Fix the System to Give Most People Most of What They Want

Nearly nine in ten Americans — Republicans and Democrats alike — think members of Congress shouldn't be trading stocks in the companies they regulate. It may be the most popular idea in American public life. Last month the House finally voted on it.

What passed still lets members sell whatever they already own. It got bolted onto an unrelated fight about voter ID. It split down party lines anyway.

An idea almost everyone in the country agrees on, and Congress still couldn't do it cleanly.

The obvious explanation is that they're crooks — that people who profit from trading stocks wrote themselves a bill letting them keep trading. I held that view for years. I no longer think it's right, and what actually happened is worse.

Every person in that story can defend what they did. The members who voted against it had real objections. The ones who tied it to voter ID were running an ordinary legislative play, the same one every majority has run for two centuries. Nobody took a bribe. Nobody broke a rule. Nobody had to.

The system produced that outcome while working exactly as designed — and it does the same thing to almost everything.

That's what this document is about. Not any single policy. The conversion rate between what the country already agrees on and what the country manages to do.

Start with the claim everything else rests on: on most of what determines whether this country works, there is a deal most Americans would take tomorrow. We don't make it. That's the problem — not the disagreement, the failure to convert agreement into law.

Democrats and Republicans each guessed that roughly 55% of their opponents hold extreme views. The measured figure was about 30. Both sides were off by nearly double, in the same direction. Heavy consumption of news media is associated with a wider gap, not a narrower one — and education makes it worse rather than better.

The honest limit: this does not mean disagreement is fake. Americans have real, deep, non-negotiable value conflicts about abortion, guns, religion in public life, and what we owe people who aren't citizens. Those aren't misunderstandings and no institutional design will dissolve them.

The narrower claim is this: on a wide band of problems the public is far closer to agreement than the political system reflects. These aren't 51-49 questions. They're 70-30 and 80-20 questions that produce nothing. This is the load-bearing assumption of everything that follows: if there is no latent majority, institutional reform just produces prettier gridlock.

I am not proposing a comprehensive platform. That's a deliberate choice, because the instinct in politics runs the other way. Every serious agenda document I looked at while thinking this through tries to be exhaustive.

Project 2025 ran to roughly 900 pages, 30 chapters, agency by agency, 35 primary authors. It was an impressive piece of work on its own terms and it became a political liability, because a document that specific hands your opponents 900 pages of attack surface.

The opposite lesson comes from the Contract with America: ten bills, one page, a promise of floor votes within 100 days. The items were chosen because polling showed at least 60% support for every one of them. It was a promise you could check — and checkable promises are the only kind that build trust.

The AI Action Plan the administration released in July 2025 is worth studying for its architecture: three pillars, a short plain-language diagnosis for each, then a bulleted list of specific actions with a named actor attached to each one — around 90 actions in under 30 pages, readable in an afternoon.

The structural lesson from all three: be specific enough that people can hold you to it, bounded enough that you can finish it, and honest enough that your opponents quoting you back at yourself doesn't hurt. So: three pillars. Nothing else in the first term. The restraint is the substance.

A reader is entitled to ask whether a document diagnosing an institutional failure has quietly substituted its own policy preferences for Pillars Two and Three, since only Pillar One repairs machinery.

The honest answer: they stabilize the ground the machinery has to stand on. Immigration and retirement insecurity are not merely two items on a list of unsolved problems — they are the two forces that most distort how people vote. Fear about a border nobody controls and fear about not being able to afford to stop working are precisely what make a voter reachable by the candidate who promises never to compromise with the other side. Leave those two open and you have repaired the machine while leaving the load that bent it in the first place. Institutional reform takes a decade or more to take hold; these two buy it that decade.

And closing them proves the central claim. This document asserts that deals exist where both sides win at the same time. That assertion is easy to make and impossible to verify in the abstract. Immigration and retirement are the two hardest available test cases — the most emotionally charged and the most fiscally frightening — and a "yes, and" bargain plainly exists in both and has never been closed in either. If it can be closed there, the premise holds nearly everywhere. Both things are true at once, and neither has to come first. The repair and the proof are the same act. Full detail on all three repairs is on its own page: see The Repairs.

The order is part of the argument.

1. Publish the contract — three pillars, the specific mechanisms, the bargain in each, and the measures of success. Short enough to read. Specific enough to be held to.

2. Ethics and transparency first. These are legible, popular, and self-binding. Passing rules that constrain yourself before asking anyone else to sacrifice is how you earn standing for everything after.

3. Election reform through the proper channels — federal and state pathways, without pretending a president can rewrite election administration by decree.

4. Immigration as one integrated statute, with simultaneous implementation triggers, funding, deadlines, judicial review, and public dashboards. Not two bills. Not sequenced halves.

5. An independent fiscal commission with a fast-track vote — but with a politically explicit mandate reflecting the agreed bargain, rather than outsourcing value judgments to technocrats and pretending the answer was arithmetic.

6. Phase retirement and education finance over time, while delivering something visible early. Seeded accounts do that — people need to see one thing arrive to believe the rest is coming.

7. Then the deferred work: health care, housing, education, climate, AI.

Term limits, considered seriously and rejected. Between two-thirds and five-sixths of Americans support them — probably the most popular structural reform in the country, and they'd pass the adversary test cleanly. Leaving them out costs something.

They were left out because they make the central problem worse. Deals get closed by people who expect to deal with each other again; term limits truncate that repetition and end every career in a lame-duck term with nothing left to trade. Fifteen states have legislative term limits, and institutional knowledge doesn't disappear there — it relocates to lobbyists and permanent staff who stay when legislators don't.

The deciding argument is simpler: term limits are a workaround for elections that don't work. Fix the elections directly — as this platform proposes — and you don't need an artificial cap on service, because you have the real thing: voters who can end it whenever they choose. They would also require a constitutional amendment; states cannot impose them on Congress, per U.S. Term Limits v. Thornton (1995).

Then the subjects, rather than the mechanisms: health care, housing, education beyond the lending question, climate, AI. Some of these matter more than what was included — health care in particular has misaligned incentives across insurer, PBM, hospital, provider, employer, and patient that resemble the political failures described here, which is a reason to think it becomes tractable using the same lens, later. On AI, this platform holds the least confidence of all; it's the one place deferral might be a mistake.

But it keeps landing in the same place: none of this gets solved with a government that can't pass a budget on time, can't stop its own members from trading on their oversight work, and can't close a deal three-quarters of the country already agrees on. Not because the other problems are small — because the machine used on them doesn't work.