In his Hue-y Decimal System post, David Weinberger is referring to Trevor Bechtel's book-cover-color-sorted book collection.

David and Trevor have been talking about the attributes inherent to books and those added from the outside by their placement in a book collection. I think maybe they are using this as a model for discussing self and identity. But, I am not going to say anything about this. These are just a couple comments from a librarian / information architect.

I once had to find a book for a certain patron given only the description: "It's the big red book on Buddhism". This was during a time when the entire collection was packed in boxes. Subsequently, I decided that book cover color (like page count and physical dimensions) is actually useful information to have in a library catalog!

I have a friend who (much to my chagrin!) organizes his book collection by the height of books, from tallest to smallest, from left to right. The reason he does this is that it makes his bookshelves look incredible neat and tidy—it has nothing to do with providing any kind of utility in accessing the books.

The size and use of a collection hugely affects what kinds of schemes can be used to organize the collection. I recently started keeping most of my CD collection in alphabetical order, even though I would prefer to use various totally idiosyncratic and objectively inconsistent association methods (like: I don't like listening to these two CDs right after each other, so I put them next to each other to remind me not to put both in the car).

With information architecture, and library collections, as the collections of information or books get bigger, the pattern of segmenting gets more and more important. So, alphabetical segmenting (i.e., there is a section of the collection for the letter A, followed by a section for the letter B) lets you do that binary search (am I past the book I am looking for, or before the book I am looking for) to work with smaller and smaller segments.

On the other hand, alphabetical segmenting as the top level organization of a large collection, isn't necessarily the most convenient structure to work with. So, with my CD collection for example, I have some parts of it separated out by genre because, for some genres, I want to make a genre selection before making an artist selection.

Organizing a book collection by color or by size works well if you choose to read books based primarily on their color or size. But, if the book subject or author or title starts to be more important, then color and size organization will start to be a hinderence when you can't segment the way you want to within colors or sizes.

For example, let's say you have 1,000 books of which 999 are red and one if blue. If you want to find the red book on Buddhism, the red / blue organization helps you by eliminating one book. Then, how do you sort through the other 999? Are you then reduced to using an alphabetical order or some other sorting system like shades of colors?

Personally, as I indicated, I like very personally idiosyncratic organization systems for my own collections, almost all of which are pretty tiny at this point. But, like with my CDs, once you get to a certain size, you really need an information architecture. And, with catalogs (both electronic and paper), you can maintain various cross-references.

On the web, there is a certain physical organization across domain names on one level and the link-browse navigation one level down from that. Then, search engines and indexes provide various types of cross-reference systems. And, in-line links are like citation-type cross-references.

I am curious about experimenting more with wikis and these concepts, because I think wikis tend to grow organically in idiosyncratic organization patterns, albeit being reflective of some group idiosyncrasy rather than some individual one. A lot of wikis makes me think of someone else's CD collection before they decided to organize by subject and alphabetize within the subject. I guess it is fun to imagine why it is organized how it is?

Maybe when I get iCites up and running more, I will start a wiki on iCite development. I want to build an integrated wiki-iCite prototype at some point anyway, and maybe this will be the way to do it.