design interactions, not applications

Among my readings of the weekend, the paper by Saul Greenberg, who revisits the concept of context and analyzes why designers find it so difficult to build context awareness applications.

He gives a brief explanation of three theories of context: Situated Actions, (our most beloved) Activity Theory and another approach I have never heard of before, the "locales framework".

According to the author "the Locales framework was developed as a principled approach to help people understand the nature of social activity and work, and how a locale (or place) can support these activities ... In many ways, the Locales framework is about the social construction and use of context, and it too recognizes its dynamic properties. Locales (the site and means) are the external contributors to context; although locales can be fixed, most are fluid.

His overall argument, though, is more interesting: if in many cases some contextual situations are fairly stable, discernible and predictable, there are many others that are not.

That implies that 1) it may be impossible to determine an appropriate set of canonical states of context 2) determining what information is necessary to infer a contextual state may be difficult and 3) determining an appropriate action from a given context may be difficult, because the kind of responses that people expect from a context-aware application are very situation-dependable.

So here I must agree with Suchman when she says that human beings don't plan anything: they just react. We do, we are opportunistic creatures who optimize the path to accomplish a task that will require the least effort.

From his brief description of the "context theories", general perceptions arise:

- interactions evolve over time, therefore the historical character of activity is relevant for understanding the interaction

- people have particular views of the same things (or places, or practices) so the individual's history counts when analyzing a context

- artefacts are the media by which people achieve goals, they are not (or shouldn't be) the goal itself


We should aim to design interactions, not applications.