Very loosely coupled

In 1934, LEGO founder Ole Kirk Christiansen created the name LEGO by combining the first letters of the Danish words “LEG" and "GODT”, which mean “play well” – unaware that in Latin one meaning of the word LEGO is “I put together.”

Why are learning objects like LEGOs? In 1958, the LEGO brick was launched, with its new socketing system. Children could snap together the bricks to construct different things, limited only by their imagination. LEGO bricks have become the archetype of interchangeable, universal parts.

Similarly, learning objects were supposed to be reconfigurable. String 'em together to build individualized learning paths. Make 'em small enough and they become like wet plaster you can shape by pouring into a mold. Instructional designers wouldn't need to reinvent (or repurchase) the wheel. Learners would get just what they needed.

While learning objects have taken hold in limited applications, their recombinant capabilities have not exactly set the learning world on fire. You can have a boatload of LEGOs but they're not going to assemble themselves. Making something requires intelligence -- the child. Likewise, learning objects need an active component in order to self-organize.

Objects are things. Content can be a thing, but content is not learning. Learning requires context, too. Content + context = a learning process. Processes don't snap together as seamlessly as LEGOs. Processes join with one another in unpredictable ways, with many more variations than one can represent by coupling the male and female aspects of little plastic bricks.

Real-world applications require flexibility. Those of us who are too impatient to await the arrival of nano-objects that conform to the shape of a contrainer can find a better model in intelligent software agents. Agents are always on the make, looking for optimal connections.

Their standards of connection evolve over time. Intelligent agents adapt. Unlike LEGOs with their single way to connect, software agents are like sticky candy. They can come together in any number of configurations.

The learning processes of the future must be responsive, self-maintaining, ever-improving, cost-effective, and malleable.

I'll suggest we start thinking about learning agents instead of learning objects.

Posted by Jay Cross at May 10, 2004 12:05 PM | TrackBack

At eCornell we've been experimenting with an approach called learning molecules (, which looks a lot like your diagram of agents, Jay. Although a learning molecule is composed of what are basically learning objects of five different types, it is the relationship between those objects that imply a learning process. Molecules have at the center a Scenario learning object, which provides the context and motivation for learning (this is basically a problem-based learning approach). The Scenario is surrounded by other learning objects categorized as Resources, Collaboration, Utilities, and Evaluation elements. One interesting thing I've observed about our model is that we are seeing less value in reusability of learning objects to build new molecules than in the ability to easily modify some elements of a molecule to create a new custom-made learning molecules, tailored to a specific group of learners. For example, we can change the Scenario of a molecule to fit a different type of industry, while leaving the rest of the learning objects in the molecule intact. Of course designing, producing and customizing learning objects is likely to continue to be a manual process, but I can see how intelligent 'agents' can take over the on-demand reconfiguration of molecules to fit the specific needs of a learner.

Posted by: Ulises Mejias at May 11, 2004 06:59 AM

I think we do have very primitive intelligent agents in the form of search engines, particularly those that shape their activities around the needs of the individual and reform around specfic needs. Some, like Google, are database miners, others which are emerging are people, or human biocomputer, miners (active and passive, depending on access) These are reflected crudely in Neil Stephenson's The Diamond Age's intelligent book. Or they are represented in some of the advanced ideas emerging in the arena of "chatbots"

Posted by: tom abeles at May 11, 2004 08:28 AM

30 Poppy Lane
Berkeley, California 94708

1.510.528.3105 (office & cell)

Subscribe to this Blog

Enter your email address to subscribe. We vow never to share your information with anyone. No Spam.

Subscribe Unsubscribe

Reference Pages

Building Community
CSS, Semantic Mark-Up, and codes
First Principles
How People Learn
Knowledge Management
Learning Links
Learning Standards
Making It Work (Implementing)
Metrics & ROI
Social Software
String theory
The eLearning Museum
Visual Learning


Our Infrequent Newsletter
Sign up for our sporadic newsletter.

Entries by category...

Customer care
Emergent Learning
Just Jay
Recycled from Blogger
The Industry
Workflow-based eLearning


Internet Time Group

© 2004 Internet Time Group

Click for Berkeley, California Forecast
Berkeley, California

Recent entries

New Blog
Blogger Experience, Housekeeping, Something New
Loosely Coupled
Above all
Demographics is destiny
Are you setting the bar high enough?
Virtual Apps
Aerobic Learning
Work as Video Game
Oracle and Macromedia, Sitting in a Tree
The Blogosphere
ASTD Silicon Valley
Performance Support
Kingsbridge Conference Center
First Post by Email
Inactive Blog
RSS Feed for New Site
Comment Spam
Testing ... testing ... 1...2..3
IT Doesn't Matter - Learning Does.
All blogging is political
Mutlimedia Learning
Damn, damn, double damn
Nonverbal impact?
The New Religion
Wolf! Wolf! Wolf! Wolf! Wolf! Wolf!
Business Process Management (2)
Really Big
Business Process Management Conference
Don't Lose a Common Sense: LISTEN
It's only natural
Go with the flow
Time Out for the Fair
Informal get-together in SF this Wednesday
Repetition, reverb, and echoes
Who Knows?
Cognitive Mapping
Push vs pull
The Big Picture on ROI
Art Break
TDF Finale
New Community of Practice Forming
More TDF04
Training Directors Forum 2004
A Rare One-Liner
PlaNetwork LIVE 2
PlaNetwork LIVE
ASTD 2004 Leftovers
Worker Effectiveness Improvement, not KM
Upcoming Events
eLearning Effectiveness?
Jay's Talk at ASTD
Mintzberg & Cooperider
Lest ye forget
ASTD International Conference & Exposition 2004
Knowledge Tips
What is Workflow Learning?
ASTD msg 1 of n
Look out, it's Outlook
Collaboration at ASTD Next Week
Tell me a story
User indifference
The shortest presentation on metrics you will ever hear
Back to Blogger
Windows fixes
The Alchemy of Growth
Grab bag
Very loosely coupled
E-Learning from Practice to Profit
Robin Good kicks off Competitive Edge
China Bloggers
Sonoma Dreaming
Upcoming Events
Emergent Learning Forum: Simulations
The Best Things in Life Are Free
Metrics and Web Services
OpEd: ROI vs. Metrics
e-Merging e-Learning
Loosely Coupled
Search me