Creative Destruction and Disruptive Innovation
Summary of Wayne Hodgins's talk at AACE E-Learn, Phoenix, Arizona.
The future of learning... it's not about predicting, it's about designing... we design our future.
Sobering thoughts: history is not kind. Think of the ice man. And ships, typewriters and photos. Almost no companies survive disruptive innovations. Remember Smith-Corona. When the plant closed, the president said, "This (the last typewriter) is the best product we have ever produced. But what we ended up doing is perfecting the irrelevant."
And very many groups can be in this process. They confuse activities and value proposition. Ask Smith-Corona. They thought they were in the typewriter business. But that was merely their activity. Their value proposition was helping people do do word processing. The examples multiply. Almost no ice manufacturing company survived into the age of refrigerators. If you confuse activities and value propositions, you are destined to be anhililated. Now 'education', 'training' and 'learning' are also activities. What's the value proposition in what we do?
Now consider things like the classroom. When something new comes along, the most common talk is about what will get replaced. But in actual fact, almost nothing got replaced. EDxample: when television came along, people predicted the end of radio. Things change - people don't sit for an evening listening to the radio. We keep it, though, in a changed form.
What if... the impossible isn't?
Consider: Personalized learning experiences for every person on the planet...? Just for me and just right: time, place, amount, medium, way, on demand, adaptive, in all forms, formal and informal, and not just online. Is this possible?
This dream isn't new. But what is new is that this is now possible. And what would it mean if it were possible?
The enabler's enigma ('enablers' - things that make things possible that were't possible yesterdat): enablers shift responsibilities from those who create things to those who use them, transferring responsibility to the implementers. Now consider how this impacts your planning. What if, during your planning, the impossible becomes possible? Personalized learning has become possible: the only question is, how do we get there?
The next big thing is: getting really small. This is already here (but not evenly distributed - Gibson). The notion of 'standardized uniqueness' - sounds ridiculous. But can be based on very small standardized components. Consider the building industry. 80-90% of any building is of pre-built components. That means 80-90% of your house that you're planning is already in a factory somewhere. The size of the components needs to be considered - not too small, not too large. Standardizing on screws, etc. is too small. Modular homes is too large. But door-units are just about right.
In our world: what we're looking for is to have the same kind of a dynamic assembly model possible - pieces in the form of single animations, images, etc., all designed to be assembled together. When buildings are built, they don't talk about who made it, they talk about requirements: you specify what you want, they return with something that fits your requirements. If we can help the learner (etc) be very detailed in specifying requirements, then we can start assembling.
This reasoning also applies elsewhere. For example, most code is written in 'objects' - this was supposed to result in mass customization of code. But this was messed up, because we insisted on packaging them in monolithic applications. You have to buy the whole big thing, and you have to find the pieces. The same with .Net and web services. The promise is that they will enable the same impossible dream. Web services promise the capacity to assemble something out of just the right pieces. So the idea is - take things down to very small, standardized componenets, and then assemble.
Another example: Dell computers. They define a computer as a set of components. They don't say, 'pick what model you want'. They say, 'build your own'. You describe what you want; they assemble it. So now, if it is possible to have mass customization at scale, and affordable, then it becomes a revolution.
Human competencies are following the same path. We thought of them as monolithic. But we are breaking these down into individual skills, knowledge pieces, experiences. Then we assemble a 'just right' collection of these skills, in order to put a person into a project (not a 'job'). Then we can define how to create a project team effectively.
Content model: smallest items are raw elements (images, etc) joined to create 'information objects'. The aggregation of these creates 'application objects' (such as 'learning objects'). Change 'content' to 'code' and you have the same model. Being able to assemble just the right pieces results in an application. These in turn are joined to create aggregate assemblies, such as lessons, and finally collections, such as courses. Now reusability increases the smaller the object, context-speciicity increases the larger the aggregation becomes.
Context is everything when it comes to learning - if there is no context, there is no learning. So the better I do at making something highly personalized, the better it is for you - but the less reusable it is. Thus we hear people say, we must reject the notion of reusability because there is no context. If we disassemble a house - the house is gone! There is no context, there is no learning, there is no relevance. But by taking the context out, we get a high degree of reusabillty.
How mucgh context is there in the material itself. Very little - context is created in the assembly, how we put it together. And if context comes from assembly, we can have our cake and eat it too.
Consider PowerPont slide shows - if you could make no changes at all, it's not reusable. Eg., it has a specific date on it. But consider slides - people can re-use previous slides in a current presentation. But have you ever tried to find a specific slide in your file system? Study - 28% of time is spent unsuccessfully searching for things. We should be desperately uninterested in searching, and deeply interested in finding. (Expertise location, web services location, competencies, code, equipment).
If we have small standardized components that are eminantly findable, which can then be intelligently assembled... then we have the possibility. Now - learning objects do not constitute learning. D'uh. Of course not. But it is part of the reason we are going to be successful at it.
Think in terms of a liquid, not a solid. Think of melting glass and metal. You get introduced into a whole new world when you realized it can be melted and poured into forms. Then we have personalized molds creating unique 'solid' solutions. Now we can say: solids are when things are actually useful, but we want to be able to turn things into liquids in order to allow reuse. The key is in being to reshape the mold. And we are getting to that point - some of the sessions here have started to talk about that.
The 'bow-tie' model: on one side, mass contribution - everybody can contribute. This doesn't mean turning everyone into authors. But consider phenomena like blogging, making PowerPoints, etc. Even if only a very small percentage is actually useful, then if we could find it, we could have mass contribution. At the center we have the work that produces the other side of the bow-tie, mass customization. It is the pieces in the middle which are worth your attention.
The middle: identifiers, metadata, onjects, taxonomies, and ontologies. (iMOTO). Metadata - it is the lack of metadat that is the problem of the web. And let's dispell a myth, that it has to be manually created. No. It can be generated automatically. RExample, go look at the last file you opened, and look at 'properties'. It's amazing how much metadata is created. We need to stop asking people 'what is the name of the file', 'what is your name', 'what language do you speak'.
Creative destruction: tearing down the old in order to create the new as it all 'gets small'. Think about, for example: filing, folders, and storage. I want to see a world of 'zero folders'. But I need some reaosnably good indexing and finding technologies. Consider storing wine: why would I store 'whites', 'reds', etc? It's a ridiculous pursuit. All I need is a globally unique wine identifier and a globally unique shelf locater. Now look at your hard drive. Don't you find yourself wanting to put the same file in more than one place? That's because the directory structure is broken. We want a dynamic, infinite number of categories, and a locator system.
This is where ontologies come in. I don't need to record the fact that it's a white wine if I record the fact that it's Chardonnay. Now think about content management systems. Are you wanting to manage 'files' or 'content'? Do you want to manage slide sets or slides, or contents in slides. In software, we will see in the next five years the obviation of the notion of a software application. You won't ask, 'what product is that'. Because it won't be a product: it will come from a collection of code. But of course the software companies are concerned - it won't come from one of us, it's not clear what the business model is.
The same with courses - the construct of courses won't go away, people will still have them, but the creation of content as courses will go away, because it makes no sense. Testing, credentialling... these will go the same way. People will look for (and be tested in) specific skills. It will go modular or micro. I want to know Robby. I want to know the precise skills I ill get if I bring himn onto the scene.
So let's look at the transformational technology:
- automated metadata
- pattern recognition
- expertise location
- recommender systems (TiVo, Amazon)
- Dynamic 'smart' assembly
Some examples: Amazon's search inside a book feature. Find one word in one page (because looking for 'books' is too modular. Consider 'TiVo'. Has a 'thumbs up / thumbs down' button. Metadata is stored in the system, recording your preferences. On the basis of this, it begins to look flor patterns in what you like or don't like in order to 'pick' programs to record that you never asked for. There are systems that are getting good enough to recommend poetry, songs. Imagine transferring this capability to our world - imagine being able to predict or recommend to a learner something that would be highly relevant in a particular context.
McLuhan: as new mediums come along, the previous medium becomes the content of the new medium. What used to be the medium - mtelevision, radio, movies - has become the content of the new medium. So the question is: do you want to be a medium or a media?
We have achieved the 'state of A' - it is possible to have anything, anytime, for anyone. But we need to equalize that distribution. But equally, having that capacity, we need to move to the next step - from 'any' to 'right' - this is the next challenge.
Get over it, get on with it. It's going to be messy. Consider food - food is messy when being assembled; we have chosen that we don't want 'food pills'. It could be neat, but it won't be, and we don't want it to happen. Also, plan for 'permanent whitewater'. There's no calm port in the river coming up. Don't wait - learn whitewater rafting.
So... what if the impossible suddenly is?
File finding programs - finding a single element on a PowerPoint slide, finding a single email - inference and pattern matching. Visual dictionaryes, combines a dictionary and a thesarus. It's the visualization - show it to me in context.
Comment: there is a focus on a tightly personalized content delivery... but what about the concept where people learn in a community. How does this mesh? We need to make sure we don't build something that isolates. Other people's projections of the future is where people are isolated, that this is where youth are headed. But today's youth are the most social - but it's still matched up with personalization. Think of things like Skype, or instant messaging.
But - what does this child need to learn? Is it going to be an individual prescription? Well, any definition of learning is going to be some sort of personalization. So we do need to go in that direction. The context is a soial one. It doesn't rule these things out. They have to be supplied. We need to differentiate between institutions. These are fixed assets. So we want to maximine the utilization of the asset. The most expensive airline seat is the one that takes off empty. So we want the individualization to be occuring still in the classroom. This is happening in commercial training centers - the notion of a class starting at 2pm next Tuesday is unworkable. When I show up at 2:17, that's when the class begins.
Comment: we are using broadband in Canada, and what we have found is that it's the cocialization piece that intrigues them. Now they've built bonds; the social piece was absolutely essential to making it happen. Still problems, the technology is clunky. Teachers are saying for the first time ever, I don't feel isolated in the classroom. Right. Consider recommender systems. What is the function of a best-seller list - as though I'm going to use that to select a movie? But if I can use the system with people like me - imagine the best read list from the people at this conference. Social - it's a group of others that are my peers, people who are like me. Consider people with rare diseases - they are the only person in their town, but they join with people around the world.
Comments: Jobs and projects: the idea that jobs are disappearing. Well... I don't think jobs exist any more; in our jobs everything seems to be project based. We are usually engaged in multiple projects. This leads to the obviation of departments - they don't make sense, their function doesn't make sense. The functions make sense - but the origanizational model doesn't make sense. But gets us to the hiring issue. Look at the ads... they have a project-like deployment in mind. They say, we just re-orged. And there's another re-org coming. So we need to make everything on-demand functions.
Think about staffing. When I hire one person, that changes what we're looking for in the next person. We want to hire for combinations of experience and skills. My requirements are unique - we want to hire unique people. This what allows us to put together a team. In learning, we're back in the barn-raising stage - we do everything from scratch. Today, we need architects and designers who never think they can create every door, every window. This divide and conquer model in differentiating sets of skills is what we're after; the team is not going to suddenly acquire much greater capacities.
Comment: we have to really work with out instructors, to get people to think that they don't need to create the wheel every time. We need to say, how am I going to use that for my course, and how are students going to customize it? The value proposition is the same - we don't want to get rid of the classroom and dump the lecture - but we can do additional things. Reply: the control of timing is being ripped out of our hands. The publishing industry is beside itself on this issue. Time shifting occurred with VCRs, but that was nothing compared with PVRs. People making 6 figure salaries sitting in rooms thinking about hat to program on Sunday evenings - gone. Just-in-time learning is starting to do this.
Comment: think about the other extreme. You are assuming ou have an intelligent audience. If I miss the West Wing, I miss it. Now let's push this. I like Friends, which means maybe I like such-and-such. But television isn't as big as it was. But this brings it right back to us. It has the result in dumbing down society. Reply: get over it. Could we have couch potatos on steroids? Absolutely. But I have a fundamental faith in human nature. Look at the declining use of TV time. But also, what was a medium - TV - becomes content in a new medium.
Me: Autonomy. Control. Darwinism that eliminates bad business models. Robby - "I've heard a lot of talks that start 'we are providing', 'we are producing'." Maybe the teacher isn't going away. But social behaviour is going to change. What will be behaviours and skills that the learner has to have when personalized learning is here. Not the teachers. The learners. Reply: I am intrigued by the notion of decision support. We have overwhelming choice. But choice becomes its own prison. The notion of decision support augmenting the capacity to make smart decisions - that's where this is going to. We want to push the decisions down to the level of the individual. Jack Welsh - pushed decision-making down, made hoarding of information a firable offense. Think of it - what if your co-worker was putting moiney into their own bank account?
Comment: decision support, just-in-time learning are good ideas. But this presumes foundational knowledge. We need an integrated view - we need to prepare people with this foundational knowledge. Apprenticeship model. Reply: yeah - the basic of knowledge: there's two types, tacit and explicit. It's as important to get expertise location. It's not all just knowledge assets. I need to be able to say, Can I find the right person to talk to. Or (commentator say) to work beside. Reply: yes, this is one of the top two recruitment tools - people choose jobs according to who they can work with.
Comment: a just-in-time education builds on a just-in-case education. What do students need to learn. A lot of he talk is based on this just-in-time, but not a lot on the fundamentals. You should learn something about mathematics, history, philosophy, theology... Reply: we have the temptation to give people what you think they need, versus guiding them in their decision. Until you have a motivated learner, nothing happens. That gets to teachable moment - when that's presented, that's when it happens. But (comment): the values you want to pass on. Reply: the former model was the programmer at NBC made the decision. The model is now dad makes the decision.
Views Today: 0 Total: 440.
SUBSCRIBE TO OLDAILY DONATE TO DOWNES.CA
Web - Today's OLDaily
Web - This Week's OLWeekly
Email - Subscribe
RSS - Individual Posts
RSS - Combined version
JSON - OLDaily
National Research Council Canada
All My Articles
Stephen's Web and OLDaily
Half an Hour Blog
Google Plus Page
Huffington Post Blog