Content-type: text/html ~ Stephen's Web ~ Design, Standards and Reusability

Stephen Downes

Knowledge, Learning, Community

Jul 30, 2003

In 'Connecting Instructional Design to International Standards for Content Reusability' Michael D. Bush (2002) emphasizes two major points: first, that standards are essential in order to achieve widespread educational content interoperability, and second, that efforts are underway to associate existing standards efforts with principles of instructional design. His article is based in discussions at the ID2SCORM [Footnote 1] conference held in Utah last year, where participants adopted and forwarded both points.

The argument in support for world-wide standards for learning content is a familiar one. Bush cites examples from the early days of the industrial age, such as the need for consistent fire hose connections, or the need for a common railroad track width, in order to show that a failure to adopt standards is both prudent and efficient. The adoption of e-learning standards, he suggests, is also consistent with existing efforts to design instructional materials that could be used by many colleges and educational institutions. Hence he and his colleagues turned enthusiastically to SCORM, a standard developed by Advance Distributed Learning for the U.S. military, and hence, widely used by both universities and corporations.

A certain degree of vagueness exists within SCORM, however, and Bush notes that the standard leaves questions such as the definition of a 'learning object' and the appropriate granularilty of such materials unanswered. For example, to cite an oft-quoted difficulty, sharable content objects (SCOs) within SCORM are expected to be independent of learning contexts, and yet, at the same time, be used in learning. The challenge, he suggests, is to create media that are at the halfway point between individual media objects on the one hand and a full course on the other.

Though learning design was deliberately left out of SCORM, he observes, there is some committment to include it in the future, and it is with this in mind that the participants of the ID2SCORM conference convened. Learning design has been approached in other standards efforts, most notably Europe's Educational Modelling Language and the Instructional Management Systems's Learning Design specification. But these efforts, he argues, are missing support from recognized instructional design experts, who, collectively, would be able to describe how content and instructional strategies may be represented, how popular learning theories may be expressed in software, and how interactions between learners and instructional systems may be represented.


In what is a bit of a misuse of the term, instructional theorists have been talking about the instructional 'contexts' of learning objects. One might think of an instructional context as the manner or way in which a learning resource will be used to foster learning. An instructional context thus defines the relation between resources with each other, such as the manner in which they are sequenced or presented to the learner. Alternatively, an instructional context may define the role that a given resource plays in a learning scenario: it may be an illustration, an example, an explanation or an exegesis, for example.

In the design of learning objects, and in e-learning generally, the definition and location of the instructional context becomes a central question. For an object to be used in learning, it must be used in some specific way, and arguably, it is not a learning object (as opposed to a mere content object) unless the definition of the object in some way also describes the manner in which it is to be used. As various commentators have argued, a mere picture is not a learning object because there is no instruction inherent in the picture. Presented merely with the picture, one has no real idea what to in order to use it to advance learning. The presentation of a picture, therefore, must be accompanied with some context. The context would describe what is to be learned from the picture, and possibly, suggest ways in which this learner must be accomplished.

Instructional design theory is in a nutshell the study of instructional context. It considers different ways of presenting different types of materials, and different uses to which these materials may be put, in order to foster learning. Although instructional design is typically practiced in concrete form, as in the actual design of an online course or program, the theory approaches this topic in the abstract, suggesting methodologies that may be used in a wide variety of circumstances. It is understandable, then, that one would see a natural fit between learning objects, which are supposed to be reusable, and instructional design principles, which are also supposed to be reusable.

At its simplest level, instructional design theory may be thought of as a series of instructions to teachers. "First, present the learning objectives (which are grounded in associated competencies," the theory might say. "Then provide some concrete examples (preferably rooted in the students' own experiences or culture). Following that, draw out the generalizations inherent in the concrete examples and show, through a demonstration, how these principles may be applied in a larger number of circumstances. Provide some exercises in which the principles are applied in additional circumstances. Finally, depending on the educational level desired, test for the student's ability to remember the examples, to remember the pronciples, or to apply them in novel circumstances."


Of course instructional design theory in practice is much richer than the simple example provided. It elaborates on the types of materials appropriate for each activity. It shows how the activities flow from one into the next, and how an internal consistency is maintained from the initial assessment of learning needs through to the metrics employed in the final testing process. But most importantly, a more mature approach to instructional design will inform the designer of means and methods to anticipate, and design for, variable circumstances.

For example, given the same instructional material, one student may successfully acquire the concepts and demonstrate the associated competencies, while another may fail. At the conclusion of the test, or at the conclusion of intermediate assessment stages, therefore, a point of decision is reached. If the student has failed to acquire the previous material, then it makes no sense to proceed to new material. Minimally, the student must loop through the instruction and try again. Consequently, a major component of any theory of instructional design will revolve around the definition, capture and use of interaction between the student and the instructional material.

Another major component of instructional design involves prior assessment or pre-testing. Before any instruction is attempted, the student's ability to demonstrate his or her competencies is assessed. Based on this assessment, a selection of a set of appropriate instructional materials may be made, so that the student's time is spent in the acquisition of new knowledge, rather than the mere rehashing of previously acquired knowledge. Alternatively, an instructional system may query the student regarding what sort of learning he or she would like to undertake, and based on the student's selection, present the appropriate instructional content. While the previous sort of design created loops, these designs create branches.

Another type of decision revolves around how the learning content is to be presented. The same content may be presented in different ways. The medium employed may vary: a student may be asked to read some text, to listen to an audio tape, to view some pictures or to watch a video. The langauge of the presentation may vary, from very simple for young learners or those just learning a second language, to very advanced for the adept. The pacing of the presentation may be speeded or slowed. The manner in which the presentation is approached - expository, explanatory, descriptive, argumentative - may be varied. Any of these decisions may in turn be based on properties of the material (is it easy or difficult, concrete or abstract, cognitive or affective) or on properties of the learner, sometimes described as their 'learning style'.

The point here is that instructional design may be characterized as a set of decisions regarding the type and order of the instructional content to be presented. Whether the instructional content be, in one instance, a classroom discussion revolving around a certain theme, or in another instance, a series of self-study exercises, or in another instance the reading and summary of a body of text, in each case, an argument needs to be made showing why this material, in this context, for this student, ought to be presented. And instructional design theory, in its grandest (and most futile) sense, is the advocacy of universal principles by which such decisions may be made.

This diagram (Anderson, 2002) is typical of the model (the terms may change, but the principle remaims the same). Objects and strategy, joined, yield learning:

Diagram of learning objects


The question of which educational resource (whether it be a knowledge object, learning object, or whatever) to use in a given instructional setting appears to be an open ended one. But in fact, the range of options is surprisingly narrow. Even in a world of a million learning resources, if we know what topic, level, lesson and language we are using we could narrow the field of possible candidates to a dozen or so. It becomes possible, even with such a crude characterization, to imagine simple rules for the selection of material.

This is, in fact, the approach taken in learning design. The IMS Learning Design specification, for example, is characterized in terms of what are called 'condition-action' rules. The 'condition' is a specification of the values of one or more properties in the learning environment. The action is the specification of what resource to launch (or what task to undertake) should the statement contained in the condition be 'true' at the time of assessment. At any given decision point, therefore, a set of conditions is evaluated, and depending on the result, a given action is undertaken.

This logic may be seen clearnly in the IMS Learning Design specification. Consider the following extract (edited for clarity) from one of the examples:

We can see that in the 'if' statement, we test to see whether the hazards lesson has just been completed. The action that follows is the showing of the next bit of instructional content.

The bulk of the Learning Design specification is devoted to clearly defining the sorts of conditions that could obtain. This includes the description of roles, the description of an environment, and the creation of certain properties. This is the core of learning design:

properties and conditions, and notifications are required. Levels B and C of the Learning Design Specification provide these. Properties, specified at Level B, are needed to store information about a person or a group of persons (role). So for a student its progress may be stored, perhaps in a dossier; for a teacher information on papers graded may be stored. Conditions, also part of Level B, constrain the actual evolution of the didactic scenario. They are set in response to specific circumstances, preferences, or the characteristics of specific learners (e.g., prior knowledge). An example of a condition would be 'when the learner has learning style X, present the activities in random order'. The idea is of course that randomness allows the student to freely explore the materials. Notifications, specified in addition to the properties and conditions of Level B at Level C, are mechanisms to trigger new activities, based on an event during the learning process. For instance: the teacher is triggered to answer a question when a question of a student occurs; or the teacher should grade a report, once it has been submitted. (IMS, 2003)
(It should be noted that when the IMS learning Design specification uses the word 'condition', it is referring to the entire conditional statement, and not merely to the antecedent ('if') part of the statement.)

What learning design buys the instructional designer is flexibility. A lot of flexibility. Given the same set of learning objects, for example, two very different types of instructional material may be created by using two different learning designs. Even within a single learning design, the presentation of material for a given student may be changed by changing the role of the student. And even within a single role, the presentation of material is changed depending on the student's interaction with the material, or by instructor intervention.

But it buys this flexibility at a price. And it is arguable (and I would argue) that the price may be too high.


We begin to see where Learning Design goes astray when we consider the reusability of a given learning design.

By hypothesis, let's consider a simple learning design, which can be characterized informally as follows:

Show the learner a learning object. Then give the learner a test. If the learner passes the test, then show him a new learning object. Otherwise, show him the first learning object again.
Without getting into the details of how to express this design in IMS Learning Design, we can see that it is a reusable learning design. It is reusable because we have not specified which learning objects or tests are to be used. The design, as expressed here, is an abstraction. It could be used by any number of designers in any number of settings. In fact, it is possible to see just this sort of description offered in any number of learning design documents.

How this would work is also straightforward. The instructional designer would select, from a repository, one or more learning objects. The instructor would select a learning design. These would then be bundled together to form a learning package. The completed entity would then be stored or shipped to the eventual end user. It is possible even that some designers may comtemplate the placement of two separate learning designs in the same package, to give the resulting package a degree of flexibility. Thus, the package, when run on an LMS, could, depending on (say) the student's learning preferences, select one of the other learning design to apply.

In practice, however, we see that the situation is not as it seems. Consider a more abstract representation of the same scenario.

We have three learning objects and a test (L1,L2,T) and a set of rules:

  • Show L
  • Show T
  • If T(passed) show L, else show L

This formalization is a bit of a charicature, but the question pops out: how do we tell the learning design which learning object to show at any given point?

No doubt this formulation has occurred to the reader:

We have three learning objects and a test (L1,L2,T) and a set of rules:

  • Show L1
  • Show T
  • If T(passed) show L2, else show L1

And, indeed, this would work. But, importantly, it assumes that the learning objects have somehow been named in the learning design.

As we look again at the examples offered in the IMS specification, we see that this is exactly what happens. Look again at the sample XML markup:

As we can see, in both the condition and the action, a specific learning object is named.

And if a specific learning object is named, then the learning design is not reusable. It can only be used in that precise context where the learning objects in question were (from the example) 'P-Hazards-Lesson' was the last lesson, and 'LA-knowledge-test-components' is the current item to show. This learning design cannot be used anywhere else except in this specific course.

It is tempting at this juncture to consider various approaches to the resolution of this dilemma. For example, instead of referring to learning objects by name, perhaps the learning design could refer to them in order. Thus, the conditional statement might read "If the student has passed the first test, then show him the second learning object; otherwise, show him the first." But this merely pushes the problem back one step: at some juncture there would need to be a specification of the order of the learning objects, and this order would have to refer to the learning object by name.

Another possible approach is to try to specify the rule by appealing to some property of the learning object. For example, the rule might read, "If the student passes the level 2 test, then show the student the level three object, otherwise show him the level two object." But this merely pushes the problem of specification into the learning object itself; the designer of the learning object would have to know that this is a 'level two' object, with respect to this learning design, and code it intop the metadata. If the designer referred to an external schema or ontology to specify level (or some other property) then the learning design would be tied to that particular schema or ontology, and once again, would not be reusable across contexts.

There is, I submit, no way out of this problem. In order to use a learning design with a set of objects, the learning design must specify the objects to be used, and if the objects to be used are specified, then the learning design is not reusable.


Having described the problem, let me now add a few words about the only way in which a learning design can be used. In a nutshell, learning design requires a designer.

The following scenario may be imagined. A learning designer is working in an integrated development environment. She conducts a search, based on a prior knowledge of the topic to be studied and the level of the student. She then selects from a list of learning designs, and then (maybe using drag and drop) moves each of the selected objects into the (blank) learning design template. The resulting product is a completed learning design that specifies each object to be used. This is then packaged and shipped.

It sounds like an appealing prospect, and no doubt software designers are already implementing this methodology. However, it must be noted, first, that the designer must already know the design, and second, as noted previously, that the word of the designer cannot be re-used. It's a labour-intensive one-off production, and therefore hardly an advance over having the designer create a customized course from scratch.

The designer needs to know the design in advance because she must know where to place each learning object into the design. Thus, for example, she must know that L1 comes before L2 in the sequence of events. This means that not only is the design process labour intensive, also, it must be implemented by a subject matter expert. And it is still not reusable. Moreover, the more detailed the design, the more the designer must know about not only the field, but also about the pedagogy behind the design.

So while it is clear that the learning design specification can indeed be used to create learning packagaes, it should be clear that:

  • The learning designs themselves are not reusable in any meaningful way
  • The completed package is not reusable at all
It seems, therefore, that the project envisioned by the ID2SCORM group is doomed to failure, not in the sense they they could not accomplish the task - of course they could, as just described - but in the sense that the task, once accomplished, will not have produced a system or methodology of any meaningful value.


In my opinion, this is a defect, not in the intentions and deliberations of the ID2SCORM group, but a flaw inherent in the concept of learning design itself.

Expressed generally the flaw is this: Learning design and reusability are incompatible.

Design requires specificity, and specificity prohibits resuability. Or conversely, Reusability requires generality, and generality prohibits design. It's an all-or-nothing proposition. If you want to tell your students what to do and when to do it, then you must tell your students what to do. But the minute you tell them what to do, you have precluded the possibility of telling them to do something else.

The director of a play cannot on the one hand instruct an actor to say "To be or not to be" while at the same time leaving it open for the actor to say something else. And if the director leaves the options open, then the actor does not know whether to say "To be or not to be" at any given time.

So where to now?

That is, of course, the subject of another paper. But essentially, the solution requires the embrace of one or the other of the horns of the dilemma. And that is what we are seeing more and more in the field.

As some commentators have already noted, we are beginning to see less and less talk of re-use in online learning. Indeed, as Patrick Lambe suggested in an entirely different context, what we are after is not reusable objects, but disposable ones. This, it seems to me, is the approach favoured by more and more institutions and corporations, as they begin to look at an instructional design system, not as a means of reusing objects, but as a means of producing them to be used once, then discarded.

In my own work, I have embraced the other horn of the dilemma. That is, I believe that - for reasons of cost and accessibilty - reusability is still an attainable goal. However, it means that, in order to achieve this goal, the idea of reusable learning design must be jettisoned. So, in a sense, what I am arguing for is disposable design, that can be created once, as needed, and then discarded, never to be used again.

Such an approach to the dilemma, of course, shoots an arrow straight into the heart of the discipline known as instructional design. It suggests that, at some fundamental level, instructional design is not about considering different ways of presenting different types of materials, and different uses to which these materials may be put, in order to foster learning. It is not like presenting a series of instructions to teachers (now some people may say that this was the case all along, to which I must ask, what is the IMS Learning Design specification, then, if not that).

In my view, the difference between the two horns of the dilemma is the difference between writing a play and creating a game. It is the difference between telling people what to do and when to do it, and creating an environment where people decide for themselves what to do and when to do it. It is the difference between requiring a director and requiring a coach. It is the difference between giving a person directions to the Forum and giving them a map of the city and letting them choose their own route.

And - crucially - it is in my mind the difference between the way learning was and the way learning will be.

Footnote 1: The ID2SCORM conference home page is - incredibly, a Google search for the conference (as of this writing) reveals exactly seven results. Only one of the conference presentations was online. Bush's article, to which this article responds, is not available online.

Anderson, Thor. 2002. Efficient Development through SCORM Standards. ID2SCORM.

Bush, Michael D. 2002. Connecting Instructional Design to Internationational Standards for Content Reusability. Educational Technology, November/December, 2002.

IMS Global Learning Consortium. 2003. IMS Learning Design Best Practice and Implementation Guide. Version 1.0 Final Specification.

Lambe, Patrick. 2002. The Autism of Knowledge Management.

Stephen Downes Stephen Downes, Casselman, Canada

Copyright 2024
Last Updated: Jul 24, 2024 8:53 p.m.

Canadian Flag Creative Commons License.