I’ve given a lot of talks over the past few years, and in virtually every one (along with quite a few of the articles I’ve written), I’ve rhapsodized about the Espresso Book Machine (EBM) from On-Demand Books (ODB, a truly unfortunate acronym for a bookseller). I’ve gone so far as to characterize this machine, despite its clunky ugliness, as incredibly sexy — while allowing that, as a librarian, my threshold of sexy may be lower than average — and I’ve said that it has the potential to change utterly not just the nature of the library collection, but the whole world of publishing. By freeing publishers from the tyranny of the print run and libraries from the enormously wasteful practice of building huge just-in-case collections based on inevitably erroneous guesswork about future patron needs, the EBM could greatly increase both efficiency and effectiveness, allowing a library or bookstore to give researchers exactly what they need within minutes of the realization that they need it, all while reducing the clear-cutting of rainforests, the carbon emissions from pallet-laden delivery trucks, and the twin scourges of returns and remainders.
Reader, we bought one.
And almost two years later, I don’t regret it. However, in the spirit of “How We Done It Bad,” I want to share some of the lessons that we’ve learned from our experience so far.
Nothing is ever as good as it sounds. I’m a middle-aged man, which means I actually learned this lesson long ago, and I honestly wasn’t expecting our EBM to really usher in a millenial day of perfectly friction-free access to an unlimited supply of high-quality books. But I did expect that a machine that promises “instant distribution of over 3,000,000 titles” would offer something like very fast production of lots and lots of high-quality books. Technically speaking, the EBM does print books very quickly. It takes about five minutes to print a 300-page book — as long as the machine is warmed up. If it isn’t, you’ve got to let the glue melt, which will take 45 minutes to an hour. And, of course, 3,000,000 titles sounds like a lot of content, but much of it consists of very old titles in the public domain, only some of which represent content that anyone cares about. We knew this going in, but it’s still been a bit disappointing that more current content hasn’t been added more quickly to the database. (On the other hand, one of the great strengths of the EspressNet book database is its depth: shortly after installing our EBM, we were able to find and print an obscure 300-year-old German text for a faculty member who had been trying, unsuccessfully, to find a printed copy of the book for years. That EBM-powered serendipity changed the structure of his course.)
Great concepts don’t print books; functional machines print books. We were very fortunate in being the second desert-climate library to purchase an EBM. The first was Brigham Young University, and ODB had to scramble to repair some unanticipated (and, to be fair, probably unanticipatable) climate-related problems. One of the more amusing ones involved static electricity: because the ambient air is so dry, pages wouldn’t pile up cleanly as they emerged from the text-block printer. ODB had to send a technician to Provo to cut a hole in the top of BYU’s machine and install an ionizing fan. By the time we installed our machine in the very similar climate of nearby Salt Lake City, that particular bug had been worked out. But being a young technology, other technical problems remained: we had issues with balky and leaking ink jets, malfunctioning sensors, and recalcitrant cover feeds, all of which have been fixed or mostly fixed at this point. And we’re still waiting for a color text-block printer that will communicate effectively with the machine, despite having been offered that option initially. For now we’re still making do with black-and-white (we can print covers in color without any problem).
No matter how sexy the delivery mechanism, the content matters more. As I mentioned above, we’ve been disappointed (though not shocked) that publishers are generally slow to allow frontlist titles to be printed and purchased through the EBM. To some degree, this can be explained by the EBM’s very small installation base: 45 machines in 41 locations worldwide, 12 of which are libraries (where publishers might not expect many sales to happen). I would imagine that the cost of making one’s frontlist available via EspressNet is significant, and if the road to recovery of that investment is lined with fewer than 50 purchase points, I can’t really blame a publisher for putting off that step until the market has proven itself. This raises something of a chicken-and-egg conundrum, of course.
No matter how sexy the content, a bad search interface will make it inaccessible. EspressNet was hobbled from the beginning by a simple but ineffective Google-style search interface: the only option is to search by keyword. ODB is promising a more advanced search interface shortly, which will be a welcome development — but this brings us to the next point . . .
No matter how sexy the search interface, bad metadata means bad search results. The real problem with search in EspressNet isn’t the inflexibility of the interface, but the abominable quality of its metadata, much of which comes from Google Books. At this point in time, searchers cannot assume that their results are accurate, which is hugely frustrating. Inflexible search is a problem, but bad metadata in a 3,000,000-title database is an enormous problem, one that can’t be solved without significant expense. ODB is a young company with a smallish staff, and it has no realistic way of fixing and upgrading these records itself. In the course of several phone conversations with ODB executives, I’ve strongly encouraged them to contact a wholesale broker of metadata like OCLC, but to my knowledge nothing has yet come of that. Until it becomes possible to search its database of books effectively, the EBM’s incredible promise will remain largely unrealized.
You can’t predict what people will get excited about. We thought that the EBM would generate excitement primarily for its ability to give people quick and easy access to millions of books, many of which had been virtually inaccessible before. There has been such excitement, and more will likely be generated when some of the issues described above have been resolved. In the meantime, we’ve been startled by the unanticipated ways in which the EBM really has caught people’s attention: there’s significant demand for blank-page journals that we print up on the EBM, bound in covers featuring images from our library’s rich digital collections; we sell these, steadily, for $7. We experimented successfully with printing and shipping the annual proceedings for a scientific society, and have seen great demand for our self-publishing services. We are currently in talks with our campus bookstore about cooperative selling arrangements, whereby the EBM can act as a sort of expanded backlist warehouse. The lesson we’ve learned here is that just as expected opportunities may fail to materialize, unexpected ones will almost surely crop up if you’re watching for them.
Being an early adopter is expensive. We knew that by buying an EBM relatively early, we were going to end up paying more than those who hop on the train later, both because the machine itself will shrink and become more efficient and cheaper over time, and also because buying into a young technology is kind of like adopting a toddler or a teenager: they’re awkward and they eat a lot. The EBM has “eaten” a lot of staff time and energy, and we fully expect that it will continue to do so for a while.
Start-up companies don’t always fully know what they’re getting into. The EBM and its attendant technologies are not only new to us and our patrons; they are also new to the company behind them. My impression is that ODB is still struggling to figure out the right balance between treating the EBM as a retail tool and a library technology. This is not an either/or question, obviously (we use our EBM for both purposes), but ODB faces a significant challenge in figuring out what the right mix of emphases will be, and how it will position itself to serve well those two very different markets. As mentioned above, the company also clearly did not anticipate the importance of quality metadata and is now scrambling to deal with that problem. These are young-company and new-technology issues, and are to be expected.
The bottom line is that I’m kind of cheating by characterizing this as a “How We Done It Bad” piece. In fact, I’m very glad that we bought our EBM. In hindsight, I might have waited a year or two — but given the unpredictable budget environment, it’s not at all clear that we could have done so. Having an EBM has been fun, exciting, and frustrating, and I fully expect that it will continue being all of those things for the foreseeable future — with the mix gradually shifting away from “frustrating” and towards “fun” as the technology matures and as we keep discovering new ways to put the EBM to good use for our patrons.