Bozo Filter for RSS

In the days of usenet, people found that it was nearly impossible to make bozos go away, but it was very easy to set up a bozo filter to eliminate them from one’s view of the universe.

Cory Doctorow is very excited about Feed Rinse — an RSS pre-scrub service that “Automatically filters out syndicated content that you aren’t interested in.” It’s a cool idea, but subject to “the Tivo effect”: by intentionally gravitating toward your own interests and shutting out everything else, you lose the serendipity of chance encounters.

Flipping through LPs in dusty bins is a very different experience from searching for MP3s on a P2P network. Not because the dust is missing, but because you greatly diminish the number of accidental discoveries. With Tivo, you have the same problem: Your plate is full of stuff you like, so you stop channel surfing, i.e. stop finding things by accident.

I may not be interested in reading your posts about baseball, but I prefer to skim over those rather than miss the opportunity to read your post about some freak baseball accident I never would have heard about otherwise.

These problems are parallel to the echo chamber effect, where people in online communities expose themselves only to information that reinforces their existing world view, rather than challenging it.

Technorati Tags: ,

4 Replies to “Bozo Filter for RSS”

  1. I kind of agree with you, but a bozo filter seems like a great idea in an RSS feed-reader. After all, feed readers are (for me, in part) about reducing my reading time to the bare essentials. I would apply the bozo filter to my “must read” feeds to keep them trim. Randomness and delightful serendipity come in my “blog lots” feed group which includes a wide variety of voluminous feeds (like Boing Boing, Ask Metafilter, Metafilter, Monkeyfilter, and a few others). I would not want to apply a bozo filter to that group.

    A different way to filter without Feed Rinse is to do a complex Google Blog Search. For example, if I wanted to read all Birdhouse.org entries that appear in feeds but not read about RSS, I would just search for “site:birdhouse.org -rss” and then add the results URL to my RSS feed reader (or fish out the RSS URL from the results source code). The main weakness to this technique is that it doesn’t let me apply the -RSS to ALL my Google Blog Search feeds at once; I have to add it manually to each of them.

    Where a bozo filter is needed most, for me, is in Google Alerts. I have more than 800 alert searches. I’d love to be able to tell it “never send me an alert for any of my keyword searches if the article found contains the following words anywhere in it.” Instead, I have to add the “do not show” words to every single alert search. Client-side filtering doesn’t work well for this since the alerts contain only abstracts of the articles.

  2. I think these filters could be useful, but they can be abused… It’s like any kind of smoothing or normalization of data… A little bit can be helpful to remove noise, but too much and you have a straight line.

    I would have loved to use a bozo filter to block all those damned “subway map remix” posts that the BoingBoing editors thought were actually interesting about a month ago.

  3. Scot, did you ever read David Brin’s Earth? 1990 near-future Sci-fi, most notable for the world network that plays a pivotal role in the book. (1990, remember – pre-gopher, pre-web, pre-rss, pre-google).

    One of the main characters had a filtering program (written for her by a hacker who was serving jail time) that filtered her email and ferreted out news articles and online conversations for her daily perusal.

    The important part was that her program had a certain… perversity built into it. Intentionally, at her request. It would go out and get items that were outside her interests, things that were bound to piss her off, let through hate mail now and then.

    All to prevent the kind of calcification you just mentioned.

    A neat concept… I’m waiting to see it in action some day.

Leave a Reply

Your email address will not be published. Required fields are marked *