotson

flotson — Fri, 08 Feb 2008 21:29:15 -0800

Good idea. I'm sending one, too.

By: dersins

dersins — Fri, 08 Feb 2008 21:44:27 -0800

lupus, I'd copy everyone on that contact page, since you can't tell for sure. For those mentioned by name but not by email address, email address appears to be firstname.lastname@insead.edu . You might also want to mention exactly what he's doing (essentially printing out google searches for the disease in question, and, apparently, pimping his own books in the reviews), and link to the reviews mentioned in this thread. Hell, link to this thread, as well. It might also be worthwhile to cc: the writer of the Guardian articles and their public editor / ombudsman.

By: dersins

dersins — Fri, 08 Feb 2008 21:46:45 -0800

Since cc'ing is cheap, it might not hurt to copy his colleagues in the marketing department/

By: not_on_display

not_on_display — Fri, 08 Feb 2008 21:54:42 -0800

^lupus_y - let us know if you get a response.

By: stbalbach

stbalbach — Fri, 08 Feb 2008 22:20:28 -0800

Who knew that the power of self-publishing would lead to "book-spam" - books automatically created in the 100s of thousands using data scraped from the Internet and assembled in template fashion. It's a lot like email spam that scrapes content so it looks legit enough to get past spam filters.

By: stbalbach

stbalbach — Fri, 08 Feb 2008 22:33:28 -0800

oh wait.. I just checked and I actually own one of his books! Published in 2002 (it's a medical condition book so I won't give away the title). It is "not bad", I found it useful. It was not that expensive. It has a template feel to it, a lot of stock info and links to more info (which is fine for what it is, it's more than a Google search, well organized and presented). The condition is so rare there is not much else out there available so I was happy to have the book which did have information that was new to me. Anyway, just want to add that to the conversation. Still, the self reviews on Amazon are inexcusably unethical, the high prices for some of the books are ridiculous and there should be a lot more clearing of the air what these books and how they are made.

By: kurtroehl

kurtroehl — Sun, 10 Feb 2008 00:21:28 -0800

This reminds me of a short story by Roald Dahl called The Great Automatic Grammatizator. Of course the guy in the Dahl story sticks to fiction. Publishing unoriginal and potentially obsolete data about medical conditions is unethical, but apparently also legal and profitable. Reminds me of the scumbags exposed in this NYT editorial who purport to run charities for veterans but actually spend 75% of the money on executive bonuses, mass-mailing and exploitative "donations" to soldiers like phone cards for obtaining sports scores. "If we disclose, which I'm more than happy to do," he said, "we'd all be out of business. Nobody would donate. It would dry up." As an aside Dahl's abilities as a short fiction writer are underrated. I recommend that anyone interested in that kind of thing check out his work, which includes several stories based on his service with the RAF during WWII. A Google search for the Dahl piece turned up this Digg post about a computer program called MEXICA designed to automatically generate stories of some redeemable quality. Seems a little dubious to me, but interesting nonetheless. For added meta-goodness here is some Jorge Luis Borges: The Circular Ruins and The Witness. ~~posted by MEXICA~~

By: cortex

cortex — Sun, 10 Feb 2008 08:34:23 -0800

This reminds me of a short story by Roald Dahl called The Great Automatic Grammatizator. Also, Trurl's Electronic Bard from Stanislaw Lem's The Cyberiad. There, one of Lem's odd couple of inventors invents a machine that can generate poetry to specification, on demand. In the story, there are some false starts as Trurl hurries to fix errors in the machine while showing it off to his friend and rival and foil Klapaucius looks on with amusement. The first verse generated in the story is gibberish: Pev't o' tay merlong gumin gots, Untle yun furly päzzen ye, Confre an' ayzor, ayzor ots, Bither de furloss bochre blee! After a couple of incremental improvements, the machine churns out this: Oft, in that wickless chalet all begorn, Where whilom soughed the mossy sappertort And you were wont to bong— And then, after "Trurl yanked out a few cables in a fury" and Klapaucius continues to laugh and mock, "there was a crackle, a clack, and the machine with perfect poise" generates this epigram: The Petty and the Small Are overcome with gall When Genius, having faltered, fails to fall. Klapaucius too, I ween, Will turn the deepest green To hear such flawless verse from Trurl's machine. From there on the machine shines in the face of every challenge Klapaucius can manage. Meter, subject, rhyme scheme, any arbitrary constraint he can dictate: the machine complies, and with grace. Six lines, cleverly rhymed, and every word beginning with the letter 's'? Seduced, shaggy Samson snored. She scissored short. Sorely shorn, Soon shackled slave, Samson sighed, Silently scheming, Sightlessly seeking Some savage, spectacular suicide. And so on. The machine eventually produces a wonderful eight-stanza "love poem, lyrical, pastoral, and expressed in the language of pure mathematics. Tensor algebra mainly, with a little topology and higher calculus, if need be. But with feeling, you understand, and in the cybernetic spirit." It's too long to type out in full here, but the poem finishes out thus: I see the eigenvalue in thine eye, I hear the tender tensor in they sigh. Bernoulli would have been content to die, Had he but known such a² cos 2 ϕ! The story more or less proceeds from there without any other specific poetical cites. It's a wonderful read—all of The Cyberiad is, as is basically everything Lem ever wrote. And as a wonderful added twist on all this, the man was a Polish literary giant whose books are available in English thanks only to translation (or, in the bizarre case of Solaris, very little thanks); so not only did Lem compose carefully constrained poems for this machine, but Michael Kandel had to re-write those poems across to English. Creating some sort of dumb-but-compliant nonsense poetry generator, able to keep up with things like the second example quote above, is a project I've been thinking about off and on for years. Meter and rhyme are things that can be quantified; so too initial letter, and even, loosely, subject given some semantic classification work. Good poetry? No. But bad poetry, idiotic poetry, I think that might be doable.

By: WestCoaster

WestCoaster — Sun, 10 Feb 2008 18:56:58 -0800

I want to offer a belated counter-hypothesis: For the very expensive ($450) books, he is creating only the title and book cover; should someone actual order the book, he then creates it. If that takes (say) four hours, that's not a bad rate of pay. In other words, for many (the majority) of his "books", he's really putting out a feeler as to whether anyone really wants to pay for specific specialized research report. That can be fairly easily automated.