Sunday, March 02, 2008

How to get rich with OPML in three easy steps

opml-icon-64x64.pngI continue to be fascinated by OPML thanks to my new job, where I deal with OPML files a lot. Late last night I couldn't stop thinking about OPML-related products. I am now convinced this format will revolutionize search and social media.


Here is how I would beat Google (at search) and get rich using OPML, had I the time:



  1. Build an RSS reader that supports "live" OPML subscriptions using OPML URLs, as discussed here, with the ability to pull in arbitrary sections of the outline. It should similarly support "live" OPML export.


  2. Figure out who people's friends are and where their friends' OPML files live. We care about these files even if they are not being viewed in the reader in any way. One way to find these files by allowing people to specify their friends ala social networking. If their friend uses your reader, you already have an OPML file for the person.


  3. Build a search engine off of this reader that is personalized for each user. It searches all your feeds, of course, but it also searches your imported OPML feeds, then your non-imported friend OMPL feeds, then the cosmos of sites/URLs closely related to all these feeds, then the Internet at large.


The key idea here is to augment Google's PageRank with something I'll call FeedRank.


PageRank ranks pages based on the number of links to each page. FeedRank ranks pages based on how close they are to your feed reader.


FeedRank gives heaviest priority to your feeds, next-highest to feeds of others (OPML) you subscribe to (live OPML folders), then the feeds of others who are like you (OPML of your friends). Only then does it look at links on the Internet.


FeedRank is more resistant to tampering than PageRank, because while PageRank ranks baded on links culled from the entire Web, including tons of sites placed there with the intention of gaming PageRank, FeedRank would contain the crucial added information of who you personally have chosen to trust OPML is key to the concept because it it allows you to increase the number of feeds in your cosmos by an order of magnitude thus providing fuel for FeedRanking. Note that while you'll want to read some of the OPML feeds (ala Dave Winer's Reading Lists idea, which I call "live OPML folders"), others will just be in your account for purposes of building your personal search engine.

1 comment:

Marshall Kirkpatrick said...

As a hack, I've been publishing OPML files that include one Feedburner file populated only with like an empty del.icio.us feed, then splicing in other feeds I discover later via FeedDigest, pipes, etc.