Beginning of the Attention Engine race

Wow, I missed this the first time around…

Megite is going letting me do what I’ve been asking Memeorandum (or enyone else that will listen) to let me do for ages  – to pivot off my own OPML file. The feature isn’t switched on for everyone yet, but I’ve pinged the Megite developer, Mathew Chen, so hope to hear from him soon.

(Update 14 Feb: Matt has got back to me – it’s live! See. Thanks Matt! Robert Scoble has one too. )

Richard MacManus has more info on this.

However, it turns out Findory can let me import my OPML too. Today. So I did. This is what I saw after:

The main pane provides ‘Top Stories’ based on my OPML file (I uploaded 642 feeds – the other items in my OPML aren’t RSS feeds, they are links to stuff). On the left is a selection of the blogs I’m subscribed to.

The upload process is clunky – your need to copy and paste the OPML file source rather then just pointing to an url, but hey, I’m not complaining.

I don’t know how the ‘Top Stories’ are decided, but the end result is pretty good. Practically every item is of interest to me. Not suprising really – it’s pivoting off my OPML / Attention data.

It’s funny – I’ve been thinking about this a lot recently, so to see services actually switching on some of this stuff is very exciting. But now I’ve actually seen it, I’m asking myself more questions….

So, I ask myself – are the ‘Top Stories’ just randomly selected items from the RSS feeds from my OPML file? If so, am I getting a more ‘relevant’ experience than if I go to a news tracking site (or ‘meme tracker’) that doesn’t take into account my interests? If so, is that ‘real value’? I’d say maybe not. Randomizing my feedreading experience is not what this is about…

What’s the algorithm at work here? Is there one beyond randomization? What data other than my own OPML file is Findory using to determine ‘relevancy’ to me?  How could it be improved?  I’ve got to play some more…

Now I’ve seen it live, I’m more convinced than ever that the ability to render a personalized experience based on Attention data is where its at. And I’m not talking about just clickstreams.

Your OPML file (specifically your list of RSS subcriptions) is one example of this Attention data set. It says alot about you: the topics your interested in and the people you listen to, and much more. There is plenty more Attention data that can be leveraged though. My tags, my wishlist, the books I own, etc.

We’re just at the beginning of the Attention Engine race.

It’s going to be great.

Related: My Attention writings

Tags: , ,

Comments (13)

  1. <i>but the end result is pretty good..<br />

    ..<br />

    What’s the algorithm at work here? Is there one beyond randomization? .. </i>

    Make up your mind, either it is good or it is just random. You got what you say yourself are some random results but somehow this is all ‘going to be great’? c’mon.

    My thoughts are on my blog

  2. Greg Linden says:

    Hi, Alex.  Thanks for trying Findory!

    The relevance rank is not random nor is it solely based on your OPML file.  Findory decides what is relevant based on the articles you read.

    Play with it, click a few articles, and watch how it focuses in. Findory learns very quickly.

    Hope you enjoy it!

  3. MSDN Archive says:

    Matt: Brilliant, thanks! Am playing with right now…

    Nik: A page that takes into account my interests in better than one that doesn’t. But randomizing the posts on a page is not really interesting. That’s my point. I think I wrote that.  I read your post too. I disagree. I’ll write up why later.

    Greg: Sounds like you’re using clickstream to augment the Attention data. Will play more.


  4. Kevin Burton says:

    Dude… I got your back ! 🙂

    TailRank has had this live for 2 weeks now.

    It was as soft launch though as I’m working on some more functionality here.  

    We actually reworked this from before (as I’m sure you’re aware)…


  5. The following is a continuation of the connections between new media network interaction and Deleuze &amp; Guattari’s concepts in A Thousand Plateaus. That said, while it doesn’t directly mention many other topics bouncing around online communities, parts

  6. After my post about Personalized Megite, I got taken to task by both Gabe Rivera from Memeorandum and Nik Cubrilovic from OmniDrive – two developers who have had a lot of experience trying to develop such systems. As Gabe wrote…

  7. I created a quick screencast to show what I mean by the term ‘Attention Engines’.

    Before I sign out…