Jakob Nielsen has a couple of nice things to say about BBC News Online in this week’s Sunday Times.
“BBC Online is a breathtaking website, much more integrated with its television and radio channels than, say, CNN. It is another good example of clean design, particularly the home page, which is well categorised, so it is easy to know where to look for breaking news. Where it fails is that it is hard to find things that have been archived. It fronts a huge quantity of rich content, and a vast archive, but lacks an efficient way of digging into it, or even listing what was there yesterday. The web puts out the latest information, which then builds to become a resource. Finding a means of dealing with accumulated content is a weakness of many websites.”
One of things we did concept work on back at BBC News in early 99 in the run up to producing the redesign (the present design) was an idea for ‘storytools’, that could have made use of the Autonomy technology that powers the BBC News search to create ad-hoc trails back into archived content.
Autonomy works by creating and matching the relevence of ‘concepts’ about the meaning of content from the content itself. The Storytools could lead you on a trail back to the origintion of a story by calculating and tracing relevance of the concept the story you were browsing from over time back through the BBC News corpus.
At any point, you would probably be presented with multiple path choices around personalities, stories, issues, places etc to trace your interest back through the archive. I guess the nearest experience to how we imagined it would be Amazon’s recommendations links: ‘other people who have bought this…’.
Again I find myself refering to Ellen Kampinsky’s awesome Amazoning The News
However, it wouldn’t be shaped by the community of users, but purely by the interelationships of content. It never got further than some mocked-up screens, but the tech guys from Autonomy expressed enthusiasm for the concept.
Does anyone know of any content sites that have done something similar in the last two years? E-commerce sites have pioneered innovative ways of encouraging ‘cross-sell’ and ‘up-sell’ but what about the equivalent for content?
If not maybe now I’m back at the BBC I can have another crack at it….
Petafloptimism 
HA! Jacob is totally correct. (The design staff at CNN.com has always been jealous of the BBC site, btw) Could not the community of users determine the interrelationships of the content? For example, ‘other people who read this read this”. Show other paths through the site. Of course this presumes that the reader has choices of other stories once s/he gets to a story page, w/o going back to an index page.
Well – that’s kind of what is suggested in that ‘amazoning the news’ piece My fear is that it would lead to an extent to the results achieved by Yahoo with their ‘most popular’ features.
Yahoo’s most popular indexes as far as I can tell, tend to reflect a pretty consistent mix of big-hitting stories that would be top of the news agenda that day, tech-gizmos, US Sports, britney spears, catwalk models who’s nipples are showing through their haute-couture and pictures of cute tiger-cubs.
The community features that you and Kampinksy suggest would demand more sophisticated ‘story-physics’ to avoid this. It would have to be much more focussed around issues.
To what extent the collaborative-filtering may lead to ‘false-focus’ – trails ‘collapsing’ to revolve around a very clumpy slection of popular stories – intrigues me.
Amazon.co.uk’s recommedations seem ‘stagnant’ to me… i don;t visit them any more as they seem to have fallen into some LaGrangian orbit around my purchases (maybe i’m a boring shopper???)
Whereas Amazon.com seems to be delibrately introducing chaos to the model in the form of users-as-editors… it’s listmania features allows others to carefully compiled lists of products – serendipitous spikes to the system that i can follow to escape out of the ever-decreasing circles of collaborative filtering.
Again though – fine for shifting product… but not for increasing understanding of an issue???
Very true, Ellen’s collaborative story paths were a point of subsequent criticism of the article. Clumpy is very descriptive of what would happen as paths of non related content would amplify into something that looks related. Perhaps primary meta data could be a constant variable that could filter out the false amplification, providing a check and balance so to speak. But is it really even worth it if a sharp editorial staff can provided these connections to other stories without the use of a fuzzy newtonian equation. And i think you are touching on a good metaphor here. Could information architecture better be described as information phyics?
i should say “fuzzy quantum equation”. It has been 10 years since i left the physics department.
But meta data as a constant variable would only resolve clumping if the content was unrelated and being associated by amplification of a “most popular” feature. This brings up another big problem many saw with “Amazoning the News” , that is, secondary (user generated )meta data is presented as more important than the story on the basis of their relative position on the screen. Matt, how about google? does it not do what you are asking? that is present associations without excessive clumping based on self prophetisizing paths?