News Scrobbling

Here's an idea for all you VC's out there. I was ploughing through my RSS feeds in my newsreader the other day and thought "But this is what RSS feeds are supposed to stop - information overload and time wasted finding the things I want to read". True, I have a lot of feeds and could probably weed them down...but then again I added them all for a reason in the first place. What I need is the semantic web to decide for me which posts of my chosen feeds I'm likely to be interested in - or which my friends might think I'd be interested in.

Which then got me thinking about Last.FM and scrobbling - how it monitors what you play, compiles it, compares it with others and offers you things you are likely to want to hear.

So why can't we have News Scrobbling?  I'm sure the technology exists. I want something on my laptop which monitors which blogs and news sites I visit, the subjects I'm most interested in, and recommends out of my rss feed what I most need to read  - maybe even compiles them into a newspaper screen for me. And perhaps has a "serendipity" option - "others who liked these subjects also read this". Maybe you could have a serendipity dial which you could turn up or down depending how random a selection you were in the mood for. And an authority slider like Technorati used to have.

It's hardly a new idea. Back in the early 90s, when the internet was accessed at 9.6baud dial-up via compuserve  on my first Apple Powerbook 100 (the one with the b&w screen and tracker ball) there was a  program called Dvorak which claimed to compile overnight the news, stock prices and weather you wanted  to wake up to in the morning. It never worked. Then there was the Pointcast screensaver. Drove me mad. Personalisation has consistently overpromised and underdelivered. But now, the technology must be there for something which works, surely? Or do I need to wait even longer until Web whatever.0?

(In the unlikely event anyone takes this idea and makes a fortune from it, all I ask is a credit "From an original concept by..."  -- or maybe a few shares. Actually don't bother. I expect Google have it in beta already...)

Comments

[Robin Lustig]

Richard: There may already be a readily-available technology which does pretty much what you want. It chooses the things it thinks you're most likely to be interested in, and then delivers the content in an accessible and portable format. I'll probably be drummed out of the blogosphere for suggesting this ... a newspaper?

[David Rendall]

Have you looked at tha latest version of Nick Bradbury's FeedDemon? He's experimenting with solving this problem using attention data. There are also some very interesting developments in the area of social-ising newsfeeds in Newsgator Enterprise Server.

[Ken Payne]

Is this the mysterious idea that interrupted your blogging a while back? As a last.fm addict, I want a piece of the action!

In fact, why leave it at print news? Watched a doc about the Iraq war? Maybe you'll like this other doc about Iraq ... or watched one Arena - here's a South Bank show on a similar subject ... tv of the future?

[Nico Macdonald]

The last.fm model doesn't quite transfer to journalism/writing as music 'scrobbling' uses the number of times one plays something as a measure of how much one likes it. People's behaviour with journalism/writing isn't the same, ie: people don't re-access/read stories they like. It is possible, however, that re-accessing a _site/publication_ could be taken as a recommendation for that site/publication. I am keen to find a way to determine an article is implicitly recommended. Being printed is one, saved to a shared bookmark service another, as it being recommended in a Digg.com fashion. And I am still surprised that the media industry hasn't come up with a better recommendation model -- or better models for debate around the important articles that we _do_ manage to find.

[Ken Payne]

Nico - how about the number of times someone reads the same byline? I'm a sucker for George Packer, for example. Or the number of times you visit the same IP? Or some composite of indicators like that. And is that what they mean by 'semantic web'?