My problem with the formal models of instructional design (e.g. ADDIE for process), is that most are based upon a flawed premise. The premise is that the world is predictable and understandable, so that we can capture the ‘right’ behavior and train it. Which, I think, is a naive assumption, at least in this day and age. So why do I think so, and what do I think we can (and should) do about it? (Note: I let my argument lead where it must, and find I go quite beyond my intended suggestion of a broader learning design. Fair warning!)
The world is inherently chaotic. At a finite granularity, it is reasonably predictable, but overall it’s chaotic. Dave Snowden’s Cynefin model, recommending various approaches depending on the relative complexity of the situation, provides a top-level strategy for action, but doesn’t provide predictions about how to support learning, and I think we need more. However, most of our design models are predicated on knowing what we need people to do, and developing learning to deliver that capability. Which is wrong; if we can define it at that fine a granularity, we bloody well ought to automate it. Why have people do rote things?
It’s a bad idea to have people do rote things, because they don’t, can’t do them well. It’s in the nature of our cognitive architecture to have some randomness. And it’s beneath us to be trained to do something repetitive, to do something that doesn’t respect and take advantage of the great capacity of our brains. Instead, we should be doing pattern-matching and decision-making. Now, there are levels of this, and we should match the performer to the task, but as I heard Barry Schwartz eloquently say recently, even the most mundane seeming jobs require some real decision making, and in many cases that’s not within the purview of training.
And, top-down rigid structures with one person doing the thinking for many will no longer work. Businesses increasingly complexify things but that eventually fails, as Clay Shirky has noted, and adaptive approaches are likely to be more fruitful, as Harold Jarche has pointed out. People are going to be far better equipped to deal with unpredictable change if they have internalized a set of organizational values and a powerful set of models to apply than by any possible amount of rote training.
Now think about learning design. Starting with the objectives, the notion of Mager, where you define the context and performance, is getting more difficult. Increasingly you have more complicated nuances that you can’t anticipate. Our products and services are more complex, and yet we need a more seamless execution. For example trying to debug problems between hardware device and network service provider, and if you’re trying to provide a total customer experience, the old “it’s the other guy’s fault” just isn’t going to cut it. Yes, we could make our objectives higher and higher, e.g. “recognize and solve the customer’s problem in a contextually appropriate way”, but I think we’re getting out of the realms of training.
We are seeing richer design models. Van Merrienboer’s 4 Component ID, for instance, breaks learning up into the knowledge we need, and the complex problems we need to apply that knowledge to. David Metcalf talks about learning theory mashups as ways to incorporate new technologies, which is, at least, a good interim step and possibly the necessary approach. Still, I’m looking for something deeper. I want to find a curriculum that focuses on dealing with ambiguity, helping us bring models and an iterative and collaborative approach. A pedagogy that looks at slow development over time and rich and engaging experience. And a design process that recognizes how we use tools and work with others in the world as a part of a larger vision of cognition, problem-solving, and design.
We have to look at the entire performance ecosystem as the context, including the technology affordances, learning culture, organizational goals, and the immediate context. We have to look at the learner, not stopping at their knowledge and experience, but also including their passions, who they can connect to, their current context (including technology, location, current activity), and goals. And then we need to find a way to suggest, as Wayne Hodgins would have it, the right stuff, e.g. the right content or capability, at the right time, in the right way, …
An appropriate approach has to integrate theories as disparate as distributed cognition, the appropriateness of spaced practice, minimalism, and more. We probably need to start iteratively, with the long term development of learning, and similarly opportunistic performance support, and then see how we intermingle those together.
Overall, however, this is how we go beyond intervention to augmentation. Clive Thompson, in a recent Wired column, draws from a recent “man+computer” chess competition to conclude “serious cognitive advantages accrue to those who are best at thinking alongside machines”. We can accessorize our brains, but I’m wanting to look at the other side, how can we systematically support people to be effectively supported by machines? That’s a different twist on technology support for performance, and one that requires thinking about what the technology can do, but also how we develop people to be able to take advantage. A mutual accommodation will happen, but just as with learning to learn, we shouldn’t assume ‘ability to perform with technology augmentation’. We need to design the technology/human system to work together, and develop both so that the overall system is equipped to work in an uncertain world.
I realize I’ve gone quite beyond just instructional design. At this point, I don’t even have a label for what I’m talking about, but I do think that the argument that has emerged (admittedly, flowing out from somewhere that wasn’t consciously accessible until it appeared on the page!) is food for thought. I welcome your reactions, as I contemplate mine.