The knowledge that must be created in software design involves two dimensions: mental and material. The mental dimension is a shared understanding of how a new computer system will change existing practice. The material dimension is an artifact - either the system itself, or some description by which a system can be built.
This section presents constructionism as an epistemological framework concerned with building things, both in the sense of building understanding and building artifacts. This framework will serve as a basis for investigating the question of: what does it mean to construct knowledge in software development?
Figure 4.2. A Model of Knowledge Construction
Artifacts and Understanding are two dimensions of knowledge. Knowledge construction occurs is when artifacts and understanding co-evolve.The following sections will describe the model of knowledge construction presented in Figure 4.2. Like the two hands in Escher's Drawing Hands (see Figure 4.3), artifacts and understanding in this model are at once the product, as well as the producer, of the other. Artifacts produce understanding through interpretation. Understanding produces artifacts through representing, which in this thesis denotes the act of constructing an explicit representation, or artifact. Iterations between representing and interpreting cause both artifacts and understanding to evolve. This co-evolution between artifacts and understanding is knowledge construction.
Figure 4.3. Drawing Hands, by M. C. Escher
Knowledge Construction is a way to think about the relationship between constructing artifacts and constructing shared understanding among people. Like the two hands in this picture, building artifacts and building shared understanding each seem to give rise to the other, and at the same time, be the product of the other.
In the constructionist paradigm (which encompasses both cognitive and philosophical ideas of constructionism), our understanding of the world is fundamentally tacit. Normally, we are not conscious of these understandings. When we stop to think, however, we bring our understandings into the open were they can be examined. To become aware of understandings by thinking about understandings is to make them explicit.
Stahl introduced a theory of interpretation as the transformation of knowledge through a progression of forms, from tacit to explicit, to external, to codified [Stahl 1993]. The act of transforming tacit knowledge to explicit is interpretation, as is transforming explicit knowledge in the head to external knowledge in an artifact. The product of each of these transformations is more explicit and precise, but also further removed from the original and fundamental tacit understanding. An important aspect of Stahl's theory of interpretation is that understanding must be tacit to be operative. Therefore, interpretation produces temporary transformations of understanding toward the explicit, but when we stop thinking about what we are interpreting, it returns to tacitness. Of course, the experience of interpretation can change our tacit understanding.
The particular transformation of understanding between tacit and explicit (in the sense of becoming aware of something) is called a breakdown [Ehn 1988; Winograd, Flores 1986]. As described above, in the constructionist paradigm our everyday operating knowledge is tacit and we act in an expert mode - without explicitly thinking about our actions. We reflect "only when the spell of involvement is broken" [Ehn 1988]. Breakdowns cause us to become aware of something. They require us to stop and reflect upon what we know. We must come to grips with the cause for the disruption in our expertness, and find a way to get around the disruption.
For example, service provisioning representatives may be able to find and fill database fields for a familiar type of order without consciously thinking of the necessary steps. But when a customer requests a new type of service, this expertness may break down, requiring the representative to consult external documentation or to devise a new way to provision the requested service.
Breakdowns trigger interpretation and the construction of new understanding. Through interpretation, a new appreciation of a situation is formed, based on our previous background knowledge and the situation in which interpretation occurs [Stahl 1993].
Papert's notion of constructionism builds upon this constructionist framework, but he has a special concern with the role of physical artifacts, or objects, in thinking. Papert writes that "construction that takes place in the head often happens especially felitiously when it is supported by construction of a more public sort `in the world' . . . Part of what I mean by `in the world' is that the product can be shown, discussed, examined, probed, and admired. It is out there" [Papert 1993].
Papert argues that we learn through interacting with artifacts - that we create an understanding of the world by creating artifacts, experimenting with them to see how they work, and modifying them to work better. Since artifacts are in the world and can provide an anchor for understanding among a group of people, they can also address a problem that faces the constructivist cognitive scientists - namely, if we construct knowledge as an individual act, how can understandings be shared?
Objects-to-think with are cognitive artifacts that provide a link between sensory and abstract knowledge, and between the individual and the social worlds. Papert's example of such an object-to-think with are the sets of gears that fascinated him as a boy, and that gave him a concrete way to think about the more abstract qualities of ratios in mathematics.
Don Norman writes about two modes of interacting with artifacts that correspond with Papert's sensory and abstract: experiential and reflective [Norman 1993]. In the experiential mode, information is perceived and acted upon with no apparent effort or delay. The reflective mode requires "mental effort to think of and contrast the various courses of action." In experiential mode, objects and knowledge are ready-to-hand (i.e., invisible and taken for granted), while in the reflective mode the world becomes present-at-hand (available for inspection and reasoning).
Experiential artifacts allow us to interact with the world. They provide information that enables us to interpret a situation through our perceptions. The danger of experiential artifacts is that they don't in themselves provide us with knowledge - instead they provide us with information that is tacitly interpreted. The usefulness of experiential artifacts is that they can trigger breakdowns that surface tacit understanding. When what we perceive is different from what we tacitly expect, a breakdown occurs, and the cause of this breakdown is brought to the surface where it can be interpreted and knowledge can be constructed.
Reflective artifacts are also interpreted, but they are more explicit in the knowledge they contain. The danger of reflective artifacts is that they leave things out (mostly things we don't know how to represent, which is not the same as things of little importance [Norman 1993]), and thus the information we reason with might not reflect the world. This danger notwithstanding, reflective artifacts are the most precise and common means of communicating explicit ideas.
Constructionism implies a process of building, both in the sense of building understandings of the world and building artifacts. Artifacts provide a conceptual anchor for shared understandings. Shared knowledge is constructed when artifacts and shared understanding are coupled through cycles of representing and interpreting.
A dilemma for constructionism is that, on the one hand, it postulates that our understandings of the world are essentially private - our knowledge is built upon our prior experiences. On the other hand, it is clear that people are able to build a shared understanding of the world that enables them to cooperate and communicate.
In most cases this shared understanding comes from participating in a particular practice over extended periods of time. In the case of software development, the time frame for building a shared understanding is considerably shorter, and the nature of the shared understanding must be more explicit. In system development, developers cannot be expected to learn the user's practice. In the service provisioning domain, for example, this would require months. And service provisioners cannot be expected to become software developers, either.
The shared understanding necessary for software development does not require a complete overlap in understanding, but rather a sufficient overlap to communicate a shared vision of what the new system should be. This understanding must be made external so that stakeholders can discuss and refine it, and ultimately so that a computer system can be built that embodies it.
Because interpretation is essentially an individual act, understanding between people is never absolute, but it need not be absolute for people to perform coordinated activities and to communicate. Artifacts that are created to externalize knowledge are representations. Representations, like analogies, capture certain features and ignore others. A representation is never a complete duplication of one's understanding - this is a source of its strength as well as its weakness. The shared understanding representations support is "not a fixed relationship between a representation and the things represented, but is a commitment to carry out a dialog within the full horizons of both speaker and hearer in a way that permits new distinctions to emerge" [Winograd, Flores 1986]. Thus, representations serve as shared objects-to-think-with that enable a shared understanding as well as provide the foundation for new understandings. As representations evolve, so do the shared understandings they enable.
|Previous||Home||Contents||Next - 4.3. Representations For Mutual Understanding|