Traditional Learning Ecosystem, Complexity and Infrastructure

March 21, 2022

Two years ago in March 2020, as Covid was just beginning to have a severe impact outside China, I blogged about Recreating the Traditional Learning Ecosystem Online. That blog is about how the online learning ecosystem is falling short of its potential, and ways we might address that by leveraging approaches that have worked for thousands of years – i.e. engaging learners in specific learning environments where learning is highly specific and consequential, such as learning in the course of work or other activities. These ideas are core to our mission at LearnerShape.

Now, as the world enters another crisis following the Russian invasion of Ukraine, I want to reexamine this topic. My basic thoughts have not changed, but we have learned a lot in the past two years at LearnerShape that informs a more sophisticated view. I have two main observations.

First, online learning systems fail to replicate the success of the traditional learning ecosystem largely because they fail to engage appropriately with the complexity of learning interactions. Learning is more complex than almost anyone believes.

Second, effective online learning tools should enable complex human learning interactions without attempting to manage that complexity. Put differently, a simple system can allow complex interactions.

Let me take a closer look at each of these points.

Complexity of Learning

Imagine a football game (I mean football the way most of the world understands it, known as soccer in North America). The number of player configurations and situations that can arise on the field is practically limitless – which is one of the main reasons that the “beautiful game” is so exciting. There is no way to design an approach for every situation. Instead, players learn skills, team behaviors, flexible set plays and other techniques that can be assembled for football success. We need to think about learning in the same way.

Learning is exceptionally complex and varied – with an accent on “exceptionally”. Maybe that statement seems obvious, but most people don’t get its full import. I did not appreciate the full complexity of learning when I founded LearnerShape, and even in the learning and development sector very few people do so. Most of us think we understand what learning involves, but if you try to reduce that to a fixed set of activities, you quickly run into a wall. Few (if any) aspects of human activity involve as much complexity as learning does. (As an aside, this is related to why the human brain is so complex.)

Let me use an example to illustrate the complexity. There has been significant recent attention in the semiconductor industry to the complexity of extreme ultraviolet (EUV) lithography machines built by Dutch company ASML – some of the most complex machines ever built. But learning is more complex … much more complex. EUV lithography solves a massively challenging engineering problem, but aimed at the fixed goal of building denser semiconductors.

It is not effective or even possible to reduce learning to fixed goals. Of course, for childhood and even university education, there are many defined curricula that attempt to do so. But this approach comes at the extreme cost of failing to address the needs of individual students. For an elegant explanation of this problem, I recommend System Failures: The Education System and the Proliferation of Reductive Thinking by Leyla Acaroglu. Among other excellent points, Acaroglu notes the importance the traditional learning ecosystem that I described in my blog two years ago:

“Early on in the human experience, children learned what they needed in order to become effective adults through the socially-learned tools of the group, from hunting and gathering to weapon making and finding life-sustaining resources (such as water) through story lines.”

She concludes, with a somewhat political bent, but equally relevant to the practical effects of linear education:

“If we don’t change the way we share knowledge and prepare young people for life, then we will continue to encounter the same problems in the world that we are grappling with today, since we will reinforce linear and reductive thinking as we go about creating the same type of thinkers that created those problems to begin with.”

The problem of complexity is arguably even more serious for adult and continuing education, where the number of variables is much greater than for childhood education. Adult learners need to acquire skills for their specific work situation, and to learn them in a way that fits their work and personal context, experience, aptitudes and existing knowledge.

When one fully appreciates this complexity, it is easy to understand why current online learning platforms often fail to deliver on their promise. There are a huge number of learning management systems (LMSs) and learning experience platforms (LXPs) competing for corporate market share, and they provide solutions that can do an excellent job of serving standard enterprise learning needs. Some of these are making remarkable progress in managing learning content with advanced artificial intelligence techniques – I am most impressed by EdCast.

But anecdotally, from conversations we are having at LearnerShape, even LXP market leaders like EdCast and Degreed are struggling to gain employee engagement with their sophisticated solutions – the same problem that has long dogged LMSs.

LMSs and LXPs tend to be complex, with many features, in order to capture the huge complexity of learning applications. And even with all this complexity, they often fail to create learning environments that truly engage learners. We believe it is possible to do better.

Simple User Experiences for Complex Behavior

LearnerShape believes that the way to improve online education is to create an environment that allows teachers and learners to generate learning experiences that fit the exact needs of learners and the learning context. In some ways, this is what LXPs are seeking to achieve by personalizing learning through AI-driven automation, but their efforts are running into the challenge of complexity discussed above. Truly generative learning experiences can be much simpler.

Take the example of Kahoot!, a gamified quiz platform that is one of the most successful edtech products of all time, with tens of millions of active accounts (and over 5 billion cumulative plays). Kahoot! is a deceptively simple game, allowing users to create quizzes with text and image questions with four multiple-choice answers each. This simplicity enables amazing flexibility, so that a quiz can be about anything, with customization residing primarily with the user rather than the platform. This also enables integration of Kahoot! online quizzes into real-world experiences.

At LearnerShape, we aim to enable a wide range of targeted learning applications using AI-enabled open-source learning infrastructure. Our infrastructure is a set of cloud-based microservices, including:

At present, combining these components to build targeted learning applications requires coding and other technical expertise, either from LearnerShape or from others who adopt our open-source technologies. For example, we are currently using our infrastructure to build:

Each of these applications has a very simple user experience, enabling a wide range of flexible and complex user interactions. This is what Kahoot! does, and what was possible in the traditional learning ecosystem. This approach of enabling complex behavior with a simple user experience avoids the need for service providers to manage complexity. More important, users can generate powerful learning experiences without being dissuaded or distracted by complexity. (My sailing friend Greg Brougham recently co-authored an article making very similar points about emergent complexity and simple scaffolds for software development.)

LearnerShape will continue to evolve our infrastructure, so it becomes steadily more powerful and easier to create learning applications, with steadily reduced need for technical expertise of builders. Ultimately, we intend for it to be possible for our infrastructure to be assembled like a set of Lego bricks. We also intend to introduce simple but powerful end-user products built with our own infrastructure. And we plan to make most of our code open-source to enable the widest possible adoption.

Our vision is that online learning tools should interact simply and flexibly with human interactions, as was experienced by “cavemen and cavewomen, blacksmiths and babies” in the traditional learning ecosystem, as I posited at the beginning of my blog two years ago.

Please get in touch to join us on this journey. We like talking with humans!

Maury Shenk, Co-Founder & CEO, LearnerShape