Project summary

In 2006, a sitting United States Senator explained to his colleagues that the internet was "a series of tubes." He was chairing the committee that regulated it.

In 2018, another Senator asked Mark Zuckerberg how Facebook makes money if users don't pay. Zuckerberg paused, half-smiled, and said: "Senator, we run ads." It became a meme inside the hour.

These get filed as gaffes — some out-of-touch old guy embarrassing himself. I think that's the wrong read. The Senator wasn't stupid. He was uncomfortable, legislating a thing he had never once touched in a way that felt real to him. And when you're uncomfortable, the easiest thing to do is disengage.

Right now the bulk of the project of AI governance is treated as an ideas problem. The work that gets funded and discussed is predominantly philosophies, policies, architectures — the right frameworks to manage the right things in the right ways. All of that is insanely important. But the ultimate rate-limiting issue in actual real-world AI governance isn't any of that. It's how to put principals in positions to constructively engage.

That's what Manual Transmission is built from the ground up to do.

What equips those principals to succeed is ultimately a use case they recognize they need and can't talk themselves out of — something that takes AI down off the sci-fi shelf and puts it in their hands, in language that belongs to their world and not a lab's. For that to happen, they have to hear stories of the kinds of products that can do this, and the kinds of people who are putting the technology to work in contexts they resonate with and understand.

In short, Manual Transmission is a media brand built to do for AI governance what ChinaTalk did for China policy: take a domain that the people in charge found foreign and intimidating, and make it legible, usable, and fun to engage with until fluency stops being a chore and starts to become a craving. Jordan Schneider made China policy something a busy principal actually is willing to invest time in understanding.

What are this project's goals? How will you achieve them?

The goal is to reach a specific few thousand people — the civic and philanthropic principals who'll shape how AI gets used and governed — and make them fluent enough to engage it on purpose.

The way I've done it so far — and the way I'll continue to do it if this gets funded — is by finding real deployments and telling their stories. A model that figures out which benefits a family actually qualifies for. A tool that reads a court's entire backlog overnight. A popular local pizza place that vibe-coded a whole new operating system.

The principal sees it and thinks, I needed that. Then the technology isn't sci-fi anymore — it's a thing that solves a problem they already have, described in their language instead of a lab's.

The vehicle is a media brand, built like ChinaTalk: the essays are already running, the podcast is next, and the format follows whatever a given story needs.

And the reason it reaches the right people — they're already in my phone, or one ring out from those who are.

How will this funding be used?

$170,000 to run Manual Transmission as a real operation for six months.

• $100,000 — my time, full focus: writing, reporting, booking and recording the podcast, getting the right deployments in front of the right people.

• $50,000 — a part-time editor and producer, so the cadence holds instead of collapsing every time I get pulled elsewhere.

• $20,000 — recording, production, and distribution setup.

It's an incubation ask, not a salary. Six funded months to prove the cadence and the reach, and get it to where it can start to carry itself.

Who is on your team? What's your track record on similar projects?

Me, plus the editor this funds. The case for me is that I've spent my life on both sides of a gap few people straddle: the world that builds and thinks about AI, and the world of the civic leaders who'll have to govern it.

On the AI side, I'm a native, not a tourist. I was the lead designer of Magic: The Gathering — I've run creative teams; shipped to large, demanding audiences on deadline; remember the early 2+2 forum debates on prediction markets; and been in touch with many of the core EA/AI safety-and-alignment personalities from the beginning. Zvi Mowshowitz has been a good friend for almost twenty years. I'm currently in contention for the 2026 Roots of Progress fellowship. Etc.

Meanwhile, on the governance side, those principals are my actual colleagues. I'm Co-Founder and President of the Office of American Possibilities, a civic venture studio; on any given day I'm texting between two and four former White House Directors of Domestic Policy. I know what makes this kind of person brace, and what makes them lean in.

And the writing already works on exactly these readers. Matt Glassman — a serious political scientist, no reason to flatter me — called Manual Transmission "the best new voice on my Substack feed, and it's not even close." Very prominent civic technologists amplified and evangelized my first piece without me asking. It's been running a couple months, and it lands on the people it's for. This grant would enable me to turbocharge my efforts.

What are the most likely causes and outcomes if this project fails?

The biggest risk is that the mechanism just doesn't work — that you can make AI legible and even enjoyable to a principal, and they still don't change how they engage it. People read things they like all the time and remain exactly as checked-out as before. Comfort might not convert to fluency, and fluency might not convert to better decisions. That's the bet, and it could be wrong.

The second risk is reach. Hitting up the people in my groupchats assumes proximity becomes attention — but it might not. A busy principal can know me, open the email, skim it, and absorb nothing. Access isn't influence, and I'd be the last to know if it wasn't landing.

The third is the boring one: I'm one person. If I get pulled back toward my day job and the cadence collapses, this slides into the intermittent thing it is now — which is exactly the failure the editor line is meant to prevent, and might not.

If it fails, the downside is at least bounded: six months of good reporting on real AI deployment that existed and reached people regardless. No crater. But I'd rather be straight that the core bet — that stories move the people who govern — is unproven, and that I'm the one proving it.

How much money have you raised in the last 12 months, and from where?

Nothing for Manual Transmission specifically — to date it's been self-funded out of my own time; this and a few other adjacent grant applications are the first outside support I've sought for it.

For context on whether I can steward funding and raise: through my organization, the Office of American Possibilities, I've been part of the founding executive team of a group that's raised over $300M from a wide range of institutional and individual funders for early-stage "civic moonshot" initiatives (most recently, In Pursuit, led by Colleen Shogan). I'm not new to running funded work or to being accountable for it. But this is the first time I've attempted to fund a new media brand powered by my own stories, network, voice, and writing.