Productive Friction
What a chain-smoking Soviet film professor taught me about learning in the age of AI

My first class in college was a nine-person seminar on the Soviet filmmaker Dziga Vertov (it came with a free trip to St. Petersburg, which was my true motivation for signing up). But it turned out that trip came with a cost, and that cost was the professor of this class. Grumpy doesn’t begin to describe this guy’s mood. On the first day, he scowled at us while introducing himself and then, 15 minutes in, went outside for a cigarette break, something he’d do periodically through the semester whenever he was especially fed up with us. Perpetually disheveled and constantly nudging his glasses up his nose, he looked like a kindly academic, but acted like your meanest gym teacher.
“This will be the hardest class you ever take!” he barked, once he finished his cigarette. (He was right about that one.) He made it clear from the jump that he didn’t believe in our thinking or writing abilities and while he would do his damndest to teach us, he couldn’t be held responsible for what he said was effectively an impossible task. Needless to say, it was in that moment that I really felt the jump from high school to college.
In the first week, we watched one painfully long and confusing movie and then we were tasked with writing a paper about it over the next two weeks. And that was it as traditional teaching goes. Each class thereafter was devoted to all of us critiquing one person’s work – 90 brutal minutes in the hot seat. “No positive feedback allowed,” he clarified. “It’s not productive and a waste of time.” Afterwards, we’d have a chance to refine the paper before turning it in. That one assignment would dictate our grade for the class.
But first, we all had to turn our papers in to him for an initial review, at which point he sent the following email on the Friday before parents’ weekend:
I’ve read your papers and right now, five of the nine of you are failing.
Sincerely, Professor M
Cue nine overachieving freshmen having breakdowns to their parents a month into college. The class was difficult in a way I’d never experienced before. My time in the hot seat was stressful to say the least, but with every critiquing session, I found myself rethinking my entire paper. More than that, I found myself rethinking how I think. That’s how intensive his feedback was.
It was, in many ways, a nightmare. And yet as the semester wore on, I could feel myself really learning. And as we all started thinking in deeper and more rigorous ways, he started easing up on the angry professor schtick. In the end, we all passed. And that paper remains one of the best things I’ve ever written.
Throughout history, we humans have been working to eliminate friction.
The wheel to get us places faster. The telephone to connect us to each other instantly. The search engine to put information at our fingertips. And until now, this orientation made sense. Because the world was full of friction that if reduced or eliminated would pave the way for progress.
But now we seem to be losing something in the process. The friction of connecting with people is gone – we can reach almost anyone through a platform. But the friction of getting to know someone in person is what leads to a deeper relationship. The friction of writing a first draft is gone – we can outsource to the LLM of our choice. But the friction of wrestling with an argument, or articulating an idea, is how you learn what you actually think.
How do you develop your own sense of self without the friction it takes to understand what you believe?
Productive friction
Professor M’s tactics were questionable at best. I generally don’t support scaring the sh*t out of 18 year olds in the name of serious teaching. But he created something that I’ve grown to value over time: what technology investor William Van Lancker calls productive friction. Productive friction is the twists and turns of the journey – the setbacks, the struggles, the misguided leaps and the obstinate frustrations that turn you into a critical thinker. It shapes not just a clear point of view on the world, but one rooted in hard-earned conviction.
Productive friction, in other words, is a key piece of how we learn. And with so many innovations eliminating the friction from our lives, it can be hard to tell the difference between performance and learning, even within ourselves. It’s confusing because research has long shown that a lot of learning can be happening even when there is no obvious change in performance. And we can improve our performance without any significant learning underneath it.
Think about it: you could perform Martin Luther King Jr’s I have a dream speech perfectly from memory, in exactly the right tone and with sizzling charisma, all without a deeper understanding of what it was really about. Conversely, you could listen to that speech and see no obvious change in your behavior, until months later, when the lessons it taught you surface exactly when you need them.
This has always been true but AI and this era of supercharged innovation is blurring that distinction at a massive scale. And that’s a problem because real learning is often invisible, even to yourself, until suddenly it isn’t and you realize you’ve changed. The reason the gap between performance and learning exists at all turns out to be structural — it’s how the brain actually encodes things.
I knew none of this when I was sitting in Professor M’s class.
But at the end of the semester, I knew that something had shifted – not just in the paper itself (though 7 revisions will do that), but in me. That kind of learning, and the conditions around it, had done something new.
The psychologists Elizabeth and Robert Bjork call it “desirable difficulties” – the conditions of learning that, while they make things more difficult at the time, actually lead to better, deeper learning by triggering “encoding and retrieval processes in the brain that support learning, comprehension and remembering.”
These conditions include varying the conditions of the learning (like changing the environment, using experiential learning alongside lectures or, say, learning through film). One study showed that studying the same material in two different rooms led to better recall for participants than studying the same material twice in the same room.
Spacing out learning is another condition – better to learn something slowly over time (like, say, critiquing the same movie for 4 months in a row) than trying to learn it all at once (like cramming for an exam). And an obvious one – looking up an answer or having someone (or some-LLM) tell it to you is far less effective than using your own critical thinking and prior knowledge to generate the answer on your own. Even if that means having a certain chain smoking professor push back on your arguments over and over again.
The modern project of developing technology to remove friction has succeeded.
And in many ways, that’s great. It’s hard to argue that the invention of the wheel was anything but helpful to humanity. I didn’t lose any critical thinking skills when GPS came along – I just finally stopped being lost all the time.
But now we’re eliminating friction so fast that we’re not distinguishing between the kind that wastes time and the kind that builds something in us. And it’s hard because the discomfort of productive friction is often indistinguishable from the discomfort of something that is just bad. Most of us have optimized our way out of something we should have stayed in. Because the stuff productive friction builds – judgement, taste, conviction, perspective – is invisible when it’s being built. Sometimes we only know it’s productive in hindsight.
At the end of the semester, Professor M – with heavy encouragement from his wife, invited us to dinner. And while we walked in mildly terrified, we actually had a nice time. But the real surprise came later, when he invited us to dinner again, and every semester after that until graduation.
Which made me realize that not only did he provide productive friction for us, we provided some for him. And the experience on both sides eventually turned from stressful class time to genuinely fun dinners.
Not every hard thing is productive
Some friction is just damage. But the kind that’s building something in you is the kind that changes how you think and how you feel, even if the process doesn’t feel great.
We need to design productive friction back into our lives whenever we can see that it’s there for us to take. But for all the times when it’s not as obvious, we need to develop the capacity to stay in something long enough to see what it’s building. Because maybe that friction you’re feeling is indicative of something wrong. But maybe it’s the beginning of something else – the kind of thing that changes you so much that you end up writing about it 20 years later.



I think every technological advance makes us think we're losing something critical and essential in the trade. I've heard plenty of people saying their ability to navigate spaces has atrophied since GPS became available on mobile devices and that it's made them worse at exploring new cities and even finding their way back to the lobby of a museum. But the friction didn't actually go away - expectation increased. It's no longer acceptable to receive an address and ask what's the best way to get there. You're expected to leverage technology to determine whether subway, bus, car or taxi is best. Gen AI is similarly shifting the goalposts further. Expectations will rise to meet our new capabilities.
Love reading what you share, always so thoughtful and I appreciate your writing