The RSS Reader Model is Wrong

April 10, 2010

In the physical world, I buy Private Eye but skip the ‘funny’ comic section. I buy newspapers, but skip the TV and Sports sections. I’ll also infrequently buy other magazines like Wired, The Economist or Monocle and selective scan/read through them.

I certainly don’t read everything in each magazine, but am not frustrated that they’re still in there: after all, that’s a limitation of the printed format. I also don’t mind that there’s great content in magazines that I’m missing out on, because of the cost limitation.

But when it comes to reading content online, through RSS (I use Google Reader), I’m not sure why I’m suffering from the same problems. We seem to have too-quickly adopted the physical model of ‘subscribe to a magazine’.

The simplistic solution would be to use basic filters to remove unwanted content. I’m not sure why Google Reader hasn’t built it into the core platform: I don’t particularly want to use a clunky solution that relies on a third party, like a Yahoo Pipe or FeedRinse.

A diagram for my blog at, showing how the current RSS model is wrong.

As much as I wouldn’t mind specifying a ‘-unicorn -steampunk -disney’ filter for my Boing Boing feed, that still only solves half the problem.

The ‘pizza’ model (the bottom of the three diagrams above) feels like a better method of reading web content: it should all be treated as one amorphous blob (an ‘amorphous blog’, if you will), which – perhaps using bayesian/machine learning or another sophisticated method – is filtered into my ‘stream’.

This might better solve the two basic problems: 1) that there’s content I currently subscribe to that doesn’t interest me, and 2) that there’s content that would interest me, but I don’t subscribe to it.

I think what frustrates me most about this problem is that I thought Google was working on it. The “Like” button in Google Reader would seem to be a simple step towards solving this solution (i.e. the training of the learning algorithm). But no… all it seems to do is to flag and identify popular content.

I’m not sure why I’d want to identify and read ‘popular’ content, any more than I want to listen to ‘popular’ music or watch the latest Hollywood blockbuster – popularity is generally not a great indicator of quality.

Anyway, that’s enough of a rant for a Saturday morning.

Leave a Reply