Update: It's partly my fault for not fully doing the research needed before I could judged Feeds 2.0's technology. I only wrote about what it seemed like from what was up-front. Read the lengthy comment by Nicholas Ampazis, the CEO of Feeds 2.0 on this post for his response. I won't be going further into this. The concept of personalised RSS aggregation isn't new. SearchFox, a company I wrote about back in January (seems like decades!) which at the time closed down and supposedly got bought by Yahoo! was one of the first to grasp the concept. But since then, there's no question there have been several other attempts. Greek-based Feeds 2.0, currently in private beta although they're giving away invites rapidly, is the latest one to attempt this concept. I had a chance to try it out, and I have to say it's on the borderline of good and will-I-really-need-that.
Feeds 2.0 is your average web-based aggregator. Little bit of AJAX here and there, the whole expand/collapse thing, a usable and , it's got things going on. Although there is one difference. On the top right-hand corner of every post, you'll notice two buttons. One says "I don't like news like this one" and shows an arrow pointing down-words and the other says "I love news like this one" with an icon of a heart. Your job as a regular feed reader, is to simply click on either one of those buttons every time you read something. Feeds 2.0 will then progressively be collecting this data and personalising your experience accordingly. As simple as that.
The way it works, is that Feeds 2.0, for every post, feed, whatever, collects lists of commonly used terms, that's terms that have been used throughout or keywords. For every feed, it shows "Tag clouds for these posts" with the whole cloud/text-size-for-importance thing — except nobody told it what the tags were; it guessed. Every time you tell it that you like this kind of news or you don't like this kind of news, it adds or removes the keywords that it collected from the post to its database, your database, and then it collectively 'scores' related posts a mark. No rocket science or artificial intelligence here as it may have first seemed, but it works.
Firstly, I have to give it to them for trying out such a concept. I think it's new and brilliant, and whether or not you use it, it's there for you. Secondly, what I really think is that concepts such as this — you know, 'personalised news reading,' 'user-dependent experiences' — sound excellent on paper and those product pitches, but whether they really work or if they're any use at all to an average user is beyond me. The reason I say this, is because I think if you're going to personalise my experience for me by finding out my reading habits, then there's no better way to do it other than build a super-AI-camera which tracks my eyeballs as I read and finds out what posts my eyeballs get most attracted to while I'm skimming through my daily reading list, and then rates the incoming posts on that basis so the next time I open up my aggregator my eyeballs go crazy with all this stuff I want to read. And that's it.
I think the whole 'I like this kind of news and hate this kind and these keywords float my boat' based concept is good, well-thought of for what we can achieve now, but a waste of time. Don't get me wrong, this sort of stuff works all the time with Junk/spam-mail filters — like the one in Thunderbird which 'learns as you go' — but I think there's a difference between a computer learning to decipher with my help between 'viagra' and 'granny's-cat-funeral-photos,' than 'google maps new version launched' and 'google maps still little coverage of africa.' Note that one is something I care about and the other isn't.
As a regular RSS aggregator, I think some of the capabilities of Feeds 2.0 are excellent. It's usability is one of the best I've seen, certainly right up there with competitors Rojo and Bloglines, and the speed is staggering for web-based aggregators. Feeds 2.0 is currently monetized by AdSense which works coincidentially works relatively similar to their own service with the keywords, and the service is fully free although they might think of adding different packages later on. On the whole, I think Feeds 2.0 is just fine as a regular aggregator, but when it comes down to what it's really trying to achieve — that 2.0 bit from its name — I'm a bit skeptical on.