The bus visualization was not a failure - we have it installed in our office and I use it every day, as do many of my colleagues! In fact, out of the hundreds of research prototypes I've seen in my time in the field, this is one of the very few that I know of that are actually in real-world use.

The point I wanted to get across is that an ambient information visualization must take into account more variables than one on the desktop - it has to look good, it has to be interesting, it has to fit with the environment and habits of people, etc. This means that it has to be designed differently from the desktop, just as all other ubiquitous computing applications. When you mention "context", you also have to take into account the context of location and use - for instance, in this case we experimented with different physical settings, some of which were more successful than other.

There is a common misunderstanding that an application or visualization must be understandable without any explanation. This is wrong - if it is correctly designed, a visualization can be very powerful even without being immediately understandable, as long as it is possible to reconstruct how it works logically. This we achieved with the bus visualization by using the geographical metaphor, and it is something I tried to capture with the model.

The conference poster contains more information on the model and is a good complement to the short paper. It can be downloaded here: http://www.viktoria.se/fal/publications/2004/eval-ambient-poster.pdf