Why I'm Loving IMAP and Thunderbird as my RSS

A recent posting from the SearchFox folks talked about 2 models for RSS reading and articles like this point to general fragmentation and frustration with reading RSS. However, the one I've been using: IMAP and Thunderbird is proving to be my favorite and a 3rd option that makes the best of the current state of the tools. I've written about the bits that contribute to this current system here, here and here.For the past week or 2, I've been using this chain of tools to read the ~200 feeds currently in my OPML: a modified version of my RSS2Email PHP script, a dedicated IMAP account just for my feed reading and Thunderbird on the client. I use Thunderbird on 3 seperate machines in a typical day, all configured the same.Note that I'm not using Thunderbird's RSS reading features. There's a good reason for this. It's just too primitive. None of the really useful tools that you can use on email accounts are available to work with RSS. Beyond that, the fact that the items sit on a central server with persistent folders, read/unread flags, labels, etc. as well as spam filtering make the IMAP tools much better suited for digging through 200 feeds a day.So, here's my whole RSS reading process in unnecessary detail.

1. Windows scheduler on my home workstation pings my RSS2Email script. It parses my OPML file, refreshes all of the feeds and emails all new items to my RSS IMAP account.
2. The emails are marked as being "From" the appropriate feed and contain a link to read the original page.
3. I've got feeds tagged as "work" and "home" based on the feeds themselves. This is included as a header in each email as "X-Context". Eventually, this will go through a bayesian filter like POPFile to do this kind of one dimensional categorization. Long-term, I'd like multi-dimensional tagging instead. This would give me "context", "interesting", "time-sensitive", etc. as additional dimensions, but requires a more sophisticated neural net to make it happen. I'm sure it's possible, but I haven't set it up yet. However, even with just this little bit of filtering, I'm quite happy.
4. The IMAP box has an Inbox and an "archive" folder.
5. Thunderbird connects to the IMAP account. I use the 5 "labels" (though only really use 3 of them) that Thunderbird has to mark messages as
   • "Needs Action" in the first label slot - this is stuff I'd like to take notes on for posting here, contributing to my notes for the ebooks I'm working on, etc. I hit the "1" key and the current item is labeled in this area and turns red.
   • "Read Later" - this is for stuff where I read the first paragraph and want to read the rest, but don't have time at the moment. I hit "2" and the current item turns orange.
   • "Archive" - this is for stuff that I want to keep for informative purposes. I hit "5" and it goes purple.
6. I use the views and saved searches in Thunderbird as well as the sorting and "Group by Sort" features to change which items I see. If I'm checking new items, I open the "Unread" view. If I've got time to do some deeper reading, I open "Read Later". I've also got a saved search each for "home" and "work" contexts.
7. I grab the archive view and move items to the archive folder. Once a week or so, I use IMAPSize to pull the archived articles into .eml text files along with my email.

Basically, because the RSS items become emails in a regular IMAP account, I can leverage all of the available filtering, organizational and archiving tools that are available for regular email on the feeds. Email clients have been serving as the place we first really dealt with thousands of items that flow in daily and have much better, mature methods for manipulating than the currently immature RSS tools out there.I won't pretend that I could hand this solution to my mom right now, but the experiment is teaching me lots about how *I* manage large piles of constantly inflowing data. I've used the regular desktop aggregators, web aggregators (even recommending one that was OK). However, none of them let me work with the feeds in the ways I really want to. This also supports my views vs trees hypothesis. The flexible views, filters and saved searches in Thunderbird let me organize and screen items in completely arbitrary ways. That's magnified by the potential of the RSS2Email script, which can add arbitrary headers to any item after doing any sort of analysis (like checking for delicious tags and adding them as X-TagKeywords) as it goes through for more data points for presentation on the client. Stay tuned as this experiment continues and evolves. I have quite a few ideas on how to extend this and make it more useable than just for me (and other geeky DIY hackers).

Technorati Tags : IMAP, email, Thunderbird, RSS, RSS+Reader