Too much information to visualise...

...so why not develop computer programmes to visualise without human input. Agrawala, Wilmot and Berthouzoz (of Berkeley Computer Science lab and Adobe Systems) have done that and describe their methods of
- analysing hand-drawn visualisations to understand the design principles behind them
- relating those principles to research in perception and cognition
- and then, using visualisation tools based on the principles, testing people's response to the visualisations they produce.
In the process they explain why people find hand drawn maps easier to use than general-purpose maps when trying to find routes (see another discussion if the principles of hand-drawn maps in an interview with Paul Stiff), and they improve on shop assembly instructions for flat-pack furniture.

I like the strong user focus of their studies:
'We find it critical to involve users early on...[their] feedback is essential for identifying problems and ensuring our design principles and the visualisations converge on effective designs.'