Joho the Blog
An Entry from the Archives

Greg Linden stumbled across "The Daily Me? No, the Daily Us," an old Wired piece of mine that pointed to a disadvantage of personalized news sites: they don't build communities the way paper newspapers do. (Look, it was an interesting idea in 1995.) Here's an excerpt, chosen because of its quaint reference to that other Iraqi war:

The fact that the document I'm looking at is the same for all who receive it has other important effects. It establishes a baseline of expectations about what we, as a community, are all supposed to know. If, at the height of the Persian Gulf War, we encountered an American who said he or she had never heard of Desert Storm, we would have learned something important about that person.

In short, the act of publishing - which, at its root, means "making public" - helps to establish a public in the first place...

Anyway, the important thing is that Greg points to Findory.com, a fascinating personalized news aggregator. There are no profiles to fill in and no groups to join. Every time you click on a story to read, the site takes it as an indication of your interests. When you return to the home page, it will have adjusted the spread of stories it thinks you care about. If you click on a link to an article that it turns out wasn't interesting, you can delete it from the list of articles you've read.

The privacy policy seems exemplary. There's no registration required. No information about you is stored except for a random number stored in a cookie that's associated with the list of articles you've read (according to the FAQ). It explains how it works its magic in this hard-to-parse sentence:

The algorithm combines statistical analysis of the article text and of users who viewed the articles with information about articles you previously viewed.

I'm guessing that this means that it goes one step further than Google's "find more like this one" option. Google (presumably) does a word usage analysis of an article to find other articles with similar patterns. Findory (I'm guessing) does a word usage analysis of the article and of the article lists of others who have clicked on it, and then compares that with an analysis of your own list. (There are only two problems with this explanation: I probably got it wrong and it's more confusing than the thing it's trying to explain.)

The site also throws in articles outside of your statistically-derived profile to enable serendipity. It does not, however, get past the problem my Wired article points to. But, then, it's not intended to.

A footnote says that "Findory News is provided in cooperation with Memigo." Memigo's banner says:

Memigo recommends news articles you will be interested in based on your ratings and those of other users that are similar to you. The memigo front page can be overwhelming; check out this guide for more details.