A Post by Michael B. Spring

(A list of all posts by M.B. Spring)

Half Way There (October 7, 2010)

There was an announcement last week that the library at the University had demonstrated an on-demand printing press capable of printing and binding books on demand. (The system is the “Expresso Book Machine” -- http://www.ondemandbooks.com/)  It is interesting that during the very same period when e-books first outsold p-books at Amazon, we would find a news story about on-demand printing!  Wired magazine (http://www.wired.com/epicenter/2010/07/amazon-more-e-books-than-hardcovers) reported in July of 2010 that they had sold 143 e-books for every 100 hardcover books over the course of the second quarter, and the rate is accelerating.  While this event had shock value for news organizations, it is important to note, as Wired also reported, that the overall e-book market is still small compared to traditional publishing.  According to Publisher’s Weekly, e-books sales were less than 1 percent of the print equivalents last year, with e-books accounting for about $188 million in sales compared to an overall book market of $35 billion in 2009.

All of this is of particular interest because of my long interest in electronic printing and publishing.  In 1984, I lead a project with funding from the Annenberg Foundation and support from the Xerox Corporation to develop custom textbooks to accompany a television series called Planet Earth.  The project had several interesting aspects, but the one I was most involved with was the preparation and publication of the textbook for the telecourse.   I suggested to the funding source that the problem with telecourses was that they mandated that all of the institutions offering a course in conjunction with the TV series use the same text.  In general, faculty resist not having the text of their choice.  I proposed that we would develop a set of components that could be assembled into many different cohesive texts on demand.  What was proposed was much more than assembling big blocks.

In the last analysis, we constructed a database of components that could be woven together so as to create thousands – in theory millions of different textbooks.  Faculty could specify discipline foci, themes, complexity, or ultimately select each individual component.  Once they had made their decisions, the components were collected and structured into a seamless textbook.  For those unfamiliar with this kind of custom electronic publishing of complex books, consider some of the complications in constructing such a textbook.  A simple problem is the construction of the various tables of contents and the index.  For each book produced, these structures will be unique.  How many chapters there are and what page an index entry will point to changes.  At a more complicated level is the forward reference.  Text books contain lines like “see Figure 7.6 on page 328”.  This requires that we know in advance that the figure being referred to will be on page 328, be the sixth figure in the chapter and the chapter will be the eighth.  This presents a significant processing problem when the contents are dynamic.  We even need to account for the possibility that the section that contains the figure of interest has not been selected!  It took almost two years to solve all the research issues but we managed.  We produced more than 1000 different textbooks which were then reproduced by the participating institutions for their students.  We were proud of what we had accomplished despite the fact that the cost/original, given the overall project budget, was close to $250/master or about $.50/page.  Processing required a powerful mainframe computer and the biggest laser printer Xerox then made.  Total time for composition and printing was a little under and hour.  The system that was demonstrated last week was cheaper and faster by an order of magnitude.  At the same time, the system simple printed a book that had already been composed and stored in memory.  The on-demand nature of the process has been perfected.  The custom nature of the process is still lacking – we are only half way there.

When we completed the Planet Earth project, I looked for funding, unsuccessfully, to carry the work forward.  My vision was one where databases of information might be tapped to create one up books tailored to individuals.  Given the expense of producing these books, I looked for situations where the cost might be considered insignificant.  One of the ideas was to produce “orientation guides” for new hires and transfers within a corporation.  Imagine that “Joe” is to be transferred from Rochester to Dallas within a corporation and will work as a manger in a new division.  Joe will need to buy a house that is in the kind of school district that meets his family needs and that is situated in the kind of neighborhood the family wants.  He will need to learn where to shop, how to get around, what his new office environment will be, who he will be working with, what their skills are, etc.  Even in 1985, we had growing digital sources for much of this data.  If we could bring a transfer or a new employee up to speed a month faster, that would represent a significant savings to the corporation.  While we were unsuccessful in securing funding to pursue this kind of project, others did similar things.  For example, the Department of Defense implemented documentation requirements for major weapons systems that resulted in custom documentation for each instance.  (Military platforms often have a multi-decade life time, and over time, as each instance is augmented, it develops a unique profile.  Put more concretely, as Abrams M1A1 tanks are upgraded and modified, it gets to the point where the tanks are very different from each other.)

An ultimate vision for custom on-demand publishing might imagine a unique text book for each student based on an intelligent assessment of their learning preferences and needs.  People learn in different ways and they have different prerequisite knowledge and skills.  Imagine a textbook that provides just what the student needs and nothing they don’t.  Further imagine that the presentation is such that matches the students learning style.  (You might also imagine how this would make teaching more difficult.  It would no longer be so easy to say read pages 35-92!)  What a wonderful experience it would be for students to have materials that were not too basic or too advanced or written in ways they can’t understand.  Computer and information scientists are continuing to work on these issues.  Personalized and adaptive systems are getting more and more sophisticated, even if they are still limited in scope and functionality.  It is exciting to see the fruition of practical on-demand publishing systems, even it took 30 years to progress from research experiments to practical applications, but we are only half way there.  The shift from mass production publishing to on demand publishing only realizes it full potential when it is not only on-demand, but custom.