Models and Theories in Human-Computer Interaction/Is This Technology Helpful?
Activity Theory (AT) is indeed a thought provoking theory. It is indeed a descriptive theory. There are many different ways the parts of the activity system may play out so I think of AT as a behavioral descriptive framework. Although, I have to admit, the chapter on AT in Carroll's was a bit of a hard read; it was very abstract, and the example of the Colored Petri Nets (CPN) did't help much given that the object as well as the outcome of the tool is beyond my knowledge. Therefore, I can say I only understand the basic idea of the theory at this point.
Two things about the theory I felt were powerful and very true according to my experience. One, a tool is only good according to how useful it is within the context of the problem space it is being used. In other words, a tool is a solution. It is very common in my profession of software development where the business will ask us to design a tool such that it does X, Y, and Z. Now, they are not explaining the problem space; they are asking for a solution. Meaning, maybe their is non-technological way of solving the problem, or maybe only a subset of what is being asked for makes sense. More than the problem space (the object in AT), there is the context of people and processes that surround the tool. This is articulated quite well by AT. And all of which play into how useful the tool will be and in implementing the best design. The worst thing that could happen is to introduce a tool that makes it harder for the subject to reach their objective.
Two, the notion of systemic contradictions (or tensions) within the activity system is very true. An example of this is when multiple tools are available to work on the same object (e.g., programming environments) but only a few of them are allowed according to the rules of the community (e.g., the organization). The tools are creating tension, such that the subject is not sure which tool to use given the rules in place. Now, what usually happens is either the developers will force the issue and the rules are changed or the owner's of the rules will remove the tools not allowed.
I love how this theory helps answer the question: is this technology helpful? It identifies the variables that participate in that answer.