Tom’s Wordpress Blog

Consensus Web Filters made a big impression on me. And it’s not just me; there are lots and lots of links to that post all over the place. So when I stumbled across Man vs. Machine in Newsreader War, I was hungry for the information.

This article focuses on whether future collaborative-style news sites will depend more on human-edited or algorithmic sources. It pretty much picks algorithms over meat. Sites like Digg and its ilk depend heavily on submissions from web surfers.

But it seems the filtering services offered by these sort of sites are both too broadly and too narrowly focused at the same time. Mary Hodder of Attention Trust (a fascinating topic on its own) like the current sites, but thinks they’re too narrowly focused:

“Digg and Memeorandum are definitely an order of magnitude better than anything we got from any top-down news organization, but when I look at them, I see all the things that are missing,” said Hodder, CEO of the video aggregation startup Dabble. “Digg and Memeorandum are catching one slice, and it’s fantastic and a total breath of fresh air, because it’s not The New York Times or the L.A. Times. But it’s still only one slice. If you are really going to nail this, you have to have thousands of slices.”

The gist of her argument is the limitation imposed by submissions from a limited group, and suggests than many more perspectives need to be taken into consideration for the filters to be really useful.

Almost to support the algorithmic approach, I checked out a new filter mentioned in the article, Tailrank. When you join Tailrank, you submit a list of feeds you’ve been reading. The sites secret recipe shakes and bakes though them to make a constantly-updated customized list of recommendations. Though the algorithm isn’t revealed, it obviously depends on links to blogs; the more links, the higher the rank, probably.

A drop down menu on the user’s page lets you select the number of links to use for a filter (this is after you’ve signed up for a free membership). When I selected 2 links, Tailrank returned 136 blog posts, 8 links 28 articles and so forth. So it was a little ironic that when I set the filter to use the maximum number of links (35), it turned up just one article: Consensus Web Filters.