Where the Action is

I just finished reading "Where the Action Is" a book by Paul Dourish. My nominal reason for reading it was that I was about to meet Paul and I wanted to have some clue about where he stood on computing and whatever else the book touched on before I tried to hold a conversation with him. I don't take the time to read many books that aren't directly related to my work, so this was a little outside the norm for me. It is a philosophical work, reflecting on other people's arguments about computing and suggesting an explanation for the current course of Human Computer Interface development. As it turned out I got a lot more out of it than just some background on the author's point of view. Although it surprises me a bit to say it, this was actually the first serious scholarly work that I've read that dealt with any of the philosophical aspects of computing. I always sort of assumed that most of this sort of discussion was carried out primarily in science fiction novels and the popular press and was the realm of pugilistic keynote speakers. I also assumed that most of the debates were fairly unsubstantiatable and basically came down to the various author's opinions of what computers could do. Arguments about "weak" and "strong" AI seem to fall neatly into this perspective.

This book was primarily focussed on the philisophical background of Human Computer Interfaces, though, and in arguing that they should and are going in one particular direction, toward embodied interaction. In the process Paul also dipped his toe in arguments related to Artificial Intelligence: certainly a contentious and ego filled arena to wander into.

What I took away from the book was an argument that HCI is proceeding from text to windows to tangible computing to social computing and that this is because of the reality of the way in which tools and human action exist in the real world. The technical terms that were juxtaposed were that of constructivism and phenomenology. The former being an approach to reasoning which says that an agent can construct a plan separate from the world and then act in the world to carry out such a plan. The latter argues that action and planning are inseparable and that while an agent may plan at some level, the accomplishment of the plan is part-and-parcel of the environment in which the plan is carried out and shows the preformed plan as a very abstract representation.

The impact on HCI is the realization that an interface is never going to be used in exactly the way the designer expected. The plan for the tool and the use of the tool are not easily merged because the use of the tool critically depends on the environment in which it is wielded. And this is something that the designer can only make vague guesses about. Even when the designer knows the exact environment for the tool, they rarely can put all the meaning of a process into the computing representation of it. There are subtle nuances that are embedded in the social and physical relationships of the people using the tool that are probably not worth trying to put into a computer program. Unfortunately for the designer, they have a huge impact on the use of the tool.

The author uses the term embodied computing to capture this notion and suggests that the future of HCI rests on designers who understand that their tools need to be transparent in their use of resources so that the user can appropriately judge when and how to use the tools.

Applying the book to my work, I'm inclined to question the ability of a program to do any sort of planning for the real world that a user can find very useful. A good example of this was the shift that the Activity Compass made from trying to lead the user on a path to informing the user about options that might be available for them. The latter left the responsibility for the actions of the user in the user's hands. I think this suggests some very fundamental problems for pushing my RFID work forward in the direction that it is going.