Sunday, February 19, 2006

Discovery of A-List Posts: Or Why A-lister vs Z-lister debate is a red herring

The conversation that started about gatekeepers has moved on to about A-listers vs Z-listers. Or how Z-listers can become A-listers. You can see the part of the conversation here on tech.memeorandum.

If you are trying to be an A-lister you aren't. Now matter how hard you try it just ain't going to work.

Besides the point isn't to be an A-lister but to write A-list posts. To do this you need an interesting topic, something worth saying (I agree posts get tired very, very quickly) and a readable style of writing. Note carefully that last point. Readable style of writing. I have lost count of the number of times I stopped reading a post after the first paragraph because of really bad writing (I'm looking at you Umair Haque of Bubblegeneration). I enjoy reading Gapingvoid because of Hugh MacLeod's writing style (not to mention the cartoons). It is really simple concept. I With the amount I read every day I don't have time to waste deciphering incomprehensible writing.

Back to the A-lister vs Z-lister debate. The real issue is allowing A-list blog posts rise to the surface from whatever blog they are posted on. Robert Scoble and Om Malik don't write A-list posts all the time. The blogsphere will improve if we make the shift away from thinking in terms of blogs to thinking in terms of posts.

More is needed to help good posts to rise to the surface. Technorati has just implemented their "Authority" slider. However, this still looks at the whole blog and not the individual posts. This is a step in the right direction but the slider could do with refinement to look at per post authority. The other major problem with Technorati's "Authority" measure is that it counts links. This fails to account for the "virtual bloggers" who simply read the post and may leave a comment but do not have their own blog.

There is never going to be a magic bullet for finding good posts. What is needed is a series of measures to rate and compare posts. Some of these measures will be social/community based. Others will be algorithmic like Google News. The more methods for quantifying quality the greater the likelihood of a good post finding its way to the surface where ever its location.

To kick start the debate I propose that two widgets be included in blogs. These widgets would allow readers to rate a particular post. One widget would allow the reader to rate the quality of writing and the other widget would let the reader rate the content i.e. is it interesting, funny, flamebait etc. Think of it as the Slashdot moderation system write large across the blogsphere. The same system can be used for comments as well.

The rating widgets fall into the community/social category of rating measures. What is now needed is a algorithmic measures. Bayesian filters fit nicely. Bayesian filters can be used to find good posts across the blogsphere in much the same manner as they detect and junk spam. The filter would compare each individual post against what it has learned are good posts. The filter can be continuously trained by simple yes/no button for readers and by tying the rating widgets.

Over time further measures can be added including rating the readability of the English and creating topic specific Bayesian filters (what is considered a good post of skateboarding may be quite different from what is considered a good post on 19th century Russian Literature).

No system will ever be perfect and the system will need to be continuously refined. A combination of social and algorithmic methods will be needed to discovery good posts. Neither social or algorithmic alone can do it. And if you are a Z-lister wanting to be an A-lister, writing consistently good posts is the fastest way to becoming an A-lister. Whinging about being a Z-lister is not.

Technorati Tags: , , , , ,