Powerful Online Learning is a Distributed System

In the midst of a pandemic, universities are now rapidly shifting to online and remote learning.  I will suggest a metaphor for powerful online learning. This metaphor should resonate with our backgrounds as Computer Scientists and it also fits core principles of the Learning Sciences. As this is my inaugural blog in this venue, let introduce myself: I’m a learning scientist who works on improving STEM and CS learning with technology. I also have a CS degree from MIT. I like to connect the computer science and learning sciences parts of my brain.

Let’s start with what good online learning is not: it is not moving lectures online and continuing business as usual. Stated in computational terms, it is not about a central computational process (the instructor) doling out the same unit of work to hundreds or thousands of processors (the students) and then providing an authoritative rating of each processor’s individual work.

Let’s consider this alternative metaphor: designing powerful online learning as designing an effective distributed processing system. Stated in computational terms, it is about coordinating the active, engaged local work of separate processors (the students) towards a common goal (a community with greater shared understanding of the subject matter). It’s about organizing the connectivity in the system so processors (students) give each other help and feedback, and thus converge towards better learning.

Why active, engaged work? Because the direct cause of learning is not instruction, but rather the active, engaged effort a student commits to making sense of a challenging concept in their zone of proximal development (their area of growth). 

Why coordinating? Because learning sciences has collected a large body of evidence that when students have to elaborate their knowledge for another student (or coordinate on a shared knowledge product) they learn more. Students also learn more when they provide feedback to each other with explanations, not just answers or “do it like this.” Collaborative learning is very powerful when roles are structured so that students are required to actively elaborate, coordinate and give feedback. Together, active learning and collaborative learning expand the “zone” for learning.

In a project called GroupScribbles, we took the distributed systems metaphor literally. We created a “blackboard architecture” for a classroom. In GroupScribbles, individual processors (students) take jobs (intellectual work) from a shared space, and post partial results (learning) back to the shared space. The taking and posting of work is mediated by virtual sticky notes which students move between their own private space and a public space.

 

It’s much simpler than it sounds. In the GroupScribbles system pictured above, an elementary school teacher created initial sticky notes that asked for representation of a particular fraction. Students selected a fraction to work on by taking a blank note from the shared space to their private space (the atomic “take” operation in a blackboard architecture). They posted back a drawing of the fraction (the put operation in a blackboard architecture).

Then small groups were asked to take a collection of stickies for one fraction to a separate room. In the room, they elaborated an explanation for each representation, and then chose a single best explanation to share with the whole class. Finally, the teacher led a discussion of the smaller set of student work that was shared back. This is a powerful learning activity because students are working hard to elaborate, coordinate, and get feedback on what they understand across multiple modes: as individuals, in small groups, and in larger discussions.

Notice who is doing the work that drives learning in this distributed system: the students. Notice the instructor’s role: to find a creative way to make every student think hard about a different facet of the same problem, to use small groups to encourage knowledge coordination and feedback among students, and then to regulate the flow of information back to the central blackboard, where a teacher can add their unique value in further discussing some of the selected work of the students.

This metaphor can be fruitfully extended:

Think of students as a heterogeneous set of differently-abled processors. How can an instructor create a coordinated system so that every processor works as hard as they can? How can the cognitive diversity of students become an asset that drives learning to become deeper?

An instructor might consider how students can give peer feedback; how students can self-select easier or harder challenges; how students can promote issues or questions from small groups to bigger groups; how shared understanding of a concept can be made stronger by comparing and contrasting different student elaborations of the same concept. In an elegantly architected distributed system, the connectivity and activity of individual processors (students) overcomes the limited bandwidth of the central processor to give attention to each individual unit.

The Distributed Systems Metaphor for Online Learning suggests that an instructor should fully engage and connect the students (the processing nodes) to maximize their active effort to elaborate, coordinate and give each other constructive feedback, with a collective goal in mind (i.e. learning more deeply by harnessing cognitive diversity).

Jeremy Roschelle is Executive Director of Learning Sciences Research at Digital Promise and a Fellow of the International Society of the Learning Sciences.