Father John, Oscar Wilde and the Waterloo Boy

2026-02-17

In 1967 Father John Culkin penned an article (blocked for extremism if you’re on the VPN) about his friend Marshall McLuhan. In which he said* “We shape our tools and thereafter our tools shape us.*” A famous quote often misattributed to McLuhan himself and often simplified as we become what we behold. Culkin and McLuhan were media theorists during the advent of the information age, spending their lives building philosophies about how technology impacts society. We don’t know what they might have conjectured about the deployment of large language models, but we can conjecture, and good conjecturing starts with simple questions.

What does large language model?

The sentence before Culkin describes how tools shape us in his 1967 article, he says, life imitates art. Culkin is quoting Oscar Wilde in his 1889 essay The Decay of Lying. Where Wilde builds a case for aestheticism—believing in art for art’s sake—through a dialogue between two characters, Cyril and Vivian. Vivian, the antagonist who doles Wilde's hot-takes, begins the story with this zinger;

Cyril (coming in through the open window from the terrace): My dear Vivian, don’t coop yourself up all day in the library. It is a perfectly lovely afternoon. The air is exquisite. There is a mist upon the woods, like the purple bloom upon a plum. Let us go and lie on the grass and smoke cigarettes and enjoy Nature.

Vivian: Enjoy Nature! I am glad to say that I have entirely lost that faculty.

The Decay of Lying is a text that LLM’s have been trained on. So is the article about McLuhan from 1967. If it’s open domain and on the internet it’s training material. The Bitter Lesson is an idea in AI research that says “in the long run, approaches that scale with available computational power tend to outperform ones based on domain-specific understanding because they are better at taking advantage of Moore's law.” A large language model needs as much writing as it can get. As much content from us as it can get. As much us as possible.

So, what does large language model? Us.

LLMs are a mirror of us. It’s not a mirror of you, it’s a mirror of everyone. It can be funny because it trained on Oscar Wilde making a metaphysical joke about nature and self reflection. It can theorize because it trained on McLuhan building philosophies about what it would do to us. It can write code because you wrote some and put it on Github.

The Bitter Lesson tells us what we have already felt. LLM is the scale of hundreds of us. It gives me specifically, as a human trained with design expertise, the engineering ability to code and deploy apps in days that would have previously taken me months. It helps me write faster by helping me work through ideas faster. If computers are the bicycle of the mind, the large language model is a plane.

McLuhan said our tools shape us. Oscar Wilde said art shapes the tools. The Bitter Lesson says Moore’s Law shapes everything. Which means this can’t be the first time we’ve had a new tool represent a magnitude of cultural change we hadn’t previously felt. One comparison, after joking with a few colleagues last week that Cursor feels like riding a John Deere when I’ve been used to digging holes, pulling weeds and tossing rocks myself.

After 4 prompts, here’s the story Claude wanted to tell me.

In 1900, an American farm was a small civilization of specialists. The plowman understood soil. The teamster managed the horses. The thresherman arrived each harvest with his crew and his steam-powered machine. Binders, shockers, gleaners, blacksmiths, farriers. Each role existed because the work demanded it. Specialization wasn’t an org chart. It was a natural response to complexity. Then in 1918, Deere & Company bought a small outfit in Waterloo, Iowa, and started selling an ugly, two-cylinder kerosene tractor called the Waterloo Boy. It wasn’t better than any individual specialist. It just made most of them unnecessary. One person on a mechanical seat could do what had previously required a coordinated team with distinct expertise. The machine absorbed the skill.

The economics forced adoption. After World War I, crop prices collapsed and debt crushed small farmers. Mechanization became survival, not luxury. Each generation of equipment, the Model D in 1923, the combine harvester, the mechanical corn picker, erased another specialist. The thresherman disappeared when the combine did cutting and threshing in one pass. The teamster disappeared when the horses were sold. Between 1900 and 1950, farm employment dropped from 41% of the American workforce to roughly 12%. That’s not a reshuffling. That’s an extinction event for an entire way of organizing human labor.

The same pattern is unfolding now in product development. For twenty years, building software has required its own ecosystem of specialists. The designer who understands users and mental models. The engineer who understands systems and scale. The product manager who understands markets and sequencing. Each knows something the others don’t. The roles exist because the work demands them. Now AI shows up, and like the Waterloo Boy, it doesn’t need to be better than the specialist. It just needs to be good enough to collapse the roles. A product manager who can prompt AI into a working prototype doesn’t need to wait three sprints for engineering. A designer who can generate production code doesn’t need to hand off a Figma file. The machine is absorbing the skill.

The optimistic version says everyone “levels up.” Shed the rote work, focus on strategy, vision, taste. And maybe that’s true for some. The best farmers in 1950 were genuinely more sophisticated than the best farmers in 1900. They understood soil chemistry, crop rotation, and market timing in ways their grandparents never needed to. The tractor raised the floor of what a single competent person could accomplish. But there were far, far fewer of them. The floor went up and the number of people standing on it went down.

The question for product teams isn’t whether AI will change the roles. The Waterloo Boy answered that a century ago. The question is whether you’re the farmer who learns to drive the tractor, or the teamster still perfecting his voice commands for a horse that’s about to be sold. The machine doesn’t care which one you choose. It just keeps running.

My immediate reaction while organizing this story from Claude was bleakness. It’s hard to imagine there aren’t parallels. And it’s hard to imagine there’s an important direction to go from here. Machines were and are always going to replace the work that we can do. That’s the tool the shapes us. But it doesn’t have to stop there.

My hope is to find a way to define this idea better than “you can just build things”. We, us, the mirror that made this mirror, are bigger than defining our outcomes as output. The way we decide what’s good and what isn’t good matters. We need something more than a machine to do that.

Ideas

Junk

https://read.lukeburgis.com/p/mimetic-desire-101

Below is a collection of edits I removed from the piece while writing. I like keeping this around to see the ways my brain was organizing.

My father, like McLuhan, is Canadian. My father, unlike McLuhan, never said the medium is the message, but my father often said, usually while my brothers and I were playing Golden Eye on N64 or while stuck in traffic on I96 on the way to a hockey game in Dearborn, MI, “An object at rest stays at rest, that’s the first law of newtonian physics boys, gotta move.” Our physical world is organized by forces. Our cultural world is organized by forces too. I don’t think McLuhan was describing mimetic desire when he said the medium is the message, but I do believe he was agreeing with Oscar Wilde with his anti-mimetic catchphrase “Life imitates Art far more than Art imitates Life”. Ideas move in many directions through society to create forces. Which forces are shaping the way in which we move?

and is summarized in his 1964 seminal work Understanding Media: The Extensions of Man, where he famously states the medium is the message.

Though decades apart, both take the same anti-mimetic stances in their philosophies of aestheticism, that tools aren’t extensions of the human experience, the human experience reflects back the tools that shape it, ultimately,

Vivian, through witty exchanges and paradoxes, spells out Wilde’s philosophical stance on aestheticism, called anti-mimetics. Anti-mimetics is, according to ChatGPT, the view that art, media, and aesthetics don’t copy reality but actively shape it. Wilde approached this idea through the lens of art. McLuhan approached this stance through technology and communication. Anti-mimetics is an important lens to approach the development and deployment of new technology in understanding how they shape our work and therefore, our world. To bring this back to designer terms and summarize Jun’ichirō Tanizaki’s ideas from his 1933 book In Praise of Shadows, reality is shaped by form.

So what does large language model?

McLuhan applied his ideas to changes he saw during the dawn of the information age, but what do these ideas mean for us during the dawn of what Sam Altman calls, the intelligence age. If an anti-mimetic stance served noticing the movement of cultural tide in 1890 and 1930 and 1960, what can it tell us about how large language models will shape our reality?

Human —> idea (tool (desire)) —> human

Human —> AI (medium) —> work —> screen —> work —> AI —> human

Human —> LLM —> human

If large language models can only look backward, how do they help us move forward?

New ideas come from the collision of messy people making messy choices and rearranging things that already exist, like Emily Dickinson and the em dash.

When ChatGPT interferes in the force of communication between two humans, is that ok? Is it ok that the primary use for LLMs so far is relational therapy?

When helping hurts. Why does AI feel cringe in human interactions when it’s noticed?

Three instances of AI feeling weird

Em dashs on a consultant influencers Linkedin posts
Friends singing a tiktok meme song not knowing it was AI
Conclusion speech to a vacation being edited (written?) by AI

We talk a lot about AI interaction improving our work, that represents a certain direction AI can move. But what does it feel like when AI moves between people? What makes our connections to someone special and does AI feel like an invasion of that?