The Blog Libific

Here's an interesting development: the subscription agent WT Cox has acquired UNC-Greensboro's Journal Finder product. Mike Markwith, who previously served as president of TDNet, will be heading up the Journal Finder team.

More here.

This is a fairly delayed posting, to follow up on the events at NASIG, but I took a bunch of notes, and have found some things to add to them. The vision session that kicked off Saturday was very interesting. The speaker was Mike Kuniavsky, of ThingM, who talked primarily about “ubiquitous computing.” Mike explained that the term was introduced about 20 years ago, and imagines a time when computers are “woven into the fabric of everyday life until they are indistinguishable.”

Among other bits that I liked, were that he showed a graph of Moore’s law – and showed that Moore’s expectation of growth of transistor power really is true. I’ve always wondered, but Kuniavsky showed a graph of time versus “transistors”. In addition, he showed the incredible drop in cost of computing power; an Intel 486 cost about $1500 (in 2008 dollars) for 33MHz of computing power; it now costs about 50 cents for 20MHz. And his comments about the early use of a motor as a general purpose tool to which kitchen gadgets were attached was pretty neat. Now, motors are inexpensive enough that you can put one in each kitchen tool, and you end up with multiple blenders and mixers (to say nothing of all the other kitchen gadgets) in a single house.

In extending this idea to “ubiquitous computing,” he provided examples of a variety of items with embedded processors, such as toys, cars, phones, shoes (Adidas has had one for five years, apparently), blenders, etc. And they have led to a completely new set of products, like digital photo frames, or the QR code, which is common in Asia. I’m in Paris right now (more on that later) and I saw one – or a similar sort of code on an advertisement – in the Metro yesterday.

He talked about examples of ubiquitous computing, but this was where his talk fell down. Interestingly, when he moved on to talk about “information shadows,” and the need for people to identify and describe them, he was both much more out of his element, and much more interesting.

His examples of ubiquitous computing – where a product or service wouldn’t be practical without a processor in each item – didn’t hold up. First, he talked about car sharing schemes. But I joined FlexCar (now absorbed into ZipCar) in, like, 2001, or something, and the system then relied not on a processor or computer inside the car or the member’s keyfob knowing that a car is available, but rather on people returning the car when they said they would. Then he talked about Bag Borrow or Steal (http://www.bagborroworsteal.com) which is just the Netflix model applied to handbags and accessories. No ubiquitous computing here, either – seems to me that the term would only apply if the system somehow knew that the bag you’re done borrowing should be transferred to a person down the block who wants to rent it, and you could facilitate the move without needing to return it to the warehouse, through some system of location identification, borrower identification and location, and a lot of processing power.

However, I do think I’ve found a much better example of the system, and one that didn’t work well before the installment of thousands of individual processors, all networked together. Here in Paris, where I am between NASIG and ALA, there’s a new system of bike sharing throughout the city, called Vélib’. It’s less than a year old, and there are already tens of thousands of bikes throughout the city. You can check out a bike from any Vélib’ station, and return it to any Vélib’ station. Each of the specially-built bikes fits into a special mount to hold them until someone wants to borrow one.



They’re incredibly popular; you see them everywhere. The entire system is managed by JCDecaux, the huge billboard firm. (Apparently JCDecaux and Clear Channel are now battling each other in major cities around the world to run these bike sharing plans, and in return they get a monopoly on outdoor billboards within the city in question.)

The first 30 minutes of every ride is free. It’s reasonable to ‘check out’ a bike, ride it from a spot near your home to your office, and then ride a different bike home. It has dramatically increased the number of bikes in the city (and, presumably, reduced the number of cars). But the system only works if it’s reliable: there must be an available bike when you want to leave (so, the row of bike stands is not empty) and there must be a spot to put your bike when you want to turn it in (the only way to return it is to find an empty stand where you can attach it).

So the problem is ensuring there are enough bikes where people want to take them, and enough spaces where people want to return them. There are people with trucks that move bikes around as needed, and this, as I see it, is where ubiquitous computing comes in. Since there are processors at each stand, a particular station (comprising of anywhere from 10 to 100 bikes, or so) can send a signal when the stands are either full or empty, and guide trucks and their drivers to move bikes as needed. Also, unlike previous bike sharing schemes, this one tracks you with the ubiquitous NaviGo card, so it knows if you didn’t return a bike, or if the bike you returned is all busted up – like the one I saw today, which looked like a car had ridden over the thing.

It seems to me that the introduction of processors in each stand and each station allows the system to identify the most efficient way of moving bikes as needed, protect bikes from theft and mutilation, and also indicates to those managing the system which bikes need maintenance attention. It seems a much better example of ubiquitous computing than those presented by Kuniavsky, and frankly, I was glad to find it, because I hadn’t yet found a good example of ubiquitous computing on a large scale.

He wrapped up his talk by discussing “information shadows,” as I mentioned above, saying, for example, that in some cases, the information is as important as the item. Wine was one example, and I think it fits very well – information about wine is as important as the wine itself, for many people.

For someone who claimed to know nothing about serials, he did a great job of connecting the dots (to misuse some terms he used) between what technologists are doing and what librarians can do to help the rest of the world find what the technologists are building, but not describing.

Technologists, he said, typically leave out the information management issues when talking about technologies, and he feels there’s actually a big gulf between the two – he put librarians “at the forefront of integration of information shadows and representations of objects” because librarians wrangle information about entities forever. He said that the world of dotted-line objects (an interesting concept that I didn’t completely grasp, but one that I think he covered and conceived of very well) needs people who understand how to corral, label, and organize information shadows. The world doesn’t know that yet, but it will, and it’ll be a really big problem. “The world needs shadow wranglers,” he said in conclusion, “and that’s you.”

All in all, an interesting and provocative presentation. Kudos to NASIG for inviting him.

Well, I’ve written already about the challenges of getting to and from NASIG. They were worth it, however. This year’s NASIG was an excellent conference. I missed the opening session due to the vagaries of air travel today, so for me it started out nicely on Friday morning with a talk by Marshall Breeding about open source solutions in libraries. I felt that he provided a nice overview of the industry, though I thought the first half to two-thirds of the talk could have been significantly condensed.

He talked about “open” pushes in library automation – Open Source, Open Systems, and Open Content. Open Source is systems in which you can see, download, and edit the structure of the program itself. I think he did a good job of emphasizing a few points that are often overlooked. Especially at a smaller institution, open source won’t necessarily save money. It may be cheaper to get started and installed, but an institution could be seriously limited in their future projects if the primary (and most likely sole) library employee supporting the system leaves the institution – it may then be necessary to limit one’s search to someone who can support that system, which may not be the library’s biggest human resources need. Human resources are obviously the most important resources an institution manages, and it’s vital that human resources are used for the greatest benefit to the institution. And while a new person is being sought, the system stumbles along without any support at all.

Companies like LibLime and Equinox, which provide support for open systems, can effectively mitigate that problem, but of course they cost money. Perhaps much less than the maintenance contract on a proprietary system, but it needs to be considered when making the initial decision.

Marshall also said “vaporware is vaporware, whether it’s open source or not.” During the Q&A, one person representing a commercial service pointed out that they would often reply to Requests for Proposals that contained many absolute requirements – and when they didn’t win the contract, they’d discover that some open source solution that didn’t meet many of the requirements had been selected. So folks are applying different selection criteria to systems based on their “open source-ness”. Both Marshall and the vendor said that was a problem.

Open Systems allow one to move data into and out of systems. This is an important requirement, though not supported yet by many – perhaps most – ILS vendors, Z39.50 notwithstanding. There is a need to maintain APIs to allow data to be moved in to and out of (mostly out of) library management systems.

I felt that once he started talking about the disintegration of library automation functionality, in part three of his presentation, Marshall started to get to the heart of the issue. Should we be going back to a non-integrated system?

With greater use of APIs and Web services, this should be doable, though it probably won’t be easier. But it’s a goal worth pursuing, and which will allow libraries to select the best vendors based not on what systems will actually be able to access the necessary data, but rather based on what systems will make the most of universally available data.

I had to skip the first set of sessions because I was preparing for a talk I was giving about KBART, the joint NISO/UKSG project I’m co-chairing. I was very pleased at the turnout, since my session wasn’t added until recently. I talked about KBART and its goals, which I will discuss again in a future posting.

After my talk, I attended a session about CORE, a new standards group being put together by NISO to define “Cost of Resource Exchange,” for data moved between an ERM and an ILS’s acquisitions module. This was a good example of the challenges of Marshall’s ‘open content’ question, since one member of the audience raised a question to ask about transferring data between her III acquisitions module and her III ERM. In this case, it’s not a question of a standard, since both products are built by the same company. Obviously, they will make the data move as effectively and efficiently as they can. But how it works between ERMs and acquisitions modules among multiple vendors is the open question that CORE is trying to address.

The talk was presented by Jeff Aipperspach from Serials Solutions, and Ed Riding from SirsiDynix. Ted Koppel from Auto-Graphics was also scheduled as a presenter, but was unable to attend.

I’ll follow later with a posting about Saturday’s sessions.