Why publish Principles of Data Analysis through Cappella Archive (a micropublisher) rather than a well-known academic publisher?
Short answer: To keep the book affordable.
Publishers don't write books or print them, and rarely sell them. They do, however, provide three services:
For these services they charge a premium on the price of books.
In recent times the process of publishing has changed completely of course: fifty years ago, who could anticipate offset printing, or computerized typesetting, or online retailing? But the principle that you need quality control, a supply chain, and a propaganda machine, has not changed. A publisher a hundred years ago would not use the same terms, but would certainly recognize the same concepts. Given that publishers have been in charge of quality control, supply chain, and propaganda for more than a hundred years and into the internet era, is it not a safe conclusion that they will continue to be in charge in the future?
Not necessarily. I think academic books may have something to tell us about the future of publishing. Hence my experiment.
Academic books are modest things: 1000 copies is a successful book, 10000 is a bestseller. Small numbers also means they are comparatively expensive to produce, which would make them doubly unattractive to publishers. On the other hand, several other features make academic books attractive to publishers.
The first attractive feature is that publishers can out-source quality control to the academic peer-review system at near-zero cost. Now, of course academic publishers would like everyone to think that they are the guardians of quality in the book market. But their quality control is only as good as the peer-review system they use. To be sure, a good commissioning editor can make a big difference by recruiting good authors and good reviewers. But commissioning editors rarely have the level of expertise we expect in a journal editor. Also, book editors cannot ignore the bottom line in a way state-funded academics can. So to a large extent, academic publishers have to let the market do their quality control.
The second attractive feature is that in a low volume market the supply chain can be simplified: books can be printed and shipped to order rather like Dell assembles and ships PCs to order. This is known as ``print on demand''. The machines to do this have only appeared in the last few years, and the production cost is much higher, but cutting out warehousing and wastage makes for big savings on overheads. Some large academic publishers are doing print-on-demand, but only to return out-of-print books to print. Lots of other outfits are going for print-on-demand though---more on that later.
The third attractive thing (to a publisher) is that the market for academic books may be small, but it is easy to target and responds pretty predictably. University libraries, in particular, feel they should have everything that's going, within a broad description. Some publishers (Elsevier, Kluwer, and Springer come to mind) have worked out just how many libraries will buy their stuff and at what prices, and moved in for the kill. Here are some comments from an editor at Kluwer (about a different book):
....some concerns raised by Dr. Saha. Please find my answers below:
1) Style files
You are free to use your preferred Tex or Latex style!2) Price
....95% of academic books are bought by libraries and institutes....sales to individuals remains limited....For a publisher to recoup overhead, printing, distribution, marketing and warehousing cost (plus making a profit) the price to our main customers (libraries) is relatively high: your book would have a list price between 74 Euro (for 200 pages) and 109 Euro (for 300 pages)....
3) Free availability via the Web
....we have experience based on several experiments that we did, including access statistics and comparing sales figures of very similar titles. In general, sales significantly decreased, on average by a factor of 2-3. You can imagine that we seriously looked into this, because if sales would go up by only 10%, it could be a winning business model for our 35000 titles....
Even the pretense of independent quality control has been dropped; hardly anyone is expected to read the book anyway. There is no attempt to improve the supply chain. The whole focus is on the propaganda machine, and the sole aim (business model) is to milk the library market for all its worth.
Surely there must be a better way?
Principles of Data Analysis started out as a set of lecture notes which I put on the web in 1996. In 2000-1 I talked to a university press (JHUP) about expanding the notes into a book. An anonymous reviewer was very positive, and gave useful suggestions for the revision. Then we got to discussing practicalities.... The paperback price, I was advised, would be about $18 for 100 pages or so. The publisher did not even mind my leaving the 1996 version on the web. But giving the full text away free was not an option. For me though, after five years when the best version I had was free, anything less was intolerable.
So the nice editor from JHUP and I parted amicably. He didn't think that micropublishing would be a good idea, but he gave me some welcome advice on the book trade anyway. He was too good an editor to do otherwise.
From a friend in the book trade I heard about Cappella Archive. They are a micropublisher based in the beautiful town of Great Malvern, specializing in quality productions of specialized books, including republication of historic books. Their business model is to concentrate on the supply chain. They do print-on-demand and sell directly; since they don't pay literary agents, distributors, or advertisers, they can still make a profit on very low turnover. They don't invest in the book either, so the author has to meet initial costs. Initial costs are typesetting (nil in my case since I did that myself), the formalities of registering copyright and putting the title on booksellers' catalogs, 6 copies for UK copyright libraries, and 24 copies given to the author for resale (I sent mostly those to reviewers). Total cost to me was GBP 233. Publicity leaflets and a fair amount of book-design advice came at no extra charge. Meanwhile, I keep the copyright and do the propaganda. Print-on-demand publishers in general have a very mixed reputation, but Cappella Archive proved to be superb.
But as any mainstream publisher will tell you, writing a cheque to your publisher is a very bad sign---if the publisher isn't investing in the book, there clearly hasn't been any quality control.
It's not an argument that deserves to be taken seriously. We have already seen what sort of quality-control publishers like Kluwer provide. Even university presses, who genuinely have reputations for quality-control to protect, can only do as well as the peer-review system in your subject. But anybody can organize peer-review. I simply requested several people whom I know to have expertise and interest in the subject (I don't know them all personally) to look at the manuscript and write a few lines about what they thought. As you can see from the mini-reviews the reviewers are not people who are easily impressed. None of them says ``a masterpiece...sure to be a classic''; on the other hand, what they do say is probably worth paying attention to.
Conclusion: this book is a better product than I could have produced with any mainstream publisher. The quality-control level is higher, the paperback is attractively produced, yet it retails for less. And the online version is free.
But merely producing a good product is not enough. Your prospective readers have to know about it; you still need a propaganda machine. Here google already helps. Typing ``principles of data analysis'' at google, shows my book as the top item (at least when I last checked). About 2 people per day seem to find the book through a general search. This is a very modest rate though. To reach its intended audience a book needs reviews in trade journals. I sent books for review to journals through reputable third-parties. At first it was an: uphill struggle. Here's the book-reviews editor of---let's called it The Fourecks Physics Magazine:
Thank you for sending Saha's book on Data Analysis. However the pressure on space for reviews in The Fourecks Physics Magazine is so tight that I can only consider books from major publishers. Except that I make an exception for books from Fourecksian authors relevant to our Fourecksian situation. Unfortunately Saha's book, while I find it interesting, does not fit either category.
Fortunately, some more prestigious publications had other ideas. Once Physics Today reviews a book, publicity to physicists can probably look after itself. At which point the experiment can be declared a definite success. Not only is the book better than it could have been with a major publisher, it is also publicized at least as well and probably better.
A nice result of the Physics Today review was hearing from other people who had done or were considering publishing affordable books. The Physical Oceanography book by Robert Stewart is particularly remarkable. Available free since about 1997 (and in both printable and richly-hyperlinked browser formats) it is now being translated into Spanish and Portuguese, with a Chinese translation also planned.
In the last 20 years academics especially in computer science, physics, and mathematics have become quite knowledgeable about typesetting issues, thanks to one particular individual. Perhaps this was not quite the intention...
[Knuth] was absurdly modest about TeX, and complained that as he went round the world he found mathematicians everywhere crouched over their terminals typing in TeX when they should be in the library! .... His own lectures, incidentally, were all handwritten in several colors on transparencies.
[Timothy Murphy, usenet posting, June 1992]
I caught the TeX bug in 1986 and must be one of the worst offenders. At least I resisted the the temptation to number the pages using primes. I did try some more modest experiments though (this being an experimental book anyway).
Instead of good old computer modern, I used the Euler
(eurm) font for maths. It was designed by Hermann Zapf
and is meant to suggest a mathematician with beautiful handwriting.
For such an attractive face by such a famous designer it's amazingly
little known. The only place I've seen it is in
Concrete
Mathematics
(at least my book keeps good company!) For the text, I used Palatino
(pplr) which I thought blends well with Euler.
Unfortunately the figures use other fonts which don't blend very
well.
Being able to produce the index dynamically is a big advantage. Since the online book comes in a choice of A4 size, letter size, or paperback size, there are three different versions of the index, but the computer looks after all that. I was able to concentrate on what to index, and do that while also editing the text.
Apart from the index there is some additional cross-referencing of page numbers in the text. And the footnote numbering follows the archaic manner of starting by page rather than by chapter.
But computers aren't everything. Sometimes you just need pen and ink. And friends who are good with pen and ink and ideas. So let me end by returning you to Becky Thorn's illustration.