At the 2008 Libraries in the Digital Age conference I attended a month ago, Jeffrey Pomerantz presented results of a study of library and information science curriculum that showed that David Weinberger’s 2007 book Everything is Miscellaneous was one of the most popular books assigned in courses on digital libraries metadata.
“That’s terrible”, I told Pomerantz, “I hated that book”.
A few months ago, I had a brief conversation with Anne-Marie about Weinberger and his ongoing debate with Andrew Keen over just how much the internet is sending our culture to hell in a handheld. At the time, I felt like both Weinberger and Keen were full of it, but realized I should at least read their books before coming down too hard on them. One down.
Most of what there is to hate about Everything is Miscellaneous is that Weinberger simplifies library organization principles in order to prove the point that card catalogs aren’t a good way to organize digital information.
Library card catalogs are an example of what Weinberger calls the “second order of order” - using sortable, paper objects to summarize metadata about a more complex three-dimensional object or text. The first order, he says, is the physical objects themselves (that can each have only one place in the world at a time). The second order was better, he says, but the problem with it is that it can never contain all the information about the object in such a small amount of space.
In the third order, the digital order, every possible bit of information about an object, and every combination of bits, is a possible point for which to access or sort that object on a moment’s notice.
I don’t disagrees that the third order reigns, but it’s hard not to take issue with Weinberger’s image of the library as a stodgy outmoded institution. Libraries were among the first institutions to embrace digital tools, first for transporting catalog records, and later for storing and accessing them. Yet, Weinberger insists on presenting libraries’ tools and methods for organizing information as if they haven’t changed since 1932.
To Weinberger, Amazon.com is the exemplar of organization in the digital age. He tells us that instead of organizing books by putting each one in a specific place on the shelf, as libraries do, Amazon.com files each book in multiple hyperlinked categories. Wonderful, I agree, and libraries have hyperlinked subject headings in online catalogs since the late 1990s, so stop hating on us, dude.
I think what irks me most is that this book could have been so much better if the author had acknowledged that libraries catalog as well as classify, and that in those ancient days, each book in a library had multiple entries in the card catalog: one for each subject heading, one for the title, and one for the author or “main entry”, the last often being multiple cards listing as much information about the object as possible.
Card catalogs were constrained by space and labor, but the principles are the same as Weinberger’s “third order”: file references to a single object in multiple places.
Of course, with that kind of background, Weinberger’s argument is much weaker: if “second-order” orders rely on paper to put access points in multiple places just the same way “third order” orders do with bits, digital objects lose their mysterious omnipresence. Suddenly, it’s not the digital objects themselves that make the “third order” so powerful, only the speed at which computers can process and sort them.
Weinberger also could have taken his argument a step further and pointed out that for digital objects, cataloging and classification are essentially the same thing. What you call it can be where you put it and where you put it can be what you call it. This was the point I kept waiting for Weinberger to make, but he never does.
Weinberger shows some love for libraries by honoring library weblog godhead S.R. Ranganathan and his faceted classification system. In building up to a discussion of why tags are better than the Dewey Decimal Classification system, Weinberger describes a tree that can rearrange it’s leaves at your whim: Ranganathan had a separate slot in his call number for “place”, “time”, “personality”, “matter” and “energy”.
Faceted cataloging is pretty good, Weinberger thinks, and advanced for its time, but not as good as throwing all your metadata into one bucket and calling everything a tag. He doesn’t prove it, but what he describes as a better way sounds to me suspiciously like making a lot of cards, calling each one an entry, and filing them alphabetically in drawers.
Everything is Miscellaneous definitely has its strong points. Weinberger shows that by applying lots and lots of tags to a digital collection, the collection’s structure hierarchy and organization can be inferred. e.g. if a lot of things tagged ‘cherry’ are also tagged ‘pie’, then cherry and pie must have something to do with each other. He argues that an organizational structure built from the bottom up can be just as good as one dictated from the top down.
And there he stops. There are all kinds of interesting things to think about in regards to tagging, flat vocabularies and folksonomies as they relate to library cataloging and classification and other forms of organization, and Weinberger shows interest in none of them. It’s just a refrain, the digital objects are here, the digital order is here, Hallelujah, please make a donation, the collection box is coming around.
To make the point, consider The Library of Congress’ flickr photostream, a wealth of pleasure and intrigue. Many images, such as Going to town on Saturday afternoon, Greene Co., Ga. (LOC), have about 70 tags, in this case including such specious descriptors as ‘old’, ‘octopus’, ‘blue’, ‘dignity’ and ‘afternoon’.
Are these valuable access points? Would a library bother adding them? How valuable is a collection when some objects have 75 tags and some have only two? When is a signal to noise ratio acceptable and when is it just noise? You’ll have to look elsewhere to find the answers.
Everything is Miscellaneous makes some good points, Weinberger’s pithy examples are fun, and as a whole, the book is provocative, so I don’t actually mind that tomorrow’s librarians are reading this book. Hell, if it provoked me, I hope it provokes some library school students too.
Hi, this is Jeff Pomerantz, of the aforementioned conversation at the LIDA conference. Caleb & I have had an email exchange about this post already, but I just wanted to go on the record with some of it.
First, you may have noticed that Caleb corrected the post above. Either he mis-remembered or I mis-spoke (I fully own the possibility of the latter): Everything is Miscellaneous is not widely assigned in digital library courses. Everything is Misc *is* widely assigned in Metadata courses though. And it’s a recommended (though not required) reading in the Metadata lesson plan that we’ve developed for our DL curriculum development project.
And now at the risk of sounding like a complete egomaniac… if you’re interested in the study that Caleb mentioned, there are actually 2 papers. One is on LIS curricula:
http://www.dlib.org/dlib/november06/pomerantz/11pomerantz.html
And one is on CS curricula:
http://www.ils.unc.edu/~jpom/conf/Preprint-JCDL2007_CSsyllabi.pdf
Another part of my email to Caleb was this… oddly, in retrospect, I replied to his comments about David Weinberger by talking about Abby Blachly of Librarything.
About Caleb’s LC flickr photostream comments… Abby Blachly came to speak here at UNC recently. One thing she said in her talk that really made sense to me is that tags work only in very large numbers; 5 tags doesn’t do anyone any good, but when you get 500 tags then they start to become useful. A corollary of this is that “bad” tags (i.e., octopus, for that LC photo) tend to wash out statistically, so that while in principle they remain access points, in practice they aren’t since no one ever sees them. Are they valuable as access points? Probably not. But who cares? We should allow “bad” tags because their cost is virtually nil; it costs more to eliminate them.
Caleb doesn’t agree. He emailed me back & wrote (I’m paraphrasing, since I didn’t ask him if I could copy & paste from his email): yes, it’s true that if thousands of people add tags to a book, the “good” tags rise to the top like cream. But that’s not what libraries are after: our job is not to decide what’s important, but to help people choose what’s important for themselves.
I agree. But, in the spirit of compromise, I’d suggest that “traditional” metadata & Weinberger-style unstructured metadata can and should exist side by side. Weinberger might disagree.
Louise Spiteri at Dalhousie has done some interesting work in this area, particularly her work on folksonomies in public library catalogues:
http://sim.management.dal.ca/People_and_Groups/ProfileMore.php?id=50
And Abby Blachly made it clear that Librarything welcomes research using their data, so looking at Librarything for Libraries in this light would be I think really useful. The thing is, we just don’t know much about under what conditions, in what search situations, for what uses, etc. different types of metadata or access points are most appropriate. But then, of course I think of everything in terms of research projects.
All great stuff! I do realize that Weinberger isn’t the go-to guy for tags/folksonomies/whatever and these links actually help sort some of my questions out - thanks!
I do have a small problem with geek culture’s skin-deep take on libraries and librarianship, as evidenced by this post. I find library and information science fascinating, valuable, earth-shaking, even. Weinberger finds it quaint and outmoded, so I have a word or two for him. It’s a short book, and if I hadn’t read it, I might not be having these thoughts.
This is probably entirely obvious, but I keep thinking that this statement from the comment above —
“if thousands of people add tags to a book, the “good” tags rise to the top like cream. But that’s not what libraries are after: our job is not to decide what’s important, but to help people choose what’s important for themselves.”
and this one from a few days ago & the best source post –
“What we get from search engines and library catalogs is lists of things, and for browsing and discovery and exploration and serendipity, what we really want is groups of things. ”
are getting at the same thing. A few years ago someone described the difference between taxonomy and folksonomy to me as - taxonomy gets you to a thing, folksonomy gets you a group of things to browse. I’m sure that’s a dramatic oversimplification, but I find it useful. And I think on some level it connects these two ideas. The “good” tags aren’t just good because they tell people what’s important; they are “good” because they produce rich pools of browse-able stuff?
Huh! I am definitely getting at the same thing, but I was actually imagining those cream-rising tag clouds in LibraryThing as leading to narrower sets.
This is not really a uncontrolled vs. controlled vocabulary problem, it’s just whenever I see a tag cloud, I just see the big words, and the big words serve to rank subjects (in the context of the object) or objects (in context of the subject) by relevance.
LibraryThing’s ranked list of books tagged ‘fantasy‘ tells me more about the community tagging books than it does help me explore the bits of the genre I enjoy. I have already read or decided not to read 90% of the books in the top 20.
And yet, flickr’s list of tags in the order they appeared is totally overwhelming and unbrowsable with a lot of tags. There’s just no pleasing me.
Perhaps what unifies these ideas is a conviction that more access points are better.
Incidentally, my del.icio.us account makes the worst wordle ever.
Thanks for this post. What I love about the “everything is miscellaneous” book is that the concept so matches my mind. I struggle to file papers because it’s so hard to decide on just one place to put it. Tagging makes much more sense. I can think of 4-6 other ways I might think of any item. Your point about the number that are tagged with a particular label - such as fantasy - does indeed tell you more about the taggers than about the items being tagged. And just as with my post on the question of Wikipedia pages (quantity > quality?), the most popular tags may not be the most useful. And while a taxonomy might get me to the right thing, I also might not think in those terms. Hence we need librarians to work with key word searching. And folksonomies are definitely audience specific to particular discourse communities. How many people would tag things the same way I would - from my generation and background and interests. Not so many, I think.
But the reason that I loved the promo video about “Everything is Misc” - though I confess that I didn’t yet read the book - is that tags work far better in finding things than remembering a single location (whether it is a paper I filed or a book with a single call number). In fact, because it is so hard for me to decide where to file something, I have piles of unfiled papers in my office. My sis in law, a huge fan of Myers-Briggs Personality Types, thinks this preference may be explained by people who are quick definitive deciders - the J, Judging folks who say “don’t confuse me with any further info, my decision is made” - and folks who are endless revisers, the P Perceiving folks who re-think things. Think Bush versus Kerry.
As for how people would tag - can we make any conclusions from looking now at tags - at delicious - to see the range of tags? Someone could publish on this? Probably already has.
I use del.icio.us in exactly the same way! It’s a junk drawer and I can make any piece of it rise to the top if I remember what I might have tagged it. Actually, I tag an object based on what words I might use to recall it later.
I think when we talk about tags, we are talking about lots of things. We mean both ‘my personal tags’ and ‘the collective tags on an object or collection’, and these are very different things.
In libraries, subject headings in a catalog record are like tags but the publisher field is not. From Weinberger’s perspective, for a digital object all metadata can both be ‘tags’. I am mostly annoyed at him for his assertion that only digital objects can work this way.
Can you tell me more about what you mean by “range of tags” in del.icio.us?
[...] year alone I have read Lawrence Lessig’s Code, David Weinberger’s Everything is Miscellaneous, and Jonathan Zittrain’s The Future of the Internet and How to Stop It. I also checked out [...]