This was an interesting session and validated a lot of my thinking about how opml could be used. In fact, one fellow there actually stated my “personal memorandum” idea out loud without any prompting from me. I’ve been prototyping a “feed analyzer” in Ruby on Rails. Trying to decided what kinds of stats the analysis should generate. Do just unique links within and between feeds matter or does the language used to describe the link matter too? I think the language matters. I used the example of one person linking to the Seattle Mind camp home page using the term “Seattle Mind Camp” vs. someone else describing it as “a foo camp like gathering in Seattle”. The only two terms those two descriptions have in common are “Seattle” and “camp”. How does dumb, ‘ole software make the connection between the two and decide that they are describing the same concept? Right now, the popular search engines are just that. Text search engines. The scrape the text in your feed or site and categorize your post or site based on the language contained within.
I put forth the idea that the user should define the context and category of the feeds in their opml, or list, file. I think the key is to make it as easy to use and search as Delicious but as power as Google.