Teacher Takes Notes

“All this she must possess,” added Darcy, “and to all this she must yet add something more substantial, in the improvement of her mind by extensive reading.”

— Jane Austen, Pride and Prejudice, Chapter 8

Read the scene again and notice what Austen is doing.

Miss Bingley has just rattled off the list of what makes a woman “accomplished”: music, singing, drawing, dancing, modern languages. Every item is something you can perform in a drawing room. Someone can watch you do it. Someone can clap. Then Darcy, of all people, adds one more thing and calls it “more substantial”: the improvement of the mind by extensive reading. No one claps for that. Nobody sees it happen. It is slow, private, and cannot be faked at the piano.

Yes, I know. The standard is gendered, it belongs to a world of drawing rooms and servants, and Elizabeth’s dry reply is that she’s no longer surprised he knows so few accomplished women. I’m not borrowing Darcy’s world. I’m borrowing his distinction, because it’s the oldest version I know of the problem I want to talk about: the difference between what looks like accomplishment and what actually builds a mind.

I’m a reader first and an educator second. So this one is personal.

Two voices I kept hearing together

Recently I’ve had two very different thinkers in my head at the same time, and they’ve started finishing each other’s sentences.

Cal Newport is a computer scientist at Georgetown University with a PhD from MIT (2009). He started his Study Hacks blog in 2007 and has since written eight books, including Deep Work, Digital Minimalism, A World Without Email and Slow Productivity, and he hosts a podcast. His through-line is simple and uncomfortable: hard cognitive work needs sustained, undistracted concentration, and the tools we’ve built (inboxes, feeds, and now AI) keep eroding our capacity for it.

Andy Hargreaves is the Thomas More Brennan Chair in Education at Boston College’s Lynch School of Education. He studies educational change, sustainable leadership and the emotions of teaching. His book Teaching in the Knowledge Society: Education in the Age of Insecurity (2003) won an Outstanding Book award from the American Educational Research Association. He argues that an economy running on knowledge needs schools that are creative and caring learning communities, and that years of what he calls soulless standardization have undermined exactly that.

[If you have the exact Hargreaves line on engagement and learning, with its page number, drop it in here.]

One a technologist, the other a sociologist of schooling. And here’s where they meet.

What Harvard’s hardest class is telling us

Here’s Cal’s video, the one I’m talking about.

Harvard has a first-year course called Humanities 10. It was dreamed up over dinner in 2008 by two English professors, Stephen Greenblatt and Louis Menand. The reading load is roughly one hefty book a week. The press keeps calling it “brutal.” And yet this year 180 first-years applied, a jump of about 40 percent, and half were turned away.

Think about that. At a time when a chatbot can summarise a difficult book in seconds, students are queuing up to spend a week with one.

Newport’s reading of this, in his recent episode on Hum 10, is that these students aren’t masochists. They’re choosing difficulty as protection, against a world that rewards shortcuts: the quick social reward, the fragmented attention, the generative-AI shortcut to knowledge. They’ve noticed something many adults haven’t. You can’t keep the mind and skip the work that builds it.

That’s my biggest takeaway, and I’ll say it plainly: cognitive ability isn’t something you own. It’s the residue of effortful work. Stop doing the work and the ability goes, however clever you once were.

Why this isn’t just a mood

I don’t trust a claim just because I like it, so here’s what sits underneath.

Learning and performance are different things. Soderstrom and Bjork’s 2015 review gathered decades of studies to show that what we can observe in the moment (a good answer, a smooth task) is performance, while learning is the durable thing, what you can still do next week. Conditions that make performance rise quickly often fail to produce lasting learning, and conditions that feel slower and harder often do better for long-term retention and transfer. The Bjorks call these desirable difficulties: spacing, interleaving, retrieving from memory rather than re-reading.

The signals we trust in classrooms are weak evidence. Robert Coe called visible engagement and busyness “poor proxies for learning.” Graham Nuthall found that students can be busiest and most involved with material they already know. Notice what Coe was not saying: that engagement doesn’t matter. He was saying it isn’t proof. A room can hum with activity while very little new is being thought.

The shortcut has a measurable cost. Studies of knowledge workers using generative AI have found that higher trust in the tool goes with less critical effort from the human. Research on cognitive offloading points the same way. When the tool does the thinking, the thinker does less of it.

And that’s where Hargreaves earns his place. His argument about standardization is an argument about systems: when schools are under pressure to show results, they reach for what is visible and countable. My own inference, reading him alongside Coe, is that this is the institutional engine behind the proxy problem. A system that can only see activity will reward activity. Miss Bingley’s list, in a modern uniform.

Where I could be wrong

A post that only agrees with itself isn’t worth your time, so let me push on my own argument.

• Selection effect. Hum 10 requires a placement writing exam and an application. Strong, motivated students choosing a hard course tells us little about whether hard courses work for everyone.
• Scarcity. The Harvard Crimson has argued that Hum 10 is a symptom of artificial scarcity: a flagship experience for the lucky few.
• A narrower conclusion. Careful commentary on Hum 10 points out that the evidence supports something smaller than “a rebellion against AI.” Students describing a wish for protection from shortcuts is meaningful testimony, but it is testimony.
• Newport’s evidence base. Much of it is synthesis, case study and advice rather than controlled trials. His principle lines up well with the learning science, but we should be honest about which parts are research and which are counsel.
• Not all ease is the enemy. Scaffolding at first instruction is legitimate. Difficulty pays off once there’s something to retrieve. Hum 10’s difficulty works because it’s purposeful, not because it’s merely hard.

None of this dissolves the core claim. It sharpens it: it’s not difficulty for its own sake, it’s effort aimed at the thinking itself.

What to take from this

I promised a clear takeaway for each of you, so here they are, in order.

For teachers

Stop treating a busy, engaged room as proof of learning. Measure what students can do later, without you.

• Check for thinking, not activity. Ask what students had to think hard about, and collect evidence of that: a cold retrieval question, an explanation in their own words.
• Protect productive struggle. Scaffold the entry, then resist the urge to rescue at the first frown.
• Set the AI rule at the level of the task. Attempt first, then compare. Build at least one moment per unit where nothing can be outsourced.
• And for your own learning: choose one demanding thing a year that you genuinely want to master. Treat compliance training as the minimum, not the model.

For parents

A child who finishes easily may not be learning much. Effort that feels hard is often the sign it’s working.

• Change the dinner-table question. Swap “Did you finish your homework?” for “What was hard today, and what did you do about it?” or “Explain it to me as if I’ve never heard of it.”
• Don’t rescue too early. Sit beside the struggle rather than solving it. Ten minutes of wrestling beats a handed answer.
• Protect one slow, unassisted habit at home: a daily reading block, phones out of the room, no AI summaries for the books they’re meant to read. Better still, read the same book alongside them.

For students

The thinking you skip is the ability you don’t build. Choose the harder path on purpose, at least some of the time.

• Try first, then ask. Write your own answer before you look at a summary or an AI response, then compare the two.
• Test yourself a week later. Close the book and explain it from memory. If that feels harder than re-reading, good. That’s the sign it’s working.
• Pick one hard thing each term and stay with it: one long book, one difficult problem, phone in another room. The reward is that you can do something afterwards that you couldn’t before.

Back to the drawing room

Darcy’s list of accomplishments is a list of things that can be seen, and he knows it, which is why he reaches for “something more substantial.” The things that build a mind are mostly invisible while they’re happening: the slow chapter, the idea that won’t resolve, the paragraph you write three times.

Lifelong learning isn’t a subscription or a stack of certificates. It’s a willingness to stay in difficulty long enough to be changed by it. The mind you want at fifty is the one you were willing to be uncomfortable for at twenty-five and again at forty. No tool changes that exchange rate.

Nobody will clap while you do it. Do it anyway.

My list of references:

• Austen, J. (1813). Pride and Prejudice, Chapter 8.
• Newport, C. The Cal Newport Show, “Reading As an Act of Resistance (Notes on Harvard’s Hum 10).”
• Hargreaves, A. (2003). Teaching in the Knowledge Society: Education in the Age of Insecurity. Teachers College Press.
• Coe, R. (2013). Improving Education: A Triumph of Hope over Experience. Inaugural lecture, Durham University.
• Nuthall, G. (2007). The Hidden Lives of Learners.
• Soderstrom, N. & Bjork, R. (2015). Learning versus performance: An integrative review. Perspectives on Psychological Science.
• Bjork, E. & Bjork, R. Creating desirable difficulties to enhance learning.
• Lee, H-P. et al. (2025). The Impact of Generative AI on Critical Thinking. CHI Conference on Human Factors in Computing Systems.
• The Harvard Crimson, “Open Up, Hum 10,” 6 November 2026.