A Wired article entitled Man VS Machines examines the nature of newsreaders right now (Digg, Newsvine et al get a mention) and loosely categorises them into those that favour (AI) algorithms and those that leverage their user base to create semantic links between data. This is a false dichotomy, since both approaches should complement each other - how? Leave content classification & creation (data architectures) to the user, and the rest to the machines. The challenge lies in hybrid approaches.

From a technical perspective, the challenge equates to 'how to use the user base'.

The conflict the article alludes to is the same one evinced in Last FM Vs Pandora and can also be articulated as the AI VS AAI (Artificial Artificial Intelligence - a phrase coined by Rael from O'Reilly) debate. The supposed battle-ground is social software. I'd like to reinforce that the approaches are not mutually exclusive and that the future of social software rests in hybrid approaches, where human agents can consult non-human agents for search, data analysis (mining, visualization) and forecasting; all of which will require the breed of algorithms that are being thrown down the 'machine' end of the debate. Heavyweight AI (like highly refined NLP) is a waste of time at this stage though. It's an academic black hole.