[This site will look better in browsers that support web standards, but should be legible in all.]

Friday, 03/04/05

The Seattle library catalog upgrade I mentioned recently was a bit of a bust. I knew it was going to be a later version of the same software, but it turned out to be frustrating in all of the same ways. It still builds item lists with client-side scripts, defeating tabbed browsing, and it's still slower than waiting for chilled butter to reach room temperature.

Simple searches are impaired by extraneous, weirdly prioritized metadata. An explicit title search for Walter Mosley's fabulous post-Watts Easy Rawlins novel "Little Scarlet" returns in the first position "Fairy tales of Russia", because it contains a story (recorded in the contents field, not visible from the results page) called "The little scarlet flower."

Why is the contents field given greater weight than the book title? The entire first page, ten results, doesn't contain the word "Scarlet", and the single instance of "Little" comes in the soundtrack to "Little Women" ("Scarlet fever"). Mosley's book appears halfway down the second page of results, where few Google-trained searchers will even think to look.

If you search for "Little Scarlet" using quotation marks, "Fairy tales of Russia" still has the top spot for some reason, but Mosley is number two. God help you if you're searching for a ubiquitous word like "Ivy" -- 176 results, ten per slow-loading page.

The new RSS support may eventually prove the handiest form of search, because the URL can be adjusted to provide an arbitrary number of results, and one never knows where the title you want is going to turn up. RSS support appears limited to item searches, though (I can't find a feed for, say, my hold list), and without knowing how to pick apart the query URL, there's no way to search for an item not currently in stock. Long way to go. 07:05PM «

Bits pushed by Movable Type