The Future of Learning

The text version of an event I ran at Interhackt 2020.

This is the text version of an event I ran at Interhackt, the designathon I helped organize last December.

Watch the talk in full, including the discussion afterwards, here:

A couple months ago, Azlen Elza had this idea of hosting a few learning groups on different subjects. One was called ‘The Future of Learning’. There was a group of us who would get together every week and talk about different learning strategies. We'd read different papers and try to figure out what things are broken with the current way in which people learn, and how we might fix those things. This is going to be a big, high level summary of all that — talking about the major ideas people brought up and where we got to as a group over those few months.

Let's begin with a little story about how the future might look, in terms of the education system. Imagine this: you're in the future and the year something like 2030, ten years from now. Every city has learning hubs — these places where people can go and collaborate with others on some subject. The minute you walk in, you see this big wall. And on the wall, there's a list of mysteries — all the unsolved problems in a particular field.

For example, you’d have all the unanswered questions in math listed out, and then you’d see the same for biology, and so on. You'd be able to look at this wall and see where we're at as a civilization. You'd be able to pick one of these mysteries and say, “Okay, this question looks interesting, I want to see who else is working on it,” and get matched with people working on the same thing. You might have some kind of mentor or advisor to help you though that process.

If you think about how Netflix replaced TV — replacing these bundled channel packages with on-demand content — I imagine a similar thing happening in the education space, where instead of getting a predetermined degree, students can sign up for courses à la carte. That way, they wouldn't have to specialize in a particular field right away — they would be able to take the set of courses that interest them and come away with a personalized degree.

There's also a lot that can be done in terms of apps — apps that help you get the right learning habits in place, for instance. There’s power in just checking in consistently and learning something new everyday. Habits make results compound. For instance, right now I'm learning to play piano. The main way that I'm learning is through an app, SimplyPiano. It’s structured by lessons, listens to you play and gives feedback on how well you're doing. You can check in there, play for ten minutes and be done with it. That's something that I think has a lot of potential in other domains, not just music — I call this idea the Dojo.

Reviving the Education System

When I look at the current education system, it feels in some sense dead. What do I mean by that? The people who come into prestigious universities have many hopes and dreams, and a lot of ambition — you can see the fire in their eyes. And then, by the time they graduate, they become ‘adults’, so to speak. They condition themselves to cope with all the inefficiencies of the system, and to find the boring things tolerable. And maybe the most they hope for is to find some kind of office job to be able to pay their bills. It's like all of the creativity, all of the ambition is somehow getting drained from these people. The two concepts I find incredibly useful for understanding what's so broken about the education system are Goodhart’s Law and the Procrustean Bed.

Goodhart’s Law basically states that when a measure becomes a target, it ceases to be a good measure. Let's say you're trying to do something, and you have a metric to tell how well you're performing. Well, if you start paying attention to that metric instead of paying attention to doing a good job, the incentives can get messed up.

The second, and maybe the more important concept here is the Procrustean Bed. In ancient mythology, Procrustes was this person who would offer his bed to travelers who needed a place to stay. But if the person was too short for the bed, then he would stretch them to fit them exactly to the bed. If the person was too tall, then he would cut off the person's legs just to fit them in the bed. The general point here is that a Procrustean Bed makes a diverse range of humans conform to a rigid system. Once you start thinking in these terms, you can notice a lot of instances of this in society. I'd argue that the education system is a major example of a modern day Procrustean Bed.

The question then is, how can we make education more alive? How can we revive it? It shouldn't be a Procrustean Bed. It shouldn't be a Goodhart’s Law situation where you're optimizing for metrics instead of trying to learn.

In our learning group, we came to the conclusion that education should be something that's personalized, and decentralized in a certain sense. It's easier to make a small system personalized than a large one. If you have some kind of big institution, then you run into this problem of having to accommodate many different kinds of people and you inevitably end up with a compromise.

There also has to be some kind of intrinsic motivation — you have to actually want to learn. It’s not enough to just have extrinsic motivation in the form of grades or something like that. By contrast, what I find frequently these days is a kind of cynical view of the system itself. It's like, “Ha, if you think that you're here for learning, well then, it just means that you're naïve or something. In reality everybody's just there to get high marks and a degree.”

Self-paced learning is important, or at least group-paced learning. If you're part of a group of people who are learning something, then the group goes at its own pace. The thing I've noticed in college is that you have a bunch of courses that, no matter the subject matter, get crammed into four-month sessions. Oftentimes there is no opportunity to go back and reread, because now you have to move on to the next chapter, or the next assignment.

Education should also be intense if you want it to be. Some of my best learning experiences have been in very intense, high stakes situations. But it can't be forced on you — you have to put yourself out there.

And maybe most important of all is an emphasis on exploration and adventure. Mystery is a motivator. What haven’t we figured out? What's still there to learn?

Peer Learning

A big part of what we discussed was peer learning. Specifically, we talked about a shift that happened towards the second half of the 19th century from a monitorial system — a peer learning system — where you had some mentor or teacher who monitored the class, to a teacher-led system where there was more of a top down approach. In a teacher-led system, you have an instructor and the instructor tells all the students what to learn and how to do things. There are many reasons for this transition. The Industrial Revolution probably played a big role in this — more people started going to school, so you had to scale up the system. How do you scale it up? One way is to make it more standardized. There could be plenty of other reasons — not really sure. The point is, the transition from a monitorial system to a teacher-led system, intentional or not, encouraged a lot more conformity. You weren't really talking to your peers as much anymore, you were just getting top-down instructions.

On the other hand, we've read a couple of papers that talked about this peer approach to learning. It's been found, at least in the papers we've read, that peer learning helps with the discovery of new ideas. It also helps with motivation, because when you're doing anything as a group, you're pulling each other up. High energy environments are contagious. Peer learning also helps fill in knowledge gaps through conversation with others, and encourages more agency on your part, since you're not getting told what to do as much.

We've also come across this difference between peer tutoring and peer collaboration. Peer tutoring means having one of your peers who knows more about the subject give you instruction, whereas peer collaboration is just everybody trying to figure things out together. Both are very important, but are different approaches within the peer learning paradigm.

We found that smaller groups tend to work better than larger groups do. Anywhere from two to eight people seems to be the the best — that's where you can have the most interesting conversations and everybody gets an opportunity to chime in.

On Competition

The other big thing we looked at was competition. What are the different forms of competition? Is competition good for learning?

If you look at the current system, the biggest example of competition is grades. We looked at how grades impact students’ behavior. For the most part, it seems that grades reduce interest in the subject matter. I don't know about you guys, but at my university there's a lot of talk that goes something like: “Why are you taking this course?” “Oh, I'm taking this course because it's a grade booster.” The person is not interested in the subject matter per se, but because you can get a good grade in the course, and grades are what matters, they take the course anyway. This is one of the most obvious examples of Goodhart’s Law in action, where grades serve as the metric. Everybody ends up focusing on the grades, and the knowledge part of education gets sidelined. To be fair, grades are easy to implement. But based on what we've read, I'd say that they make learning more difficult instead of easier. They benefit the administrators much more than the students.

We can also talk about competition in the education system in general. Take a super competitive environment — a zero sum environment where one student doing well means another student does worse. For example, take a program with limited spots where everybody has to compete to get in. Or, situations where everybody's graded on a curve, so how well the top student does actually impacts everybody else's grades. These kinds of things encourage people to — in the extreme case — sabotage each other. So they're not very good for cooperation. They're also not very good for risk-taking. If you're in a competitive environment and you want to do well, then maybe you don't want to take the challenging math course that's gonna drag your average down, even though it's interesting, because at the end of the day your average is what matters.

There are different types of competition. There can certainly be some kinds of healthy competition — for example, look at this hackathon. This hackathon is basically a competition — there are prizes and stuff — but it's a very healthy environment! Everybody's eager to help each other out. In high school, I was part of a competitive robotics program where it was a similar thing. Yes, there was competition, everybody were building their own robots, but every team wanted to help every other team because at the end of the day, learning was the priority.


We also talked about something we're calling artifacting. An artifact is anything that's a collaborative output from a discussion or learning session. For example, an artifact can be a drawing, a piece of code, or a video. It can be some kind of explanation of a concept that the group came up with. This idea, I think, has a lot of promise.

What we found in our learning groups, however, is that it's trickier to create artifacts than it might at first seem. Everybody's going into it with different goals in mind, and everybody has their own little thing to contribute to the conversation. It's hard to get one cohesive output from a group of people, especially if the group is relatively large, just because it can be hard to get people on the same page. People have different motivations, different goals. Regardless, that's one of the things we played around with, and plan to continue exploring in the future.

Top-Down Learning

Another thing that came up a lot was this idea of top-down learning. If you think about a university course, or a university program, the way that learning happens is bottom-up. You have courses in specific things — for example, an algorithms course, a web development course, or a calculus course. Later on, you may or may not find the knowledge you gained useful. It's this notion of: learn the specifics first, and then find a goal to which you apply the specifics. The whole talk about “becoming employable” or “finding a job” rests on the presupposition that you’re a key in search of a lock, and the university is the system that molds you into the key. Procrustean Bed.

What we talked about was: what if you flip that on its head? What if you do the complete opposite? Start with some high level goal that you're completely unqualified for, and then figure out what things you need to learn in order to reach that goal.

Here’s a true story about how I thought through these things: I wanted to create something that would improve our collective sensemaking. I felt that that we, as a society, are doing a pretty poor job of making sense of the world. So I was like, “Okay, what might solve that problem?” I thought it might be a web platform. The only question is, how do you do web development? I knew nothing about web development at the time. So, I started doing research and learning about how people do web development. What are the tools people use? What are the best practices? You start with this high level goal and get more and more specific as you go. You get more technical, too. Eventually, you might settle on something like ReactJS for your frontend framework, a MySQL database, and maybe Django for the backend or something like that.

This top-down approach to learning seems better, because you can always tie whatever you're learning back to the high level goal. You go “Okay, I need to learn this specific thing so that I can progress on a project.” It's not some kind of abstract “Oh, I'm learning calculus because calculus is useful,” or “I'm learning web development because web development is useful.” It's hard to motivate people when they can't see a direct application of the thing being learned to their goals.

For more on this, check out the essay I wrote specifically on this topic.

FIRST Robotics

And, just very briefly, I also want to tell you about my time in FIRST Robotics. In high school, I founded and ran a competitive FIRST Robotics team — Team 5897: APEX Robotics (check out our 2016 and 2017 robots, respectively). Every year, a new game gets announced with completely new rules, and you're given six weeks to build a robot entirely from the ground up. You're not given any time at all — it's this very intense period where everybody sacrifices their leisure time and their social lives just to go and build this robot. But it's super rewarding, because you end up learning a bunch throughout that process. And then you go on to compete with other teams in the States, or in Canada. You'd have teams from all over the world come. It’s this incredible, life changing experience. If any of you guys know anyone in high school, I'd highly recommend that you tell them about FIRST Robotics.

The thing that I loved most about it was the fact that it was completely project-based. The mindset is: “We have to build the best robot possible, because we want to win the competition, or we want to get some kind of reward.” In order to do that, you have to think about strategy. You have to think about design. You have to think about the actual engineering process, not to mention fundraising and recruiting. And because you're given six weeks to build, you have to do all that very, very quickly, and that encourages people to learn a lot.

We also had mentors. The mentors were either parents or industry experts. Whenever we were stuck, or whenever we had a question, we would come to the mentor and the mentor would explain how this or that thing worked. Extremely helpful.

There was also a strong sense of community and friendly competition, as I mentioned before. FIRST uses the word coopertition. I pulled this quote from the FIRST website that describes it:

Coopertition involves learning from teammates. It is teaching teammates. It is learning from Mentors. And it is managing and being managed. Coopertition means competing always, but assisting and enabling others when you can.

And you really saw that at the events people would go to — at the regionals and at the Championship. Yes, there were matches, and people would compete in the matches, but at the same time, everybody was running around trying to help fix up each other’s robots. You usually had a lot more people show up to the competitions than you needed. Let's say you have a 30 person team, but you really just need six people in the pits to fix up your robot. So you had all these high school kids with nothing to do, and a lot of them would go socialize with other teams and help them out. It was pretty magical.

The two things that that didn't work as well were knowledge transfer and cost. Because it's restricted to high school kids, when one wave graduates, all of that knowledge gets lost, unless you have some kind of mentor who acts as the knowledge keeper and teaches the younger students how everything's done on the team. The other problem was that it was pretty resource intensive. To sign up for one event, you needed something like five or six thousand dollars. And most teams signed up for more than one event because that gave you more opportunities to improve your robot and compete, and therefore more opportunities to get to the championship. So that's pretty expensive. And that's not even taking into account the costs of travel for all your teammates, and food, and the resources you need in order to build. You also need a space to build. We had budgets in the tens of thousands. It's not uncommon for teams to have budgets of a hundred thousand. So if you want to make this type of project-based learning more accessible, then the cost of some of these things needs to be brought down.

Selected Papers and Resources

Here’s a list of some papers we read as part of the ‘Future of Learning’ group:

  • Bamberger, ‘The Computer as a Sandcastle’

  • Landahl, ‘Learning to listen and look: the shift from the monitorial system of education to teacher-led lessons’

  • Damon, ‘Peer Education: The Untapped Potential’

  • Cranshaw & Kittur, ‘The Polymath Project: Lessons form a Successful Online Collaboration in Mathematics’

  • Kohn, ‘The Case Against Grades’

  • Goodman & Crouch, ‘Effects of Competition on Learning’


I have written a number of other pieces about learning, each touching on a crucial idea that a learning system of the future could incorporate:

  • Top-Down Learning, about starting with a high-level goal and drilling down into specifics on a need-to-know basis.

  • Behavior Loops and Dojos, on the importance of practice, habits and consistency in learning.

  • The Three Month Rule — if you begin learning a skill, you should at least stick with it for three months.

If you liked this post, feel free to share it with your friends! If you have any feedback or if I got anything wrong, please let me know! | @savsidorov