Elon's million-dollar writing prize was of course a scheme

On missions and anti-missions

The Architect’s Dream, by Thomas Cole (1840)

Not long after I announced my own $10,000 essay prize, coined as the “world’s largest open¹ essay prize,” Elon launched his own: $1,000,000 for the best article on X. Immediate deflation. I did not expect the world’s richest man to suddenly champion longform writing. A 7-figure essay prize was a ridiculous goal of mine, one I thought would take a decade to exist, but for the equivalent of 14 cents in Dean-adjusted-dollars, he launched it with the care of a whack. After the shock faded, I tried to be open. This is historic, right? An unreasonable sum of money will be granted to, not an essay, but at least a written thing over 1,000 words, a piece to “move culture,” untangled from gatekeepers, open to anyone (anyone who is a premium X subscriber…). So, I did what any fellow essay-prize host would do and burrowed into the terms and conditions, and unsurprisingly the altruism breaks down the further you dig.

The first thing to poke into is the criteria: after writing a 25,000 word textbook on what makes a good essay, and maybe another 10,000 on how we’ll determine the winner, I knew the WINNER DETERMINATION clause was where the real architecture of the prize lived. I found 16 words, summarizable in 3 points: originality (undefined), spelling (what?), and “Platform engagement (i.e., verified Home Timeline impressions).” I.E., did it go viral? Despite the terms saying “no politics,” that absolutely seemed like the deciding factor. The winners each had an Elonic nature: on the corruption of Deloitte’s government contracts, on COVID scams, on tariffs, on Minnesota. This is not abnormal though; the typical prize is after all just a scheme to promote an agenda, not writers.

However, most prizes aren’t openly evil enough to tell you they intend to steal from and destroy their own contestants. See Term 3, 118 words, but compressed here to break the shield of legalese:

“By submitting… [you] irrevocably grant… X and its respective parent companies… the right to use Entrant’s… likeness, voice, persona… (collectively, ‘Personality Rights’), in whole or in part, in perpetuity, throughout the world… for editorial, advertising, trade, promotional, and commercial purposes… without further notice or compensation.”

Similar to how the mafia operates through mattress storefronts, this looks like data acquisition dressed as patroning the arts, a scheme to train Elon’s xAI but veiled as a technoliterary prize. It’s also a bargain. OpenAI and Google are paying $60-70 million per year in licensing for trillions of unstructured, bot-infested words from Reddit. Now, here, for roughly $2.17m or so (there were bonus prizes), xAI is getting tens of millions of words of novel, human-crafted, engagement-ranked content—the exact thing you need to fine-tune a frontier LLM.

I share this to help frame a question that’s top of mind—now that The Best Internet Essays 2025 is shipped, and Essay Architecture is done with its first loop of projects (a textbook, a library, a club, an app, a prize, an anthology), it feels time to cohere this into a business—: how do you grow and sustain something that is actually mission-driven and incorruptible by the market?

Mission statements are usually a farce. After merging with SpaceX, xAI’s latest mission statement is “to build a sentient sun to understand the Universe and extend the light of consciousness to the stars!”—beautiful, but to colonize space, they first need to colonize this planet’s attention. To “elevate the world’s consciousness,” WeWork needed a few real estate scams. To “bring the world closer together,” Facebook needed to psychographically profile you and sell your data to foreign governments. When you squeeze a lofty ideal through the narrow pipes of TAM (total addressable market), TUM (total user milliseconds), and CLV (customer lifetime value), the mission doesn’t just die, it risks becoming a wretched inversion of itself: an anti-mission.

If my innocent mission² of “let’s use technology to reinvent how we teach writing and to help discover great essays” wasn’t earnest, it could easily be twisted into “let’s steal from writers to create autocompleted slop for professionals who don’t have time to write and probably hate writing anyway,” because that’s where the money is. That’s where the market leads if you willingly blind yourself. Techniques can be spun towards many mercenary ends, but the thing to never compromise is telos, the purpose behind whatever you do. Personally I’d rather become a great essayist/teacher with a modest company that sustains me through life, creatively and financially, than a constipated aspiring unicorn with a hyperscale slopcorp that contributes even slightly to the slowrot of everything.

I think what I’m trying to build is an institute. I see this as an entity that exists to serve a particular mission, and it inverts the many modes of the growth-centric creator (contributions not content, ecosystems not niches, members not users, legacy not liquidity). The thing is, an institute is usually supported by endowments, donations, foundations, tuitions, grants, exits, and such. I wonder if an AI-enabled solo-operator can now have the civic impact of what used to require millions of dollars in old money.³

The meta-mission of Essay Architecture is something like “use AI to protect the essay through the age of AI.” It is both poison and cure. A friend called this “mithridatism,” a practice from a paranoid ancient Hellenistic king, who self-administered small doses of poison to protect himself against a lethal dose.⁴ AI will probably be the largest ever force of institutional collapse, but also it enables us to rebuild micro-institutes that outlast the ones already falling. It’s all about knowing the limits and edges, where to use it and when to avoid it. What is worth building towards?

Some more thoughts on institutes:

---

Updates and announcements

I’m starting another newsletter, a development blog that shares (1) new features and prototypes for the essay editing app I’m building, and (2) the musings, mishaps, and epiphanies of a writer-builder. If you’re interested, just sign in to the Essay Architecture website (updated on Friday BTW).

(An example of something I’ll be shipping)

---

Some reading recommendations:

In the spirit of trying to keep up with writers on Substack, I figured I’d try curating some of my favorite reads published from the week:

• The Kinds of Short Stories People Really Want to Read by Isaac Kolding

• xerox’d flowers by visakan veerasamy

• I need whimsy but I also need to be taken seriously by Sherry Ning

• A Baudelairian rube strolling down a small town sidewalk by Caleb Caudell

• Reading is magic by Sam Kriss

• AI Populism’s Warning Shots by Jasmine Sun

• Every Resource I Use to Educate My Kids by Latham Turner

• How to Walk Through Walls by Henrik Karlsson

---

---

Other notes

---

Footnotes:

1. An “open” prize is one without entry conditions: no demographic qualifiers, no requirements to join a university or publishing deal, and no prompts that narrow the topic too tightly. When a prize has these constraints, it’s more likely to promote an agenda; without them it’s more likely to promote the medium. Technically Elon’s prize was open by this definition, but focused on articles not essays, and of course mired by the whole sacrifice-your-voice-forever clause.

2. I figure it’s required to share my existing mission statement in an essay about missions. This was written in September of 2025. I’m uncertain if it’s too bombastic or not bombastic enough. Here’s a gDoc if you have feedback. I’d like to rewrite a v2 so the voice is more personal and less institutional, maybe frame it around the “micro-institute” idea, and then probably go galaxy brain and forecast a hypothetical timeline on how writing could evolve through 2100, along with ideas on how to use technology to preserve/expand the culture around essays.

3. Almost 100 years ago, Buckminster Fuller embarked on his own mission, and in reading how he reflected on it in 1982, it now seems more possible than ever for the average person to think and dream at the scale of culture:

“I undertook to see what a penniless, unknown human individual, with a dependent wife and newborn child might be able to do effectively on behalf of all humanity.”

A final anti-mission to note: all the AI labs have some version of “to benefit all of humanity” in their statement, but first they need to commandeer the entire economy.

4. While King Mithridates was able to avoid assassination, the irony is that he was eventually cornered by a revolt and unable to poison himself, leading to a more brutal death.