2004-02-01

AlexLit

The Hawkeyes won today. There's rationalizing. Each time they win, I'll write up one of my links on the right. Good thing the season gets tougher.

Alexandria Digital Literature is a site that sells DRM-free eText, specializing on material that had, at the time, been out-of-print. Published authors included Vonda McIntyre, Robert Silverberg, William Sanders, and others. DRM-free? Customers were expected to complete a survey that verified their understanding of and willingness to comply with an "Honor System" method of protecting the author's rights.

I believe that function of the site is still operating, but I haven't tried it in a couple of years. I will say that I've enjoyed most of what I purchased from them. Even better was that I enjoyed the stuff that AlexLit predicted that I would enjoy, especially stories like "Elvis Bearpaw's Luck" by William Sanders.

I found AlexLit in 1997, and it had been functioning for a bit before that point. The basis for the story predictions is the collaborative recommender that they have in place, nicknamed "Hypatia." AlexLit asks you to rate stories as the basic rating unit. For instance, each of the stories in the Starlight 3 collection is rated individually. On the other side of the scale, The Lord of the Rings is rated only once, instead of a rating for each book. You should start by rating the works of your favorite authors, though Hypatia is happy to serve up a list of well-read (and taste distinguishing) works if you just want to start rating. Once you have rated 40 stories, Hypatia is willing to make recommendations for you, both from the AlexLit's marketplace and from her ratings database.

The quirks: AlexLit does not ask you to rank on a numeric scale. Instead, you choose from descriptive words ("Dreadful", "Boring", "So-so", "Enjoyable", "Really Good", "Excellent", and "Fabulous"), though the user can change this. The recommendations are presented with both a predicted rating and a confidence score. Think: "I predict with High confidence that you will find Tom Stoppard's script Rosencrantz and Guildenstern Are Dead to be Excellent." Recommendations are presented in lists, which can be sorted by prediction, confidence, or a weighted combination. The user also has the option to raise and lower the confidence threshold that a story needs to make a recommendation list -- i.e., items on a "Reckless" list only need "Wild Guess"-level confidence to qualify.

AlexLit is open in two very unusual ways. First, the proprietors make clear that predictions are based on the ratings made by your two-hundred "nearest neighbors." Users can even see how their neighbors have rated specific stories. The specifics of matching neighbors have not been revealed, nor have the specific weighting rules used when making predictions. Also, users are allowed to enter new storypoints into the database themselves, as opposed to making requests to system admins.

The website has a small core of users who share access to a message board. Some have used single datapoints to identify themselves as "neighbors," a relationship that is not necessarily reciprocal. Because AlexLit mostly sells Science Fiction, Fantasy, and Horror, there is a bias to the database -- I have had to go elsewhere to look for romance novel tips. Romance has a presence in the system, just not amongst my neighbors.

AlexLit is currently operated by the Seattle Book Company. I thought I knew what Seattle Book does, but their mission is murky enough that I can't tell you. The system appears to be mostly stable, with a few glitches, but if it goes down, it can take weeks for the designer, David Howell, to hear about the problem and fix it. He does have other priorities now -- some of which involve actually making money, I would guess.

So what does AlexLit recommend for me? Well it depends on my mood:

The "Reckless" recommendation is for the young adult novel, "The Players' Boy" by the late Antonia Forest.
If I am merely looking for something less daring, I should check out "The Princess Bride" by William Goldman.

No comments: