1. What should I read next? – Plug in the name of a book and it generates recommendations. If you enter a book that it doesn’t recognize, it’ll prompt you to sign up and make recommendations of your own. How does it generate recommendations?
WSIRN produces recommendations based purely on collective taste: when books are entered into the same favourites list, they become associated with each other. The more often particular books appear on different lists, the stronger that association becomes. Purely and simply, WSIRN represents mass opinion about books. Over time the recommendations should get better and better as the database grows.
I’m not actually sure whether this works or not. Entering Dickens pulled up Boris Leonidovich Pasternak (author of Dr. Zhivago), Robert Louis Stevenson and Olivia Goldsmith. It would be hard to find more disparate authors. Similarly, Jay McInerny’s Bright Lights, Big City, which I’d probably group with fellow Brat Pack authors Bret Easton Ellis, Tama Janowitz, and pomo lit like Delillo, instead lists Jonathan Ames, P.G. Wodehouse, and Inga Musico.
OK – so this is a new service and it doesn’t have a huge amount of data to draw on and so that’s why the recs are so screwy. But I question the entire recommendation model. If I’m going to write a big list of all the books I like, it’s going to range from Brenda Laurel to Judith Butler to the Gossip Girl series to Tolstoy. Individual taste tends to the quixotic, especially for avid bookworms. Plus, I’m not sure what advantage this site gives over Amazon or Barnes and Noble, both of which have huge databases of customer data informing their recommendations. You’re better off trolling the chicklit.com forums for ideas.
Simple, stripped down, easy to use interface for Wikipedia. Useful for kids and non-power users, but was Wikipedia that difficult to use to begin with? It’s open-source, so I envision this changing quite a bit in the next few months, but it would be good to incorporate some accessibility features if you’re going to muck around with alternative ways to view web content.
3. Streampad, which is a super-cool in-browser mp3 streaming server… thing. You install the helper app on your home machine and you can listen to your entire iTunes library from work, the internet cafe, your friend’s machine, whatever. Nice integration of faddish web stuff to learn about new music: listen to whatever mp3′s people are posting to del.icio.us or to mp3 blogs (through The Hype Machine) and through the Internet Archive”s concerts. (There’s also a completely unnecessary Google Maps mashup stuck in there, just for fun.)
The problem? It doesn’t work. This just isn’t ready for primetime yet, although I would very much like it to be. I couldn’t get a single track to play; the problem is that heavily linked content like songs on mp3 blogs are usually stuck on usendit or on some server that’s down or something and they just aren’t conducive to being quickly streamed to sample them. Too bad.
(This would be really interesting if linked with audioscrobbler, which I still can’t forgive for merging with last.fm and mutating into worst. interface. ever. )
Here’s the Web 2.0 Mashup Matrix for more applications like this. Suggestions?