library update

Previous post: Idea (re: libraries)

I talked to Dan the librarian today, and he had an interesting point:

Adults in small-town libraries are conservative readers. “They go back to their favorite authors from the 70’s and 80’s and read through that, then maybe ask a friend. Or they only want audiobooks and make a beeline for the audiobook section every time.”

Moreover, another interesting point: Adults go to the library when they are being thrifty in time and money. “To me, the bookstore is much more relaxed with comfy couches and coffee and tea and everything.” According to Dan’s observations, people generally go in and out, and don’t need help finding books.

He said that he was familiar with online retailers’ recommendation systems, but talked about them as “judgements” that libraries wouldn’t use. The library creates pamphlets and bookstore-like displays, but isn’t very successful with these. What IS wildly successful is the NYT bestseller list, patrons’ adherence to which he admits is a problem. [That awkward phrasing is also a problem. Sorry.]

As for what data were held: the system knows how many times a book was checked out, but not when. (that is so useful, isn’t it?) It also knows what books are checked out, and presumably, which books are on hold. [I think these “on hold” or “waitlist” data are some of the most useful.]

Finally, total quantity of books checked out is the number used for funding/grant-writing purposes.

Dan says that it would be great if he could get patrons to check out some of the older books that never made it big.

My thoughts:

  1. People don’t take books from your displays for several reasons.
    • It doesn’t look like a cohesive whole. These “displays” are arranged on the tops of rather long bookcases, about chest-height. Spread over about the length of a pingpong table, a sparse dozen books hardly look cohesive. Moreover, some are facing out, some are facing in….(these bookcases happen to form an octagon around a middle table where the card catalogs (“search”) are located. It just doesn’t look very good, and why should it? You’re librarians, not visual artists. Barnes and Noble has pretty displays because there are people paid to make them.
    • If they take a book, they’ve ruined the display! This is the same reason why retailers eschew perfectly folded shirts in favor of leaving one or two rumpled: they want you to touch and experience (and buy). This isn’t a problem for bookstores because they have multiple copies.
    • (This is why an electronically generated display is better: it effortlessly looks nice and encourages people to explore.)
  2. There is part of your population (18-25), small as it may be, that isn’t yet set in its reading habits. In fact, your other readers might not be as narrow if they’d had a recommendation engine to help them branch out. This generation is used to having recommendations pushed at it, and moreover, doesn’ t see algorithmically generated suggestions as “pushy.” They’re comical when they’re bad, of course, but then we handily ignore them.

Idea

I finally have an idea for something practical to build.

Problem: at bookstores, the bookseller wants to make money. One way he does this is by arranging books into logical clusters–for example, relatively recent travel-pop-science books (The Geography of Bliss, Paris: The Biography of a City). He hopes that you’ve heard good things about at least one of the books, that that one will catch your eye, and that you’ll start looking at the others. Now he has at least a chance of getting a purchase out of you!

Online booksellers have evolved with similar features to help people find books similar to ones with which they are already familiar. Example: the aptly named “Customers who bought this item also bought:” feature on Amazon.com. Liked The Geography of Bliss? Try Against Happiness: In Praise of Melancholy.

Web-based booksellers, coupled with search engines, also let people who vaguely remember the name of a book (“Something about Paris…”) to find the exact title.

Where do I not have convenient access to the great internet? And of these places, where might I want to find new books to read?

Yes, the public library. I don’t know if my town is behind the times, but the computers that let you search the library’s catalog are strictly for catalog searching: no web use allowed. This is enforced by a stern librarian.

I believe that small-town libraries fail if patrons don’t take books off the shelves. Yes, larger libraries or academic libraries may be meeting places, “living rooms” of the city, the third place before Starbucks (and with no food or beverages), but I believe that small-town libraries are duty-bound to keep up the public’s book consumption. After all, people in cities and universities have great access to cultural events. Not so much in the suburbs.

Book consumption seriously flags if patrons can’t find new books they’d like to read. And usually, they can’t.

1. The search system that my library uses is probably 20 years old. Unlike Google, it doesn’t have the data or machine learning techniques to figure out what result I want to see. Its idea of “most relevant” is vastly wrong, for the most part. Keyword search is an exercise in frustration, and title search is pretty useless unless you have the exact string. Bad for browsing.

2. Books on shelves are terrible to browse through. What do we have to go by: a spine? Hardly enough information. The Dewey Decimal system does sort by subject, but I venture to say that we no longer search by a single subject anymore. People want to use boolean operators (happiness AND travel AND recent AND bestseller), not “Weiner, Eric — Travel.” “Voyages and travels.”

What I propose is to design a system that increases book consumption in small-town public libraries.

This system will use search and recommendation technologies already developed by the for-profit world, and attach it to the resources and values of the small-town public library.

What value and resource constraints might there be, and why might they not be too significant?

1. Libraries should be anonymous. Intellectual freedom, right? I’m not sure how much information they keep about their patrons–I wouldn’t be surprised if it was merely what books were currently checked out. But I also know that I can opt-in to a feature that keeps a list of items I check out, and people probably wouldn’t mind if their aggregate current consumption habits were disclosed. (I haven’t thought through the ethical implications of this fully yet.)

2. Libraries should be unbiased and uncommercial. But you already have the NYT bestsellers list up. Is this unbiased? Certainly not–it’s probably skewed towards the well-promoted and well-connected (there are indie bookstore bestseller lists as well). Lots of people get recommendations off this list, so the library is already influencing its patrons.

What to do:

0. Invest in a Microsoft Surface, or other touchscreen display device.

1. Find some open system of book classification. (google google google?)

2. Based on what types of books are most popular at a given time, display a random collection of books available in the library that are on similar topics. Rotate through, oh, 5 screens of 6 books?

Interaction:

1. Forward arrow to see more books.

2. Pinch for shelf location/call number. Ideally, strips of lights in the floor would also light up to point you in the right direction.

3. Lower right corner to read excerpt/blurb?

4. Lower left corner to add tags. (scan your own card to login so you can add a book to favorites??)

Anyway, this is just the beginning. I’m sure someone out there’s made a start, and hasn’t gone further because it’s not commercially viable.

Link: notes from interview with Dan the librarian: http://wp.me/pMfzI-at