I mentioned [[http://feedtree.net|FeedTree]] yesterday — I’ve been thinking about it on and off since then, trying to judge the magnitude of the problem, and the value of the solution.
Let me step back a second. For those of you who don’t know about FeedTree, its basically a tool to help the distribution of RSS feeds (what they call “micronews”). The premise is that these days there are a large number of blogs, an equally large number of people who read these blogs, and a significant portion of readers come via RSS feeds.
Infact, this is an over simplication of the problem, for RSS feeds are becoming ubiquitous these days. Pretty much any kind of content can have an associated RSS feeds — newspapers, magazines, calendars, project management tools, software releases, email. Things are further complicated because RSS feeds are trivially easy to embed in other content (thanks to the large number of tools and libraries available for almost all languages). So you can see my del.icio.us tags on my home page, fetched via RSS. Ditto with flickr.
So the basic problem FeedTree is trying to address is that of efficient utilization of network bandwidth and efficient distribution of content. This is useful for pubishers (I pay for my bandwidth, so I want to get the best value out of it); useful for ISPs (if RSS feeds end up eating a substantial chuck of the backbone traffic, like BitTorrent does right now, you can be sure ISPs will be interested); and end users (I don’t have to wait for my aggregator to “pull” feeds every so often, news will be pushed to me instantly).
The key idea is that a large chunk of this traffic is redundant (say 50% of the participants read Slashdot) — so instead of everyone pulling data from Slashdot directly, we can share the data among each other in a peer-to-peer fashion. The technology is not new, but the idea is. You can read a lot more about the [[http://trac.feedtree.net/project/wiki/FeedTree|gory details and some pretty pictures]] on FeedTree’s website.
So if everyone starts using FeedTree, will all our problems disappear? Leaving the social aspects of adoption etc aside, I think there will still be problems.
First of all, the distribution model for feeds is not as simple as everyone pulling feeds from the publisher. A good example of this are the various [[http://www.planetplanet.org/|Planets]]. Technorati just released Favorites. Then there is Memeorandum and other meme-trackers. So some kind of network aggregation is already happening, though its hardly anything as organized as what FeedTree proposes.
Secondly, these feeds are growing richer in content. Podcasts are just the beginning. Imagine if all feeds had some rich content (audio/video) embedded in them (basically, when the size of feeds grows to more than just a few KB), a P2P distribution system will not necessarily solve the problem (we have already seen the problems that BitTorrent has run into, because its a **highly effective** P2P content distribution mechanism).
Thirdly, I would expect content distribution giants such as Akamai to step into this business soon (if they haven’t already). If NYT and CNN can use Akamai for delivering their audio/video content, they most certainly can use them for RSS feeds. Akamai has a huge, robust content distribution infrastructure that can simply be plugged in.
And finally, as an end user, my biggest gripe with RSS feeds is not the ‘pulling’ aspect. Its the way they are aggregated and presented. Its the way meta data is maintained. I don’t want to have to statically maintain my OPML — I want it to evolve dynamically, learning from my reading patterns and automatically picking up stuff that I might find interesting. I don’t want to have to read feeds in order of their arrival, or categorized by some tags — I want Google news like clustering across my feeds. When I am overwhelmed by the number of feeds, I want to see the largest clusters only and zoom in as I feel like reading more.
I’m looking forward to further results from FeedTree!