Wednesday, August 1, 2012

The future of textbook publishing is us

In my May column, I announced my intention to give a free online course (a MOOC) this coming fall, and asked for assistance from the mathematical community.

That course, a school-to-university transition course, titled Introduction to Mathematical Thinking, is going ahead, with the first lecture on September 17. There is a brief description of the course, together with a short promotional video, on the Coursera website.

I have started a blog,, to chronicle my experiences working with this new format and to provide a platform for feedback and discussion once the course gets going.

I also wrote a short textbook to accompany the course, Introduction to Mathematical Thinking. Though the textbook is not required for the course, some of my Stanford colleagues who gave the first generation of “Ivy League MOOCs” – just a few months ago, so fast has this new movement taken off – told me that many students want an old-fashioned physical book. On the other hand, all the transition textbooks I am familiar with are fairly pricey, which would put them well beyond many of the students who are likely to enroll. Moreover, none of them are designed to accompany a MOOC. So I decided to write one.

My two main criteria were: it had to be short (no more than 100 pages) and cheap (less than $10 in the US). The only option was to self-publish in print-on-demand format with Amazon’s CreateSpace service. I did not quite hit my page-limit; when I include the front material, the tally comes out at 102 pages, but that’s close enough. But I figured I’d cover my costs if I set the retail price at $9.99, just below my target.

The procedure is so ridiculously straightforward, I can see no reason why anyone should ever publish another textbook a different way, given the huge expense of textbooks. We authors have been typesetting our own manuscripts ever since Don Knuth first released TeX in 1978, and all that is required to produce a book with CreateSpace is to generate a PDF file that fits the page-size you select.

CreateSpace does not provide TeX support (by way of a style file), but they do provide sample pages for authors submitting manuscripts in Word, and I just played around with the page parameters in LaTeX until the output matched their samples for both odd and even numbered pages, which I checked by printing out copies of both, putting my output on top of theirs, and holding the two up to the light. (Low tech, but effective.)

In my case, I decided to produce my book in the standard 6 in x 9 in format, and the key LaTeX parameters I came up with are (for the record)
\oddsidemargin 1 in \evensidemargin .55 in
\marginparwidth .75in \marginparsep 7pt
\topmargin -.5in \headheight 12pt
\headsep .25in \footheight 12pt \footskip .35in
\textheight 7.5in \textwidth 4.95 in

When I submitted my final PDF file, CreateSpace’s automated checking system flagged the manuscript as possibly not being correctly formatted, but I pressed forward, since the next stage is that one of their employees examines the manuscript, and indeed that individual accepted it, confirming my suspicion that I was probably off by a millimeter or two, something that could upset an automated checking system but is close enough to pass a human eyeball test.

The point is, the whole process is so well designed, there is no reason why anyone who can use LaTeX should do anything other than self-publish from now on. With a very small number of exceptions, no one who writes a university-level textbook does so to make money. Our goal is to get material in front of students as quickly and cheaply as possible. If there were a way to do so that can save the students money, I am sure we would all want to do so, the more so given the way textbook costs have skyrocketed in recent years. With modern print-on-demand technology, we now can do just that.

You don’t need to know anything about publishing to do this. CreateSpace does for book publishing what TurboTax did for filing your tax return, and it does it in much the same way, by taking you through the entire process in a simple, step-by-step fashion, including cover design, securing an ISBN code, and selecting marketing channels.

For sure, the finished product is not quite as good as would be achieved with the professional expertise of a good publishing house. But to my mind, for a textbook, it’s close enough, especially when the resulting book can be sold for as little as a tenth the price a publisher would charge.

The one thing I paid someone else to do was copy-editing. I have written enough books to value highly the services of an experienced copy editor. (You might also want to pay an indexer. I did my own, but I have done so for several of my previous books.)

Of course, even with good copy editing, occasional errors creep through. Not long after my book was on the market, I was looking through one of my author’s copies (you have to buy them, but at an even lower price than the retail mark), and spotted a couple of small typos. A few minutes editing the LaTeX file, followed by a quick upload of the replacement PDF, and the correction was made, ready for the next person to buy a copy.

Returning to the MOOC now, let me re-iterate the request I made in my May column. I am giving my MOOC in the early fall to coincide with the many transition courses offered at colleges and universities across the US, in the hope that instructors of such courses will incorporate my MOOC in their courses in some way. My reason for this is that I think the only way to make a transition course MOOC work is to have enough participants who either are already familiar with the material (such as instructors) or else have direct access to such expertise (e.g., their students in a transition course). I see no other way for students struggling to understand the material to get the help, advice, and feedback they will need to progress. Social media provides various platforms for students to interact, to ask questions of one another and to comment on others’ work. But there has to be a mechanism for mathematical truth to find its way into the discussions!

So the key to making something like this work is, I think, to build up a Wikipedia- like community of instructors who, for five weeks each year, will make available their expertise to the thousands of students around the world who are taking advantage of a MOOC to obtain an education they would otherwise not have access to.

The benefit to the students in the transition classes given by MOOC-participating instructors is that their learning will assuredly be enhanced by acting as tutors for the students who are not so privileged. Both because teaching others is a powerful way to learn – as most of us discover when we become TAs at graduate school – and because those students will surely feel much more incentivized to understand by playing such a feel-good role.

Stay tuned to my MOOCtalk blog for updates on the project. And if you are an instructor giving a transition course this fall, please consider getting involved.


Pete Giesin said...

This MOOC looks very interesting. I am moving into the secondary education (math) field after spending 18 years as a software architect. I can use some mathematical thinking refreshing.

BTW, is there any way you might be able to release your book as a Kindle edition as will as print?


Anonymous said...

Pete, Thanks. I hope you find my MOOC useful. The problem with e-books is the display of the symbolic mathematics. At present it has to be presented as images, and the quality is low, virtually unreadable on a small device such as a smartphone. I can't see that issue being solved. A lot of mathematics is simply real estate dependent. I am looking into doing a digital version for a particular device, such as the iPad.

Christian Lawson-Perfect said...

You can use MathJax to display very nice LaTeX in ePub books.
Or Kindles can display PDFs fine, so that would work.

Peter Rowlett said...

"Kindles can display PDFs fine"

Ish. Not great. I have seen maths on a Kindle (not PDF) but it wasn't great. I expect it was the low quality images Keith mentions.

Anonymous said...

Chrsitian, Thanks. I tried that, and it works if the display is large enough and you know the dimensions of the display, so you can design the layout for it, but many of us read e-books on our iPhone, and then it doesn’t work. At least, I don’t like it, and I know I’m not alone. I’ve given up reading a number of technical e-books because the formulas and graphs were unreadable. To my mind, learning mathematics is hard enough, without fighting against hard-to-decipher mathematics. The problem is, e-books are intentionally device independent. Which is why I’m looking to bring out device-specific digital versions, where I can design the math displays to fit the intended screen.

Corey said...

You've left out one of the other reasons for publishing - tenure reviews. It's unlikely that a self-published textbook will "count." (My university has a fairly specific list of point values for publications depending on the type of review and reputation of the publisher. ) So pre-tenured faculty have a disincentive toward self-publishing.

Anonymous said...

Peter, yes, that is my issue. Use of real estate is important in mathematics. That's why I don't use the Khan Academy style presentation in my MOOC lectures. I tried it and decided it did not work for advanced mathematics. After trying various approaches, I ended up simply video-recording paper-and-pencil. I'll say more about this on, and give a sample video from my MOOC in my next post to that blog.

Anonymous said...

Corey, thanks for writing. That will surely change very quickly. A self-published book that garners good peer reviews will rapidly become a valuable item on a resume. We are already seeing that with the "unofficial" certificates from MOOCs. A certificate signed by, say Udacity's Sebastian Thrun, saying an individual has scored in the top 5% of 40,000 students worldwide, will have value once there is a means to certify that the individual was the one who did the work - something Udacity is now starting to provide.

Of course, at the moment, self-publishing really only works well for established authors who have a readership, since there is no publicity and no marketing. So publishing with a prestigious university press is the better option for someone building a career. But that too could change, and likely will. We live in interesting times.

rob eby said...

There is also the tie-in with online homework systems. At my college, we are required to use online homework for every class, from developmental through basic DE, with an introductory Statistics class and a Finite Math class.

So I agree with your comments, but I do not see how this will "kill" (my words) the major textbook publishers, as they tie the text with the online content.

I look forward to taking your class, and I am presenting at AMATYC in november if you wish to continue this further, or discuss possible solutions. (like an extension of the MAA calculus homework system)

Rob Eby

Peter said...

A few comments.

About your reply to Corey's comment. "That will surely change very quickly" is something I've been hearing all my (academic) life but nothing is happening -- academia proves highly conservative. The main problem is that the young researchers willing to seriously experiment will often not gain enough "traditional" merit compared to those who just play the game -- and those who successfully play the game will rarely see the need to experiment later.

This is a serious problem that would deserve much more effort from the few established researchers that are both influential, established, and open to new ideas: help young researcher get the credit they deserve with their experiments such as self-publishing (can't help but add: and publish open-access or even open-source). Or in other words: it's great to hear that self-publishing worked for you, this time, but can somebody else reproduce it?

Finally, LaTeX (as a binary) is nice for producing print output -- but practically incapable of doing anything else (and actually, professional typesetters will easily complain about the quality of TeX's output).

As Peter Rowlett and yourself pointed out, even the best reflow-PDF viewers (Kindle, Nook) are quite limited. However, that is actually the author's fault. It's like trying to build an iPad with manufacturing equipment from 1978 (or for that matter, teaching a MOOC in 1978).

So instead of using LaTeX to do what it can't do -- produce content for an html environment -- authors need to take the next step and switch to authoring systems that can produce both good print and good html. That's hard right now, but worth an experimental effort (good keywords: pandoc, asciidoc, restructured-text, sphinx-doc -- and I'd volunteer right away to help actually.)

After all, with the adoption of MathML3 in two critical standards (html5 and epub3) and with technologies like MathJax, mathematical content in html finally makes sense.

(Disclaimer: I'm involved in the MathJax development)

Peter said...

One small addendum. Here's such an experiment going all the way to XML: Rob Bezeer's Linear Algebra book which (due to it's flexibility) is part of IDPF's official ebpub3 sample repository

Peter Rowlett said...

Interesting. In various approaches, have you tried a smartpen? We just published a booklet on lecture recording and related issues in mathematics and some of the contributors liked this as a pen-and-paper-but-recorded approach.

The booklet is here (free):
Media Enhanced Teaching and Learning: case studies and evidence of effective use.

Anonymous said...

Rob, thanks for your comment. You may be right that major textbook publishers can adapt and remain in business, at least as far as large-sale, mass market textbooks are concerned, though they will surely have to match the POD competition in price. But I can't see how they will be able to prevent the potentially large number of authors of expected-small-run textbooks tailored for their own courses from doing exactly what I just did with the textbook for my upcoming MOOC. They will surely lose that section of their business.

Anonymous said...

Peter, I think you are being pessimistic on the pace of change - but maybe you have been at institutions more conservative than the ones I've been at. Certainly, when I was a department chair at Colby College and served on the Rank and Tenure Committee, we had several discussions that led to changes in the promotion and tenure criteria, based on reassessments of what constitutes scholarship in the different disciplines, and as a dean at St Mary's College of California I was involved in similar discussions.

As to e-publication of mathematics, if/when the technology reaches an acceptable level, I'm sure usage will grow rapidly, but the cognitive importance of page layout in mathematics clearly makes this a challenge. Since most mathematicians *do* mathematics using paper and pencil (or blackboard!), texts surely have to conform to the layout conventions of the handwritten page/board. Right now, LaTeX seems by far the best solution to me, though it does limit publication media. I guess we all need to stay tuned. :-)

Anonymous said...

Peter, interesting that you raise the smartpen issue. I experimented with that in preparing the lectures for my upcoming MOOC. I will report on the outcome of those experiments in the next post to my blog, which should appear later this week.

rob eby said...

In that regard, I think you are correct. I also see the various "custom pubs" probably going away, at least in print form.

As for your online class, I know myself and at least 2 dozen other professional math teachers are signed up, so the comments should be interesting.

Unknown said...

Dear Keith,

Well said! My open source linear algebra textbook was initiated to demonstrate exactly the points you have made in this post. And to back up your comments about your time at Colby, the University of Puget Sound has been very supportive of my efforts, in terms of time, recognition and promotion (both the book and me).

I have recently come to agree strongly with Peter's view, and it is the development and quality of MathJax that makes one rethink the primacy of LaTeX. MathJax makes mathematics in the browser of very high quality and the TeX syntax for entry is a skill most all professional mathematicians already have. So I have recently switched away from PDF being my primary target.

I've converted my semi-structured LaTeX source of my book to XML tags of my own choosing. This (custom) conversion is automated, but still painful, and not re-usable. But I can produce a wide range of output easily from the XML. Check out my current experiments, still in-progress, but look especially at the "Web Version" of the book which makes heavy use of knowls, which are being promoted by the American Institute of Mathematics. From the same source, I am making reasonable LaTeX/PDF but that effort is trailing the HTML version right now.

Thanks for using your "bully pulpit" to talk about the new possibilities for academic publishing. The mathematical community is well-placed to take the lead on this one (again).

Rob Beezer said...

Rob, I guess there are two issues here. One is putting math into Web pages for display on PCs or other devices with reasonably large screens. The other is producing e-books with mathematics, where the screen sizes of readers vary significantly, down to smartphones. I can see growth in the former, but I don't see how it could be done with the latter, unless, as Peter suggests, the community start to write mathematics in a way that allows that -- something that seems unlikely to me. But maybe retinal displays will make all these discussions moot!

[Comment resubmitted by blog administrator]

Peter said...

I find it odd that you think overcoming a system of power (publishing with established publishers to gain reputation in our community) is easy, but something natural (digital natives writing for the web first, not print) is hard.

Of course, there are huge issues (what could "responsive design" mean for mathematical content?), but it seems natural in a world where now all mathematicians grow up with tools such as edX, udacity, mathoverflow, math.stackexchange, physicsforum, wikipedia etc etc etc -- all using lightweight markup languages (and LaTeX-notation via MathJax) which compile easily to both TeX and HTML+MathML.

[Comment resubmitted by blog administrator]

Jiyuu said...

Hi Prof Keith,

I'd like to offer a suggestion, if it is worth considering. I live in Malaysia and while your book is very much more reasonably priced compared to other textbooks, the conversion rate coupled with the cost of overseas shipping still makes it outside the reach of my meagre means. My suggestion is, if you offer the option to buy a PDF version of your book, we can read it on our computer or print it out ourselves (which has the added bonus of reducing its carbon footprint from having to ship the goods^^)

Anonymous said...

Hi Jiyju, I've thought a lot about the use if different media here. Since I view the MOOC itself as an "online textbook on steroids", I don't see any gain in expecting students to read a PDF on a computer screen. In fact, I'm part of a vanguard trying to establish a new form of instructional medium, and I don't want to have old media thinking holding back new media, if the latter seem likely to be better. That means the role of a printed book is to provide a resource to use when the student does not have online access. I can imagine print media always being part of an online or digital instructional package for this reason. Certainly, other MOOC instructors have learned that some students do say they want a physical book, even if the course is designed to be completed without a book (as mine is).

I checked the cost of printing at a commercial printing shop (like FedEx), and here in the US it is $.49/page, which for my book works out at $50. Using Amazon's PoD service, they print the PDF and bind it, for 1/5 of that cost, so that seemed a great deal. But I intend to keep looking at alternatives.

There is another question here: Math textbooks in e-book format that are not designed to accompany a MOOC. Then we get into the issues that have been discussed in this blog-thread. There I am definitely looking at alternatives -- which probably won't involve PDF.

Thanks for writing.