Why All the Noise About Functional Design?

OK. So why all the noise about functional design?

Because:

• It's closer to what designers really do anyway
• It makes it possible to link theory in a more direct way to a design
• It allows the designer to tailor the process to the needs of the project
• It supplies an alternative to the traditional approach
• It corrects some of the shortcomings of the traditional approach

Functional design actually fills in a blank spot in the ADDIE/ISD approach. If you look closely at the many versions of ADDIE/ISD, you will notice that the models abandon you just at the point when you need the most insight and guidance. In this respect, the ADDIE/ISD models remind me of the "...And then a miracle happens" cartoon from the New Yorker. Most of the ADDIE/ISD processes get you to the point where you are ready to make the most substantive and detailed decisions but do not tell you how this can be done. Functional design takes up at that point, focusing the attention of the designer on the details of the artifact and challenging you to think inside of each functional area of the artifact to produce a design that is detailed, yet coherent.

Functional design also deals with an inconsistency found in virtually every ADDIE/ISD description: the mixing of instructional theory with design theory. Stop and think about it: when the design model tells you to perform task analysis, it is committing your design to a particular bias toward task-type instruction. Sometimes this is not  appropriate to the type of subject-matter you need to instruct, and this apparent mismatch between what you sense you should do and what the model tells you you should do leads to confusion and frustration. 

There is another example I could cite of where the current ADDIE/ISD models create this kind of subtle bias: in the recommendation to use taxonomized instructional objectives. I know that I am saying heresy at this point, because one of the principles that brought instructional design out of the dark ages was the one-objective-one-strategy concept promoted by Gagne, Merrill, Bloom, and others--an idea so thoroughly ingrained in instructional design today that many designers would not know what to do without it.

But the creation of classes of objectives (and I don't argue that they are not comfortable and useful--I've used them many times myself) tends to gravitate the thinking of the designer in the direction of the fragmentation they lead to, with no countervailing force to bring the designer's thinking back toward integration of the fragmented learning into fluent and competent wholistic performance. 

What remedy do I suggest? The remedy of acknowledging that there is a body of design theory that can guide a designer in the making of designs and that there is also a body of instructional theory that supplies the content of those designs. This implies that I can talk about how to design without making assumptions of what the design will look like. Clint Rogers and I make this point in the new Green Book III chapter we titled "The Architecture of Instructional Design".

This is a new idea for instructional designers. Things have not been taught this way up until now. One of the fall-outs of that has been the neglect by instructional designers of a large body of research into design across the disciplines. 

So anyway, that is the reason for all of the noise about functional design. It is a design theory, not an instructional theory, but it shows how instructional theory can be more directly incorporated into instructional designs.

Functional Design and ADDIE/ISD

Functional design and ADDIE/ISD inform each other and are capable of working in a complementary way. Probably the two themes most important in describing the relationship are sequentiality and granularity.

Sequentiality refers to the issue of design decision order. Designs are large and complex enough—even the simplest ones—that they cannot be created in a single stroke. Designing is decision-making process, and therefore some decisions have to be made before others. ADDIE/ISD models specify a general order of data gathering and decision-making processes. Logically it does not make sense for some of these processes to come before others. At the same time, ADDIE/ISD models have been criticized for their tendency to promote “waterfall” thinking, which tends to be linear and which prevents the designer from anticipating the outlines of designs early-on: frustrating clients and designers alike.

Functional design orders design decisions according to project-specific patterns in response to constraints imposed by the problem’s context.  ADDIE/ISD models represent an idealized case in which it is seldom made clear how prior constraints impact the idealized model’s processes. Functional design assumes the existence of prior constraints for every design problem that force the designer to re-assess the order of decision-making. Design order in functional design is determined by the “next most critical” rule for selecting decisions to be made, assuming that certain decisions have been made or are strongly indicated even before the project is launched, creating the “seed” from which the design expands outward in increasing detail.

Granularity refers to the degree of guidance provided to designers by a design approach. Functional design and ADDIE/ISD probably differ to the greatest degree with respect to granularity. This is because functional design operates within the same context as ADDIE/ISD, overlapping to some extent some of the ADDIE/ISD processes. However, functional design deals with the concerns of design within this context at a much smaller granularity and therefore scaffolds design efforts to a much higher degree during the critical phase where the design itself is being formed.

Functional design also recognizes that design decisions begin during what is normally recognized as a design phase but that they also continue to be made throughout what is normally called the development phase. Functional design therefore considers all decisions, from the highest level down to the lowest, to be design decisions. What are normally considered production tasks are therefore treated as design tasks, and the existence of a development phase is de-emphasized.

Functional Design V

Functional design supports the simultaneous consideration of parallel alternatives—design hypotheses—and reduces the sequentiality of decisions. It encourages the evaluation of clusters of decisions which must work together to create design unity. It enables designers to hold off firm decisions on a cluster until the impact of the decisions on cost, skill, time, and tools has been determined. A functional designer will most likely advance the design within many layers at once through interdependent decisions spread across multiple layers. The method of proposing hypothetical designs and testing them is described by Schön as a design “conversation”. Just as Schön’s domains of an architectural design define the loci of individual design decisions, the layers and sub-layers of an instructional design localize the attention of the instructional designer without obscuring the integrity of the whole.

Functional Design IV

Since it can’t be predicted what constraints will come with the problem, and since different decisions challenge different parts of a process model, functional design does not specify a design order. Instead, it follows next critical decision order. Gibbons (2009) defined several factors that may individually or jointly determine the next critical decision for a given project and moment within the design process:

• The decision most constrained by the latest previous decision.
• The decision most constrained by external factors, such as skill availability, infrastructure, resources, etc.
• The decision that best advances the central discipline of the primary generator.
• The decision that best takes advantage of an opportunity afforded by the latest previous decision.
• The decision that creates the most options for later decisions.
• The decision for which there is the most supporting data from the analyses of the target population and of the instructional context.
• The decision that represents the next highest quality priority.
• The decision that most directly addresses a major client criterion or desired feature.
• The decision that best leads to the satisfaction of an innovation goal.
• The decision that is most necessary to the implementation of a chosen theoretical position.
• The decision that responds to the latest prior decisions in other layers of the design.

Functional Design III

As decisions are made, either in singles, in clusters, or in order to serve an operational principle, each new decision imposes new constraints on future decisions, redefining again the options left in the problem solving space, making certain future decisions imperative, and shifting priorities for the next decision. This cycle of progressive placement of constraints continues until the minutest decision has been made firm. Functional design is a natural companion of design layer theory. Initial constraints on a design problem affect different layers. One project may require a particular medium to be used; another may require a particular social setting. Constraints influence the options available within different layers of the design. Moreover, constraints on one layer ripple effects to other layers, constraining decisions within them. 

Functional Design II

It is called "functional design" because it employs layers, and layers correspond with functions of the artifact being designed. I propose that instructional designs have layers that correspond to:

• the representation function (provide sensory experience for the learner)
  
• the control function (give the learner a way to respond to the sensory experience)
  
• the message function (construct the individual messages of the instructional conversation--the ones that are to be represented)
  
• the strategy function (determine high-level strategic moves that support learning and drive messaging)
  
• the content function (supply content in appropriate form to the strategic, messaging and representation functions)
  
• media-logic (execute all functions)
  
• the data management (record data from the instructional encounter).  
  

Functional Design I

I couple the use of design layers with an approach to design process that I call functional design (Gibbons, DESRIST 2009, see http://prezi.com/58265/edit/ ). Functional design is based on the principle that a designer can make design decisions in any number of orders. It supplies the designer with a guideline for determining the order that follows the best strategic sequence for the particular project.

Most designers recognize that design problems come to the designer with diminished degrees of freedom—with constraints imposed by goals, resources, client desires, and time. It is generally recognized that fixed-process models are not sensitive to this fact of life and that there are generally no guidelines for adapting the generic process to a specific project. Further complicating things is the fact that different projects place different value on full-featured models due to time, staff, or client interest restrictions.   

To deal with this, functional design removes order constraints from decision-making during design. Design decisions can be made in any order. However, functional design also incorporates the principle that each decision, firm or tentative, imposes constraints on future decisions—creating some new decisions while cutting off others. This leads designers to hold decisions tentative where possible until correlation has been made across all areas of the design to ensure coherence and unity in the whole design.

Learning versus Instructional theory

It is a general but, I believe, mistaken idea that learning theories can be applied directly into instructional designs. I think there are translations necessary to convert a learning theory into an instructional theory that can be applied directly. By looking more closely at theories, you can see that the ones that tended to be most easy to apply to designs are those that considered themselves "instructional" theories rather than "learning" theories. I think the key difference is that a learning theory posits from what is exists or is thought to exist. Instructional theory posits from what can be made to be there. It supplies terms for elements that can be created.

Two kinds of theory?

To make sense of these ruminations on layered decomposition of instructional design problems one has to accept a distinction between two kinds of theory: instructional design theory and instructional theory. To put this in general terms, it means for designers accepting the existence of two kinds of design-related theory--one kind pertaining to how things are designed, and one pertaining to what kinds of structures those designs might contain. The first of these can be called a design theory; the second can be called a domain theory. 

Donald Schon (umlat over the "o") describes domain theories in some detail in "Educating the Reflective Practitioner" (Jossey-Bass, 1989). He gives an account of a tutor helping a student solve an architectural problem. In the process, Schon reveals his view that different parts of the problem become most critical at particular times and that each sub-problem constitutes a "domain" which has its own terminology, principles, and practitioners.

At the same time as we acknowledge domain theory as a kind of theory required by designers, we can also see that designers draw on another type of theory--design theory--which is theory about how something is or can be designed.The subject of Herbert Simon's "Sciences of the Artificial" was this kind of theory.

The two types of theory relevant to a designer may puzzle the scientifically-oriented person, who is used to there being only one kind of theory--and that an explanatory , not a synthetic, kind. What answers this puzzlement is the realization that designers use theory, but that the theory is not scientific theory: it is design-related theory. Moreover, the identification of two kinds of design related-theory (with the expectation of even more being identified) means that design-related theory is a different species of theory with different properties and different uses: technological theory.

Seeing more deeply into the heart of design problems

Studies of designers and the process of designing have shown that expert designers perceive the problem in more sophisticated ways than novices, seeing different conceptual categories at a deeper level of organization. As the instructional design field continues to mature, its problem-solving constructs and processes can also be expected to evolve and consider deeper structures. Myself and Clint Rogers have proposed a theory of design layers as a way to describe the architecture of the deep structure of instructional designs (“The Architecture of Instructional Theory”, in Reigeluth & Carr-Chellman, 2009, Instructional-Design Theories and Models, Volume III, Routledge). The theory describes several levels (layers) of structure that can be used to decompose and solve instructional design problems. The appeal to a layered design architecture has been made in other design fields but is new to instructional design.

Approaching design problems differently

Instructional designers have become accustomed to decomposing their design problems in terms of general phases or steps of problem solving. This approach, however, is but one of many possible avenues to simplifying problems for solving. Design fields other than instructional design have employed approaches that break the design problem down in terms of the characteristic functions of the artifact. I suggest we think in terms of a layer theory of design architecture that leads to a method of decomposing design problems by the generic functions of the class of artifacts being designed—functional design.

What's in a design problem?

Instructional design problems aren’t really single, monolithic problems, and instructional designers don’t solve design problems all at once. Design problems consist of many different sub-problems which are identified by decomposing the large problem into smaller, more solvable problems. The manner in which designers do this decomposition makes all the difference.