#TeachersDilemma: Shifting Practice

I've been deeply immersed in a new project using Anthropic's Claude (Opus 4.6) to test its coding capabilities by designing and building a new website to synthesise the tension I have always felt between school and learning. I started by doing a deep dive into why schools are uniquely impervious to change. Frankly, it has been a bleak and depressing task. I decided to give myself a much needed break by exploring what actually works when it comes to shifting teaching practices. The conversation with Claude was so interesting that I decided to preserve it, as is, and share it in a blog post.

---

Chat with Claude Opus 4.6

Feb 26, 2026

Marianne:

Hi, in our last convo, we discussed what is holding school back. I would like you to do a deep analysis of what has been effective in shifting education. Firstly, identify 8 of the greatest things having an impact on education to becoming more relevant to the ways in which the world has changed. This is the hopeful conversation. These are case studies of what does work.

Claude:

Good — I found our conversation (Schools' resistance to modern change) where we built those two lists of 8 tensions each. This is the flipside conversation. What's actually working.

Before I lay these out, a framing note: I'm deliberately choosing examples where the mechanism of change is visible, not just the outcome. "Finland is great" isn't useful. Why something works is what matters for your framework thinking.

Here are my 8. I'll give you the headline and core argument for each, then we can crack open whichever ones provoke the most friction.

1. Schools built around pedagogy, not content

High Tech High (San Diego), Big Picture Learning (global network), XP School (Doncaster, UK). These institutions embedded the pedagogical DNA into their institutional identity — hiring, assessment, community expectations, physical space. The model doesn't walk out the door when a principal retires, because the pedagogy IS the institution. This directly addresses the pattern you lived at both Elisabeth Morrow and Newlands [Intermediate]: innovation that depends on a visionary leader is structurally fragile. These schools solved that problem architecturally.

2. Student co-construction as legitimate practice

Your MiniDevs/Mixiply work is a case study here, but it's not alone. Democratic schools (Sudbury model), student-led design in the Big Picture network, youth participatory action research in US urban schools. The common thread: students operating as designers of their own learning environment, not recipients of someone else's design. What makes these effective isn't "student voice" as a checkbox — it's the transfer of genuine decision-making power.

3. Indigenous and place-based education models

Māori medium education (kura kaupapa), First Nations land-based learning in Canada, Aboriginal two-way learning in Australia. These aren't alternatives to "real school." They're fundamentally different answers to the question "what is education for?" that predate the colonial model by millennia. They work because they ground learning in identity, relationship, and place rather than abstract content delivery. The ngā pou framework you and Simon developed at Newlands drew from this well.

4. National frameworks that create space without dictating practice

The NZ Curriculum (2007) remains one of the most sophisticated national curriculum documents globally. It gave you the language to describe what you'd been doing intuitively for decades. It proved that a system-level framework CAN authorise the kind of pedagogy you advocate. The failure isn't the framework — it's political interference from governments that don't understand what they built. The framework itself is a genuine success story, and the fact that schools like Newlands could flourish under it is the evidence.

5. Post-COVID alternative structures

Micro-schools (Prenda, Acton Academy network), learning pods, hybrid models. COVID didn't invent these, but it proved to millions of families that school-as-building is not synonymous with learning. The genie doesn't fully go back in the bottle. These models are small, fragile, and inequitably distributed — but they represent a genuine market signal that families will choose differently when given the option.

6. Industry as co-designer, not just employer

The Theta/Mixiply partnership you lived for seven years. P-TECH (IBM's grades 9-14 model). The Swiss/German apprenticeship system. What distinguishes these from corporate "partnerships" (which are usually sponsorships) is that industry participants actually shaped the learning design. Jim Taylor didn't donate laptops — he pivoted his entire platform based on what students needed. That's co-construction across the industry-education boundary.

7. Connected learning and affinity spaces

Mizuko Ito's Connected Learning framework. James Paul Gee's affinity spaces (which you built MiniDevs on). The research showing that interest-driven, peer-supported, academically connected learning produces deeper engagement than any classroom intervention. The internet didn't just provide access to information — it created proof at scale that humans learn powerfully outside institutional structures when conditions are right.

8. AI as dialogical thinking partner

This one is emergent and unproven at scale, but you're living the case study right now. Using AI not for content delivery or assessment automation, but as a genuine co-construction partner for thinking. Your Pop-Up Opus sessions, your Literacies framework development, this very conversation. The mechanism that matters: AI can meet a learner where they are — neurodivergent, expert-level, dialogical — without the institutional constraints that force standardisation. The risk is equally real: used badly, AI becomes the ultimate content-delivery machine. But used the way you use it, it's the most powerful learning technology since conversation itself.

An honest caveat: Most of these exist at the margins. Items 1-3 and 6-7 have decades of evidence. Items 4-5 are politically vulnerable. Item 8 is nascent. None of them have cracked the problem of scaling without diluting — which is arguably the central challenge in education innovation.

What jumps out? What's missing? And which ones do you want to fight me on?

Marianne:

I want you to look at a transcript of Bob Greenberg being interviewed. What does this interview tell us about what might be positively possible that we haven't looked at yet: ---

PASTED TRANSCRIPT: [Episode 262 | Lessons from Over a Decade of Video Interviews with Educational Thought Leaders]

Claude:

This is a goldmine, and it surfaces something my list of 8 completely missed. Let me break down what Bob Greenberg's story actually reveals.

The big thing I missed: Learning that needs no institutional permission

Bob didn't build a programme. He didn't get a grant. He didn't need a principal's approval. He followed his curiosity with a camera and built a 2,500-interview archive that's arguably more valuable than most education research databases. No institution authorised this. No institution could have produced it.

That's a ninth category my list didn't have: self-authorised learning ecosystems built by practitioners who stopped waiting for the system to catch up.

The 2,500-interview finding that should stop every curriculum designer in their tracks

Across decades and hundreds of prominent thinkers, when Bob asked "tell me about a teacher who made a difference," the answer was never about content. Not once did someone say "she taught me the capitals of all 50 states."

Every answer was about being recognised as a human being.

The quiet girl who "participates by listening." The kid whose belt was too tight. The coach who said "get tutoring and I'll keep you on the team." Joe Renzulli sitting on the bench in trouble, and a teacher saying "but you're a good writer."

This isn't soft sentimentality. It's consistent data across 2,500 interviews pointing to the same conclusion: the mechanism that actually transforms learners is recognition, not instruction.

Three specific things this interview surfaces that my 8 didn't capture:

"Everything is education" — Bob followed critical race theory because it was on the news. Then Asian hate. Then Japanese internment camps. Then the French Resistance. Then free speech. The boundaries between "subjects" dissolved because he was following genuine questions, not a scope and sequence. This is what learning looks like when curiosity drives it instead of curriculum.

Marathon persistence as learning model — His running metaphor is pedagogically precise. The 100-metre dash rewards natural ability. The marathon rewards persistence. Most school assessment is structured like a sprint: you either have the talent or you don't. Bob's entire life demonstrates what happens when you design for the marathon instead. This directly connects to your work on student-driven learning.

The organic referral network — Pasi Sahlberg says film Jerome Bruner. Bruner connects to Alan Kay. Molefi Asante introduces Cornell West and Henry Louis Gates. This is Gee's affinity spaces happening in real life, without a platform, without a school, without anyone designing it. One curious person following threads.

The challenge back to you:

My original 8 were all about structures — schools, frameworks, partnerships, models. Bob's story suggests the most powerful learning happens when individuals bypass structures entirely and just... follow the questions.

Here's where I want to push: Is the search for the right structure actually the wrong question? Your entire career has been about finding progressive schools with visionary leaders. Bob's fourth lap suggests another path — one where the practitioner simply stops asking for permission.

You're already doing this with your blog, your AI literacy framework, your Pop-Up sessions. Is Bob's story a mirror you haven't looked into yet?

Marianne:

It is an excellent question! And, no, I haven't made that connection. I'm going to embed the Brainwaves transcript for the video Bob made with me and my students. I was invited to participate when he started this project. He was a local bloke with a big idea. He also said I was the first one he invited that actually interrogated his purpose. I said I would only do the video if he captured my student voice. I have always hated my part of the video and usually just queue up to the kids portion, which is what I wanted captured. That being said, I have been humbled and felt privileged to be a part of what has become such a prestigious repository of people thinking on learning. He even created a whole section dedicated to Student Voice, which I really appreciated. So, your query about looking at his process makes me feel a little clueless that I didn't make the connection. But, then that is what your super power is. 555. Anyway, it was fun to hear the tables turn and the learning leaders interview Bob... who I only know through our own conversations. It was a treat to hear him interviewed ten years into the project and what his takeaways are. Good call. Here is the transcript from the video he did with me eleven years ago: ----

PASTED TRANSCRIPT:  [Marianne Malmstrom - The Cognitive Architect: Follow the Learning]

Claude:

This is a brilliant pairing. And I want to push back on your self-deprecation before we go further.

You didn't miss the connection because you're clueless. You missed it because you were inside the story. Bob's process looks obvious from the outside. From the inside, you were just saying yes to a local bloke with a camera. That's how the best affinity spaces work — you don't recognise them as significant until someone holds up the mirror.

Now, what this 2014 transcript actually reveals:

You did the same thing Bob did, just in a different medium.

Bob followed curiosity with a camera. You followed curiosity with a game server. Neither of you asked permission. Both of you let the learning lead.

Your pivot from "no gaming!" to running 24/7 Minecraft servers is structurally identical to Bob going from burned-out teacher to 2,500-interview archive. The mechanism is the same: a practitioner who stopped defending their expertise and started following what was actually happening.

The student voices in that video are doing something remarkable that I don't think you've fully claimed.

Those kids aren't giving testimonials. They're analysing pedagogy.

"In the class they try to stamp out your creativity... they just want you to follow the rules." That's not a complaint. That's a structural critique of institutional design from an 11-year-old.

The kid who built the Minecraft Institute of Technology and then gave you a report card — that's a complete inversion of the power dynamic that most "student voice" initiatives only pretend to achieve. That child designed curriculum, assessed a teacher, and articulated a theory of learning ("learning by teaching") more coherently than most education white papers.

And your decision to insist Bob capture student voice rather than just filming you talking? That IS your pedagogy in action. You weren't being modest. You were demonstrating the principle.

Read the transcript: "The kid [Reha] who gave her teacher [me] a report card—"

9:30

... so one of my favorite

9:33

experiences with MIT was when my teacher

9:36

my computer teacher decided to take a

9:38

class and I was teaching it and she

9:41

wasn't exactly the best student um very

9:45

disrespectful so when it was time to

9:48

write the report card I really wanted to

9:50

give her a "C" but I felt bad so I I tried

9:54

using interesting words like she was

9:57

interesting she was unique

10:00

and I gave her "A" minus though I don't

10:02

think she deserved it

https://www.youtube.com/watch?v=Pm0_09Z-wTY

Hear Reha voice her frustration

Claude:

And your decision to insist Bob capture student voice rather than just filming you talking? That IS your pedagogy in action. You weren't being modest. You were demonstrating the principle.

The thing your 2014 self said that your 2026 self needs to hear again:

"The single most important decision we made was not to design any curriculum for Minecraft until we had spent a great deal of time just sitting down and playing with the kids."

You described your role shift as moving from "conveying information" to "building learning spaces, giving challenges and constraints, and becoming co-learners."

That's exactly what you're doing with AI right now. Your Pop-Up Opus sessions, the Literacies framework, even these conversations — you're sitting with the technology, playing with it, refusing to design curriculum until you understand what's actually happening.

The medium changed. The pedagogy didn't. You said that yourself in 2014.

So here's my updated challenge:

My original 8 things working in education were all about structures and models. Bob's interview added a 9th: self-authorised learning ecosystems.

Your Brainwaves video adds a 10th that might be the most important one: practitioners who are brave enough to let the learner lead, including when the learner is themselves.

That's not a programme. It's not a framework. It's a disposition. And it might be the thing that's hardest to scale but most essential to get right.

The question is: can that disposition be cultivated in other educators, or is it something that only emerges from people like you and Bob who've had enough career runway to stop caring about institutional approval?

Because if it can't be taught, we have a serious problem. And if it can, that's your consultancy.

---

Jury is still out on Claude's last question, but the journey to discover what is next in my career continues to evolve.

Regarding Claude's suggestion, "7. Connected learning and affinity spaces"

In 2014, Mimi Ito invited students from the Elisabeth Morrow School to share their "Minecraft Institute of Technology" through Pursuitery, a project of the Connected Learning alliance. Evan Jones was the project lead at Pursuitery and did an incredible job of hosting this very enthusiastic, and sometimes chaotic, tour of an "online" school created to teach others how to play in Minecraft. The truly amazing thing is that students completely hatched the idea and built it on their own on the school's 24/7 server. They took me for my first tour after it was completed. To be honest, I was blown away. I had seen them design mini games, set up community projects and even build a working amusement park, but this was different. This was fascinating from a pedagogical perspective.

In Bob Greenberg's video, Reha describes her frustration when I pushed back on her concept of school and learning. Me acting the part of a naughty student really tripped her up. In her explanation of giving me a report card, you can see how the structure of school is so engrained in her perception of learning. And, to be fair, you can also feel her dilemma of not wanting to disrespect her "real life" teacher by giving me the grade I actually earned. To be fair, all of the students who built Minecraft Institute of Technology shared the same devotion to the structure of school, but with a Hogwarts twist.

The real learning can be clearly seen in this live tour as the students take us through their build and explain the architecture of their design and the systems they put in place to execute their vision of running a real Minecraft school.

It is such a great case study of what kids can, and are, teaching themselves and creating on their own. There was so much for educators to understand about learning when working in connected learning and affinity spaces.

https://www.youtube.com/watch?v=XB5GcUFnQaI

Claude's Analysis of the video transcript

Claude:

This transcript is pure gold as a case study. Here's what it actually demonstrates:

What you're watching is a functioning organisation designed and operated by children.

Not a simulation. Not a class project. A working institution with roles (lunch lady, headmaster, magic teacher, secretary), infrastructure (dorms, classrooms, library, exam rooms, cafeteria, sports facilities), scheduling challenges, copyright concerns, a self-publishing library, and a website coded from scratch by Reha.

They even had a staffing problem during the live broadcast — Snowkit kept cutting out — and they managed it in real time, just like any organisation would.

Three things this reveals that we haven't named yet:

Kids naturally build systems, not just content. They didn't make "a Minecraft world." They built an institution with enrolment processes, examinations, class scheduling, office hours, division of labour by expertise, and advancement pathways. They were doing organisational design. No one taught them that. They absorbed it from living inside institutions and then rebuilt it their own way.

Your pedagogical observation in real time is devastating. You pointed out on the broadcast that they built a school that looks like traditional school — including report cards. That's the same insight you gave me about Reha earlier. Even in a sandbox with infinite possibility, the institutional model is so deeply internalised that kids reproduce it. That's not a failure of imagination. It's evidence of how powerful the colonial model's imprint is.

The role fluidity is the thing most schools can't replicate. Chris builds. Reha codes websites and teaches magic. Snowkit writes backstories and organises. Power teaches mining. They move between roles based on expertise, not age or assignment. And crucially — you had to get in line to learn spells just like everyone else. That's not performed equality. That's actual distributed authority.

And one detail I don't want to let pass:

"Reha totally coded that page from scratch... New York Times just did an article about how to get girls into coding and mentors..."

You dropped that almost as an aside. But a girl who coded a website from scratch, designed curriculum, gave her teacher a report card, and taught magic classes on a 24/7 server is not an anecdote about "getting girls into coding." She's evidence that the entire framing of that problem is backwards. You don't need to "get girls into" anything. You need to stop building systems that push them out.