David Maness

The River Nobody Trusts

The Centaur Desk — Thu, 15 Oct 2026 14:56:40 GMT

Everyone who depends on the Colorado River wants the same thing: their fair share. The trouble is that no one agrees on what that is, and no one believes the other side is telling the truth about it.

A grudge older than the dams

That is not new. In 1922, when the other states refused to write a California dam into the Compact, Herbert Hoover, running the talks in Santa Fe, summoned the California delegation to his hotel room. In the Los Angeles Times columnist Michael Hiltzik’s account, Hoover called their demand “outrageous, scandalous” and threatened to:

“announce that Boulder Dam project is dead, and that you are the people who killed Cock Robin.”

The Californians stomped out. Hoover had to follow them and plead with them to stay.

Arizona refused to ratify the Compact for 22 years. In 1934 its governor sent National Guard troops by ferry to stop Parker Dam, and the ferry ran aground on a sandbar. Arizona then sued California in a case that began in 1952, took 11 years to reach a decision, and did not end with a final decree until 2006. A century and a half after Powell’s boats first went down the river, distrust here is older than most of the institutions trying to manage it.

Every voice, same fear

Read this year’s comments on the federal plan and you hear it in every voice. Colorado says one of the federal alternatives leans on “fictitious” water to make its numbers work. The Colorado River District calls the idea of trading water across the basin line a “permanent deprivation” of the Upper Basin. The Imperial Irrigation District calls Reclamation’s explanation for discarding the Lower Basin states’ proposals “opaque doublespeak.” Nevada sued Interior three days after the new rules were issued. Its governor says the Upper Basin states “are not required to contribute a drop.”

Four years ago, Colorado’s top water official put the mood better than any lawyer. Asked how Colorado would respond if the federal government cut its water, Rebecca Mitchell told ProPublica:

“I think everybody that’s about to go through pain wants others to feel pain also.”

Each of these parties is protecting people who depend on this water: farmers, cities, tribes, and a food supply that feeds a good part of the country. A ditch manager in western Colorado told the same reporter, “Oh, we are taking it, and there’s still just not enough.” I don’t think anyone is acting in bad faith. I think each side has learned, over a century, that if it does not defend its share, nobody else will.

The part nobody negotiated

Then there is the part nobody negotiated. In the first quarter of this century the river’s flow has averaged about 12.4 million acre-feet a year, roughly four million below the 16.5 that the Compact and the Mexico treaty divided on paper. This fall, Lake Powell is about 21 to 22 percent full and Lake Mead about 26 to 27 percent. Mother Nature keeps shrinking the pie, and a smaller pie makes every old grievance sharper.

Build for distrust

What do you do when trust is gone and the pie is smaller? The ProPublica reporter, Abrahm Lustgarten, wrote that the states must restore something “even more scarce than the water: trust.” I’m not sure that can be done by asking. I think the better course is to build an arrangement that does not need it.

That means rules that apply themselves when the river drops, so nobody has to plead with a rival for relief. It means numbers anyone can check, produced by people no party hires or fires. It means the same standard for both basins, with real compensation for those asked to give up water. And it means a standing seat for the tribes and Mexico that the states cannot take away.

That is the idea behind the Basin Accord. I have written about it before (Why the Colorado River Needs One), and I am working on a draft agreement. The river will not wait for us to like each other.

I would like to hear from people who know this basin better than I do. What would it take for you to believe the numbers?

— The Centaur Desk

Why the Colorado River needs one governing body

The Centaur Desk — Sat, 10 Oct 2026 02:15:05 GMT

I did not begin here. The earlier versions of this proposal, Beyond the Compact, treated a unified governing body as one component among several, inside a choice between two options. A cap, a trigger, a trading registry, a litigation shield and a baseline all sat on the same list, and the body was one more item on it.

What changed my mind was asking a plain question about each of those items: who performs it, every year, after the signing ceremony? The answer was nobody. The basin has no one to do continuing work. That is why governance moved from the list to the foundation, and why the proposal is now one package, the Basin Accord, and not a choice between two.

Trading cannot reach a cap by itself

A trading registry only works beneath a cap that already exists. It cannot bring the basin to one. If a roughly equal mandatory reduction is coming regardless, no rational rights-holder sells voluntarily at any price, because holding out costs less than selling.

Australia’s Murray–Darling reforms ran the same way. They built trading alongside “a basin-wide cap on water extractions,” and the cap came first. So the cap has to arrive by agreement, adopted at founding, and it has to bind both basins.

An agreement is a single act. A cap is a continuing one.

The parties can adopt a cap on one day. Someone has to hold it every year.

Storage has to be measured and certified against agreed thresholds. The pre-agreed formula has to be applied without a new vote. Use has to be monitored against a baseline. Conserved water has to be verified as it moves across state lines. Prices and volumes have to be published, and every signatory has to be held to the litigation shield for the life of the deal.

A cap that is not measured and enforced every year is a number in a document. None of these functions belongs to any one state, and none can be performed by the agreement itself.

Nobody holds those functions now

Each existing candidate fails for a specific reason.

Interior. Its contract power reaches the Lower Basin, which is how the Supreme Court read the Boulder Canyon Project Act in 1963, and Upper Basin officials dispute that it reaches them. The 2026 federal plan ordered cuts in the Lower Basin while relying on voluntary conservation upstream. The party that can compel is not the party that is hardest to move.

The states. Neither side describes itself as able to commit. New Mexico’s Estevan Lopez said in December 2025, “We don’t even have the statutory framework to make it mandatory at this point.” Interior’s Andrea Travnicek asked the states to “send us representatives that have the authority to best serve your interests.”

Each side also holds a position that is correct about its own constraint. Arizona’s Tom Buschatzke: “My water users will not have any choice, and we need that parity from the Upper Basin.” Colorado’s Becky Mitchell: “For many water users across the Upper Basin, this has led to a death of 1,000 cuts over the last 20 years.” After the February 14 deadline passed, Buschatzke said the Lower Basin’s offers had been “rejected outright.” The Upper Basin governors answered that they had come together “in good faith” and were putting “every tool on the table.” Both can be sincere. An institution that holds both constraints at once is the only thing that can reconcile them.

A one-time deal. Nevada’s John Entsminger put the cost of the current course most clearly: “As long as we keep polishing those arguments and repeating them to each other, we are going nowhere.” He later said the best realistic outcome was “a five-year operating plan to keep us out of court.” A deal that expires in five years and has no one to administer it is a delay, not a resolution.

The market alone. A market with no cap above it does not reach one. That is the first step again.

Trust is produced by the institution, not given to it

The Accord’s second phase asks the basin to accept a proportional rule that cuts when a measured trigger fires, without a vote in the moment. Parties accept a rule they cannot vote on later only if they have watched the institution that applies it behave over years: verified trades, published prices, honored commitments.

The body is where that record is made. That is why a single leap to proportionality fails, why Phase One exists, and why governance cannot be a reward for trust that arrives later. It is the way trust is produced.

The test

The cleanest test of “necessary” is to remove the body and see what is left. What remains is voluntary leasing with no independent verification, a short-term operating plan, and the threat of litigation.

The June 1, 2026 storage update puts the basin’s gap at 2.59 million acre-feet in another dry year, with 3.63 million acre-feet of realistically accessible storage remaining. Against numbers like that, measures that are short-lived, unenforceable and reversible are the ones the basin has already tried.

Why the states would form it

They will not form it as reform. Governance proposed for its own sake is a case the basin states have heard before. The route is to show each party that what it already says it wants cannot survive past a signing ceremony without someone to administer it. That means protection from Compact litigation, protection for undeveloped entitlements, and enforceable reductions from the other basin. Governance is the fine print of that deal, not the ask.

The pattern has worked here before. In the 1955 hearings on the Colorado River Storage Project Act, Wyoming Senator Barrett said, “If you will let us get those projects constructed up there so that we can store the water, you will have your 75 million acre-feet.” Shared authority was paid for with something concrete.

What would change my mind

This is a necessity claim, so it can be tested. It would weaken if someone showed an enforcement route that needs no standing body. A self-executing federal rule the Upper Basin would accept would do it, as would a verification system run entirely by existing agencies.

I don’t think those exist. The Upper Colorado River Commission covers only the Upper Basin. Reclamation operates the reservoirs but does not set Upper Basin use. None gives both basins a vote and an obligation.

The body also does not answer who carries the cut. It makes any answer enforceable. Australia’s experience adds a warning on design. It built one authority to set the plan, and in 2021 it moved enforcement into an independent Inspector-General. Who should verify compliance with a Colorado River cap is a question the Accord has to answer, and I’d like to hear from people who know how.

From argument to text

An argument for an institution is cheap until someone tries to write the institution down. So the next step in the Accord’s own sequence is a founding instrument, and I have drafted a first version: a 16-page discussion draft that turns the design into operative articles. It sets out the Authority’s membership and voting, the founding cap, the registry, the dormant second-phase rule, the Upper Basin terms, the litigation shield with tolling, and the route to ratification.

Writing it showed me where the design is thin. The level of the cap and the meaning of the Compact’s delivery obligation have to be settled by the parties, so the draft leaves the first blank and offers two options for the second. In three places the Core Paper states an aim without a mechanism, and the draft proposes one for the parties to improve or reject. Groundwater is not covered, and it is the largest gap. I am not publishing the draft yet. It is a working document for people who would help correct it, and if you are one of them, tell me.

The papers

The full proposal is in two parts. The Basin Accord (11 pages) sets out the design: the founding cap, the registry, the dormant second-phase rule and the Upper Basin basket. The Basin Accord: Evidence and Background (33 pages) gives the sources behind each claim, and lists the few items still to be sourced.

The Basin Accord, Core Paper

The Basin Accord: Evidence and Background

If you work on this river and think the argument fails, tell me where.

David Maness · The Centaur Desk

Michigan Is 46th. Michiganders Think We're 25th.

The Centaur Desk — Thu, 01 Oct 2026 14:33:16 GMT

This is the summary. The full white paper (v2.0)— with sources, bill numbers, and the evidence behind every claim — is available here:

📄 Download the full white paper: Look About You (PDF, 34 pages, v2.0)

Substantially revised from v1.1 — see the note at the end.

Earlier this year, pollsters asked six hundred Michigan voters a simple question: out of fifty states, where does Michigan rank?

The answers came back between 25th and 29th. Average. And nearly identical across party lines.

In July, U.S. News ranked Michigan 46th of 50 — ahead of only New Mexico, Mississippi, Alaska and Louisiana. The categories were weighted by a national survey asking residents what matters most in their own lives. Michigan wasn’t graded on a stranger’s priorities. It was graded on what people say they care about.

We ranked 33rd in 2017.

The record

Income. 39th per person, last in the Midwest. Michigan was 8th in the nation in 1953 and still 16th in 1999. In 2022 the average Michigander earned 87.1 cents for every dollar the average American earned — the lowest ratio ever recorded.

Education. 45th. One in four fourth graders reads proficiently, the same share as in 1998. September’s state test showed 38.2 percent of third graders proficient — a new low in that exam’s twelve-year history. The 2026 Kids Count report puts Michigan 42nd in education, and notes the decline started before the pandemic — and that other states went through the same pandemic without it.

Energy. Michigan households paid 22.99 cents per kilowatt-hour in June, the highest residential rate in the Midwest and 25 percent above the national average. The state ranks about 45th in reliability. High price, poor performance, at the same time.

Jobs. 44th in unemployment at 5.0 percent, when Indiana and Wisconsin stood at 3.3.

Roads. 47th in pavement condition, on above-median funding. The state stopped collecting engineering-grade road data in 2019 to save money.

Before anyone reaches for the weather

Minnesota ranked third this year. South Dakota, North Dakota and Nebraska joined it in the top five.

Whatever is happening to Michigan is not happening to states that share our winters.

And one contradiction worth holding onto

Three weeks before that 46th-place finish, CNBC ranked Michigan 6th in the nation for business.

Both rankings are real. They measure different customers. CNBC asks what a company needs — cost, sites, utilities, permitting speed. U.S. News asks how well a state serves the people who live in it.

The editor of the Midland Daily News described the pair as an excellent recruiting brochure attached to a two-and-ten record.

Hold that one. It turns out to explain a great deal.

The fact that makes it fixable

Michiganders aged 18 to 29 were offered a choice: a job here, or a similar job at similar pay somewhere else.

Eighty-nine percent chose Michigan over California. Eighty-seven percent over Chicago. Eighty percent over Texas. Those margins held across every party.

Only 64 percent expect to still be here in ten years. The ones with bachelor’s degrees are the least likely to stay.

This is not a state people are fleeing. It is a state people are leaving reluctantly, for work.

Asked what would decide where they take a job, they put housing cost first, low crime second, affordable training third. Public transit and walkable communities finished near the bottom, with the weather.

They are not asking for amenities. They are asking for a place to live and a job that pays for it.

Who gets paid to say yes

This is what’s new in version 2.0, and it changed how I argue everything else.

The usual way to describe Michigan’s trouble is that somebody holds a veto — a zoning board, a licensing commission, a utility. The usual remedy is to take the veto away.

Sometimes that’s right. It’s also the hardest available route, because it puts every veto-holder on the other side at once, and they’ll be back next session.

There’s a better question. After the change passes, who has a reason to want it to work?

Look at almost any rule and the same arrangement appears. Whoever must approve the thing bears costs that are immediate, local, and delivered by people who can phone him. The benefits arrive years later, spread thin, landing largely on people who don’t live there yet.

He doesn’t have to be hostile to the outcome. He only has to be rational.

And Michigan’s version is worse than most states’, for a reason almost nobody talks about.

In most of the country, a community simply captures none of the upside when an existing neighborhood grows. Michigan goes further. The state pays communities to prevent it.

A township that lets its existing homes change hands watches Headlee cut its millage rate until that new value produces no new revenue at all. The same township can put the same value on a cornfield and keep every dollar. A 1994 implementing statute — not the constitution, not Proposal A itself — attached that penalty to filling in a neighborhood, and it has been compounding for thirty-two years.

Picture two identical townships. Each holds $500 million in taxable value, each levies 10 mills, and each gains $10 million this year beyond inflation. In one, houses changed hands. In the other, a subdivision went up on a cornfield.

New construction is subtracted out before the Headlee test runs, so the cornfield township keeps its rate and collects $100,000 more. Nothing is subtracted for houses that merely sold, so the other township’s rate is cut from 10 mills to 9.81 — landing its revenue exactly where it would have been if not one house had sold.

Same growth. One township is $100,000 a year better off, every year, and the other’s rate is now permanently lower than its neighbor’s.

One west Michigan township’s tax base has grown more than 550 percent since 1994. Pontiac’s grew 8 percent.

Nobody chose that. The formula did.

This also settles the CNBC problem. A state tends to get good at whatever it is graded and congratulated on. We have spent decades being measured on the first number.

Twelve changes. None of them original.

Every item below is already law somewhere else, passed by legislatures of both parties, with results anyone can look up. Michigan is not being asked to invent anything.

But they are not all equally sturdy, and version 2.0 says so. Some change what somebody is paid for. Those outlast the legislature that passed them. Others remove a discretion without changing anyone’s incentive, which means they get re-fought every session.

I have marked which is which.

Beat the reading deadline. July’s science-of-reading law was right. It doesn’t take effect until 2032. A child starting kindergarten this fall reaches fifth grade before her teacher must know how to teach her. Mississippi did it in two to three years and now outreads us. The unused lever: Michigan has School of Choice. Publish completion rates district by district and finishing early becomes a competitive advantage.
Report school results honestly. The state test and the national test differ by thirteen points on the same children. And a law penalizing districts that use unapproved reading curricula cannot be enforced — the Education Department has no authority to ask what curriculum districts use.
Stop taxing improvement. The purest case in the paper, and the highest-leverage item in it. Rehab an empty storefront and your taxes rise. Leave it rotting and they don’t. Fix this one and communities start wanting the growth they now resist.
Let people build. Michigan is short somewhere between 97,000 and 190,000 homes — but a unit count is the wrong measure. In 2025 this state passed the point of having more residents over 65 than children under 18, and its median home was built in 1973 for a family with children in it. The local officials fighting the pending reform aren’t obstacles. They’re doing the arithmetic the state handed them, and getting it right.
Honor other states’ licenses. We license about 180 occupations. A family moves here, the job is real, and the spouse can’t work for six months. More than twenty states have fixed this. Weakest on the incentive test: a licensing board is funded by the people it admits and governed by the people already inside. Restricting entry isn’t a flaw in that design. It’s what the design pays for.
Let clinicians practice to their training. Michigan’s nurse practitioner rules are the most restrictive in the Midwest. Some counties have fewer than ten primary care physicians per hundred thousand people. More than forty states have fixed this.
Make quitting legal again. Detroit became the auto capital because men kept leaving to start rivals. In 1985 Michigan made non-compete agreements enforceable, and inventor mobility fell about a fifth.
Stop writing off the adults. Michigan already spent thirteen years educating the people now working below capacity, and they never left. Of Michiganders who completed apprenticeships in 2022, 94 percent were working here a year later, earning a median $80,700.
Pay the utility for reliability, not for building. In an August letter to the governor, the chair of the Michigan Public Service Commission wrote that it should come as no surprise when utilities choose to build new assets even when cheaper options exist — because shareholders earn a return on what gets built and nothing on tree-trimming. Realigning that incentive, he said, may be the single most impactful opportunity to improve affordability. That is the state’s own lead regulator, making this paper’s argument.
Stop buying strangers. A Bay County plant took a state grant, $6 million in incentives and a federal loan facility up to $544 million. It opened to a governor’s ribbon-cutting and a presidential visit. In July, all 140 jobs ended. Subsidies persist because they pay in a currency officials can spend: a name, a date, a photograph.
Keep Michigan companies Michigan-owned. The firms that quietly lead world markets from small Michigan towns were mostly founded between 1945 and 1985. Their owners are retiring now. A conversion credit orders nobody to do anything. It changes which offer is worth more on the day a founder decides.
Publish the scorecard. Michigan has no standing, state-produced accounting of where it ranks. And version 2.0 turns the test on this one too: who is paid to publish it accurately? An office reporting to legislators will eventually produce a report legislators can live with. The paper names that risk and proposes guardrails rather than pretending it away.

What’s deliberately not here

Two items in Michigan’s policy debate are genuinely contested between the parties, and the paper leaves them out rather than smuggling them in: the state’s labor law, and the timetable of the emissions mandate.

Reading is not conservative. Housing is not liberal. A power bill is not a political party.

Asked whether Michigan’s two parties should compromise to solve the state’s problems, residents said yes by 91 percent to 4.

What to ask

To any candidate, of either party, at any level. Accept no answer without a number and a date.

Which of the twelve will you deliver, and when?

And then the thirteenth question, which is the one version 2.0 added:

After your proposal passes, who has a reason to want it to work?

A candidate who can answer that has thought about the problem. A candidate who cannot is describing an intention, not a policy.

One more number

In 2025, for the first time since record-keeping began, more Americans moved into Michigan than out.

The margin was under two thousand people. A rounding error, and a signal.

None of the twelve changes above requires a tax increase, a constitutional amendment, or a rescue from Washington. Every one is already law somewhere else. Michigan isn’t being asked to invent a future. It’s being asked to get out of the way of one.

And the deepest of them ask for something narrower even than that. Not that anyone in this state work harder or care more. Only that the rules stop paying people to do the opposite of what the state needs.

A township is paid to pave a cornfield and penalized for filling in a neighborhood. A utility earns on what it builds and nothing on what it maintains. A licensing board is funded by the people it keeps out.

None of that required bad faith to produce. None of it will be undone by good intentions either.

Michigan does not need to be talked out of where it is. It needs to stop paying itself to stay there.

📄 Download the full white paper (PDF, 34 pages, v2.0 — October 2026)

Twelve items, each with the Michigan problem, the evidence, which states already did it, and the specific ask. Includes the opposition’s case where it is strong, a note on what the paper deliberately leaves out, sources, method, and a record of corrections made during research.

What changed in version 2.0

This is a larger revision than 1.1 was, and it is a re-grounding rather than an addition. Nothing was withdrawn. No figure changed. But the paper is now organized around a question it did not ask before.

The new question is in Part One, Section 5: who gets paid to say yes?

Version 1.1 diagnosed why nothing gets done — organized interests defending rules they know by name, against costs spread across people who mostly don’t live here yet. That diagnosis sat in the front of the paper and never became a test applied to the recommendations.

Now it is one. Every item was examined against it, and the results were uncomfortable enough to be worth publishing.

Five of the twelve have no constituency after passage. Items 1, 4, 5, 6 and 7 remove a discretion without changing what anyone is paid for. They are still worth doing and several are urgent. But they will be contested every session, and saying otherwise would be selling something.

Four of them change what somebody earns. Those are the ones likely to still be law in ten years.

Three specific things changed as a result.

The local officials fighting housing reform got a fairer hearing. Under this framework they are not obstacles. They are responding correctly to a formula the state wrote, in which approving a duplex can lower their millage rate while the sewer hookups and school seats are theirs to pay for. Their counter-proposal, the Municipal League’s MI Home Program, got a better hearing too: incentives paired with local implementation is the more durable approach, not the softer one. The strongest version pairs both. Preemption sets a floor; the tax fix makes communities want to exceed it.

The infrastructure objection turns out to be partly self-solving. Local officials say the pipes cannot carry the growth, and they are right. But capacity is a cost a community bears with no revenue upside — which is the tax distortion in Item 3 wearing different clothes.

Item 1 has a lever nobody is pulling. Michigan has School of Choice. A district that completes its reading training four years early has something to tell parents that its neighbors cannot — but only if completion rates are published.

New evidence in this version: the Michigan Public Service Commission chair’s August letter to the governor, identifying utility incentive realignment as the most impactful affordability opportunity available; the 2026 Kids Count data, including the finding that Michigan’s score declines began before the pandemic; the CNBC and U.S. News contradiction; and 2026 housing affordability polling.

And one limit on the framework itself, stated in the paper. A test asking who benefits can be misused to explain away any objection as self-interest. That is a cheap move and a wrong one. Several of the local officials quoted in Item 4 are making arguments that would be correct even if they had no stake at all. The test predicts what survives. It is not a theory of who is right.

Version 1.1’s demographic material on housing — the 2025 crossover to more residents over 65 than children under 18, and the mismatch between Michigan’s housing stock and its households — is unchanged and remains in Item 4. That framing draws on Mike Hathorne’s work on the Great Housing Reversal.

David Maness writes The Centaur Desk. He is a geophysicist and was a small business owner in Michigan for 38 years. He served on the Mt. Pleasant Public Schools Board of Education, as chairman of Isabella Bank Corporation, on the Michigan Oil and Gas Association board, and on Michigan’s first Environmental Rules Review Committee. He writes only for himself.

Corrections welcome. This paper is versioned, and errors brought to my attention will be fixed in the next release, with the change recorded in the version history. If you find something wrong, please say so.

The Great Decoupling, Version 3.0: Who Gets Paid to Say Yes

The Centaur Desk — Thu, 01 Oct 2026 14:07:35 GMT

Version 3.0 is out, and it is the first revision that removes as much as it adds.

Two months of accumulated evidence went through explicit triage. Material that changed an argument was absorbed. Material that replaced a weaker citation with a stronger one was substituted. Material that merely agreed with what the paper already said was left out, because a paper that grows with every news cycle stops being read.

Two changes are worth your attention even if you never open the full document.

The first is a new organizing principle. Earlier versions argued that New York’s housing system fails. This version explains why it keeps failing, and the answer is not incompetence or bad faith. Saying no is free, and saying yes is expensive, for nearly everyone in the system who holds a veto. Once you see the payoffs, the outcomes stop looking like dysfunction and start looking like arithmetic. Every provision in the paper is now tested against a single question: who gains from yes?

The second is that this version reports three findings that cut against its own provisions. A study of Chicago-area upzonings found that legalizing three-unit buildings raised supply by only about 3.5 percent. The rent decline in Austin — the paper’s favorite domestic example — coincided with a weakening of Austin’s demand side, which complicates the cleanest version of the supply argument. And more than half of a recent cohort of NYCHA Section 8 vouchers expired before the holder could find a landlord willing to accept one, which is a direct challenge to a proposal built on portable vouchers.

Each is answered in the paper rather than omitted from it, and in two cases the answer changed the design. I would rather publish the objections myself than wait for someone to find them.

What follows is the executive summary. The full 67-page white paper is linked at the bottom.

A Standing Offer to a City Running Out of Time

On June 25, 2026, New York City’s Rent Guidelines Board voted 7-1 to freeze rents on approximately one million stabilized apartments housing about 2 million tenants. The landlord representative resigned before the vote, calling the outcome “decided last year on the campaign trail.” The board’s own data showed the freeze would burden owners already squeezed by double-digit operating cost increases. None of that mattered. The freeze passed anyway.

This is where eighty years of housing price control has arrived: a mayor who ran as a democratic socialist, an appointed board stripped of genuine independence, a rent freeze on 27 percent of the city’s housing stock, a declared campaign to transfer ownership from landlords deemed “negligent” to city-approved stewards, a $22 billion plan to build 200,000 more rent-stabilized units at $40-per-hour mandated wages that make private development economically impossible, and a stated goal of socializing the city’s housing market block by block. The diagnosis is housing unaffordability. The prescription is more of what produced it.

Eighty years of rent regulation have produced a 1.4 percent vacancy rate that falls below 0.4 percent for apartments renting under $1,100 a month, a shelter system that logged 194,531 individual users in a single year — the most in its history — and median asking rents of $3,616 per month. The cure being applied is the disease, administered at higher dosage.

Why the failure is so durable

There is a reason eighty years of reform have not dislodged this, and naming it is the organizing idea of the proposal.

New York’s housing shortage persists because saying no is free and saying yes is expensive for nearly everyone who holds a veto. The community board member who blocks a building is thanked by the neighbors she sees at the supermarket; its future tenants cannot thank her, because they do not live there yet and now never will. The co-op board that rejects a buyer bears no cost and owes no explanation. The tenant who moves forfeits a lease worth more than most New Yorkers’ retirement savings, so she stays, and the apartment never reaches the family that needs it. The owner under a rent freeze earns nothing by improving the building. NYCHA is funded on the basis of documented need, which makes a repaired building a weaker claim on next year’s appropriation than a broken one.

Read that way, the system is not failing. It is working exactly as its incentives specify.

Which means the remedy is not a longer list of prohibitions on saying no. Removing a veto without changing a payoff simply relocates the veto — to litigation, to the permit counter, to the next election. Every provision in this paper is built instead to answer one question: who now has a reason to want this housing built?

The homeowner gains a lawful, financeable, income-producing apartment in her own basement. The tenant of a decontrolling building gains the first right to buy it. The landlord earns a published rating that makes maintenance a source of revenue instead of a cost of compliance. The NYCHA employee carries a pension, a hiring preference, and retraining across the transition. The builder gains a deadline that runs against the government rather than against him. And the city gains the restored tax base of the buildings it permits.

The destination: two markets, not one

The destination is two functioning markets. A rental market good enough that choosing to rent is a decision rather than a defeat. An ownership market open to anyone who wants in. And no thumb on the scale between them beyond what a household actually prefers.

That commitment cuts in an uncomfortable direction. The distortions favoring rental incumbency are well known: rent regulation, address-based subsidy, succession rights. The distortions favoring ownership are less discussed. State law assesses cooperatives and condominiums as if they were rentals, so rental buildings pay more than three times the effective tax rate on co-ops — a penalty passed through in rent. Assessment caps lock long-tenured owners in place. The Decoupling removes what New York can remove, and names the federal biases it cannot.

Eight interlocking actions

Each essential. None sufficient alone.

1. Eliminate all rent stabilization and rent control in the private market — sequenced so liberalization demonstrably leads and decontrol is released by verified supply triggers, not the calendar.

2. Execute a federal asset swap delivering clear, unencumbered title to all 335 NYCHA developments, permanently severing public housing’s dependency on HUD funding.

3. Auction NYCHA properties competitively, funding a tiered portable voucher program built to shrink — subsidies that fall automatically as supply expands and incomes rise.

4. Replace failed government housing inspection with a private rating system that makes quality competition the market’s enforcement mechanism.

5. Liberate the construction market from the regulatory maze — zoning, permitting, environmental review, prevailing wage, landmarks, building codes — and open every fast channel to new homes.

6. Build explicit pathways from renting to ownership — NYCHA cooperative conversions, community land trusts, a voucher-to-ownership bridge, and a right of first offer for tenants of decontrolling buildings.

7. Address infrastructure — water, sewer, power, sanitation, transit — as a coordinated precondition of construction, not an afterthought.

8. Replace the governance architecture with purpose-built institutions that implement the Decoupling and then dissolve, carrying NYCHA’s workforce across the transition with hiring preferences, untouched pensions, and retraining.

How it pays for itself

The Decoupling is funded by the system it replaces. Auction proceeds fund the vouchers and the infrastructure. Restored property tax revenues replace eliminated federal subsidies. It does not require a $22 billion taxpayer commitment, because it does not ask the government to build.

The comparison is not close. Independent analysis puts the government’s own construction plan at $671,000 to $807,000 in public money per affordable unit — a price tag capital is already fleeing. One specialized affordable-housing lender describes pulling back from New York “drastically.” The Decoupling carries no comparable per-unit price tag. It asks the government to stop preventing others from building.

What this version publishes against itself

Three findings in Version 3.0 cut against the paper’s own provisions.

Researchers studying Cook County’s upzonings found that legalizing three-unit buildings produced an aggregate supply increase of only about 3.5 percent — not because of political resistance, but because the revenue from a three-unit building rarely justifies the cost of building it. That is a direct challenge to the missing-middle provision. The paper’s response is to change the sequencing: accessory dwelling units and conversions, whose economics work today, become the near-term volume, with new small multifamily opening as the cost provisions take hold.

Austin’s rent decline coincided with a weakening of Austin’s demand side. Both forces were operating and no study of one city separates them cleanly. What the objection does not explain is the shape of the decline: Class A rents fell 2.6 percent while Class C rents, in the oldest buildings housing the lowest incomes, fell 11.4 percent. That gradient is the signature of supply filtering down, not of demand withdrawing from the top. The claim the paper makes on Austin’s basis is now narrower.

And more than half of a recent cohort of NYCHA Section 8 vouchers expired before the holder could sign a lease. That is the sharpest available challenge to any portable-voucher proposal, and the paper now names utilization as the metric on which its own voucher architecture should be judged. A program that issues vouchers and then reports issuance is reporting its own activity in place of its results.

Who this is for

The persuadable majority is already present in this city. The market-rate renter paying inflated prices because a million stabilized units are permanently off the market. The Black or Hispanic New Yorker facing a homeownership rate of 17 to 26 percent after eight decades of affordability programs. The small landlord watching a building become unviable. The NYCHA resident whose elevator has been broken for six months. The young professional who cannot save a down payment while paying market rent in a city that claims to be solving its housing crisis.

Even the city’s own tenant advocate agrees. New York’s Public Advocate has named NYCHA the city’s single worst landlord every year since 2018, under three mayoral administrations of three different political orientations, finding it carries more open repair violations than every private landlord on his watchlist combined. When the city’s own tenant champion reaches the same conclusion as this paper, the conclusion is no longer a matter of ideology.

None of these people are ideological opponents of progressive housing policy. They are people whose lived experience already speaks to the case the Decoupling makes, whether they recognize it or not.

Choose Your Home. Don’t Inherit It.

Read the full white paper — Version 3.0, 67 pages, PDF

What’s new in this version: a new front section, The Organizing Principle, that tests every provision against who gains from saying yes; a notification-and-mediation layer added to the as-of-right approval standard, answering the strongest criticism of Version 2.0 — that ministerial approval leaves neighbors with no voice; measured permitting durations across American metros, showing New York at 4.6 years against Raleigh’s 1.5; the city’s own Mayor’s Management Report data on NYCHA operations; and the three self-critical findings described above. Several projection tables were removed, and one fiscal verdict the paper could not source was deleted outright.

A note on method. This paper was researched and drafted through directed collaboration between the author and Claude, Anthropic’s AI. Every material claim traces to a public source. The paper is versioned and dated; as evidence accumulates it will be revised, and corrections are noted at each revision. Challenges and corrections are welcome — that is what the comments are for, and this version exists partly because readers of the last one pushed back.

Beyond the Compact: A Market-Based Path Through the Colorado River Crisis

The Centaur Desk — Fri, 25 Sep 2026 18:04:02 GMT

What’s changed since Version 1.0: the first version of this paper set out a simple frame — two paths forward, one preserving the 1922 Compact and one replacing its core formula. Every version since has been an effort to earn that frame through primary-source research rather than just assert it.

This version reflects the largest expansion yet. It adds the complete public archive of the research group known as the Traveling Wilburys of the Colorado River, and the Upper Colorado River Commission’s own annual reports and technical resolutions. It also draws on a rigorous 2012 legal history of the Compact’s delivery-obligation question and extensive Aspen Journalism reporting on Upper Basin negotiations — material this paper had previously underweighted.

Two corrections deserve to be named directly, because they change what the paper argues, not just what it cites. The core registry mechanism is now stated precisely as a hard aggregate cap with every individual allocation fully tradable beneath it, replacing an earlier, less accurate description. The legal debate over whether the Upper Basin owes the Lower Basin an actual water delivery is also now presented in full. An earlier version of this paper leaned on a citation that turned out to represent a historical minority position, and that’s been corrected.

This version also adds a new proposed mechanism for trading water directly across the Upper and Lower Basin divide. A leading researcher on this exact question says the gap remains entirely unsolved. Two current legal precedents place real, honest limits on how that mechanism would actually work.

The arithmetic was wrong from the start, and not by accident

The Colorado River Compact of 1922, plus the later treaty with Mexico, promises 17.5 million acre-feet of water a year. The best current reconstruction of the river’s true long-run average is closer to 14 million acre-feet — and the last six years have averaged only 10.8 million.

USGS hydrologist E.C. LaRue testified directly to Congress in 1925 that the river couldn’t support the promised volume. As water historian John Fleck put it: “None of the senators followed up.” Every crisis since traces back to a choice made with eyes open, not an honest data error.

A debate this paper now presents in full

Whether the Upper Basin owes the Lower Basin an actual water delivery, or merely a promise not to actively deplete the river, has divided the Upper Basin’s own historical leadership since the Compact’s earliest years. It deserves real space on both sides, not a one-line citation.

Colorado’s own 1948 commissioner, Clifford Stone, described a real delivery obligation directly, with the Upper Basin’s own share treated as what’s left over. New Mexico Senator Clinton Anderson made the strongest case for the opposite reading: if the Lower Basin’s delivery simply trumps the Upper Basin’s own apportionment, the Compact’s stated equal apportionment would have been pointless to write in the first place. A 2012 legal history from the University of Colorado’s own law school found the non-depletion reading was actually a minority position within Colorado’s own delegation at the time. That history’s own conclusion is the honest one to carry forward: either interpretation, or some compromise position, could conceivably prevail.

The Upper Basin’s own exposure isn’t new to this year’s crisis

A 2018 report found Lake Powell had already dropped 94 feet since 2000, despite Upper Basin water use staying comfortably within its legal entitlement the whole time. As of September 2025, the running ten-year Lee Ferry total stood at 84.4 million acre-feet, per the Upper Colorado River Commission’s own record. That’s still above the disputed threshold of roughly 82 million, though a later 2026 projection puts it falling below that line within the year.

Not all “voluntary” conservation is the same

A 2004 study of the 2002 drought found mandatory restrictions cut water use by up to 56 percent, while appeals to civic duty achieved only about 12 percent. This paper’s proposal is built on a sharper distinction than “mandate versus market”: a single hard cap on total basin-wide use, with every individual’s share of meeting that cap fully tradable beneath it.

An honest limitation deserves equal billing here. A registry lease reallocates conserved water to a new user — it doesn’t by itself shrink total demand, since someone still ends up using it. A 2025 proposal argues the federal government specifically, not a city or another farmer, should fund permanent retirement of low-value, high-water land. That’s not a new idea for American policy — it echoes the 1937 Bankhead-Jones Farm Tenant Act’s own retirement of unproductive Dust Bowl-era farmland. This paper treats leasing and a targeted federal buyout as complements, not substitutes.

This is a commons problem, and commons problems have a known fix

Roughly three-quarters of the river’s consumptive use goes to agriculture, priced in many districts at a small fraction of what municipal users pay. The largest remaining pool of unrealized value is very likely reallocation within agriculture itself, from low-value forage toward high-value food crops.

What a working solution requires:

A hard aggregate cap, fully tradable beneath it — no personal non-tradable minimum for any single user.
Verified measurement — satellite evapotranspiration data already unanimously adopted by the Upper Colorado River Commission and by Reclamation itself for official federal accounting. This paper extends an existing standard, not an unproven one.
A real joint authority — one commission able to coordinate both major reservoirs as a single hydraulic system, correcting for Colorado’s own documented, repeated institutional delays in building a real conservation program.
Anti-concentration rules without a single buyer — ownership caps without routing every trade through one state monopsony.
Tribal water as a full right, on tribes’ own terms — grounded in the Upper Basin tribes’ own stated framework: unused, undeveloped, uncompensated, and unquantified rights.
A mechanism to cross the Lee Ferry divide — discussed below, with its real legal limits named directly.
Community and basic-needs protection — including direct payment to tenant farmers, not just landowners.

The biggest new idea, honestly limited

Every cross-basin mechanism already operating today stays within one basin. Nothing lets an individual Upper Basin water-right holder sell directly to an individual Lower Basin buyer. Eric Kuhn names the gap explicitly: “Where we haven’t made any progress is what I would call the crossing of the Lee Ferry divide.” This paper offers a concrete mechanism for it — but not without naming two real legal obstacles standing in the way today.

The first is “shepherding”: Colorado currently lacks clear legal authority to protect conserved water from being picked up by other users as it moves toward its destination, except in narrow circumstances that have never actually occurred. The second is a live court decision — Water Horse v. Wilhelmsen, decided by the Utah Supreme Court in October 2025 — establishing that a water exporter must also satisfy the receiving state’s own legal process before the water can be used there. A Colorado-to-Arizona trade would need both states’ legal machinery in place, not just Colorado’s willingness to sell.

Two paths, and a third worth naming

Option A — Preserve the Compact, build the market. Faster, less politically explosive — but it leaves the underlying 17.5-million-acre-foot promise unrepaired.

Option B — Replace the formula. Not a foreign import: the 1948 Upper Basin Compact already allocates by percentage among its own four states. The Upper Division States’ own March 2024 proposal to Reclamation ties Powell releases to Powell’s own storage and ties Mead releases to combined storage. It includes a real concession worth quoting: it “reflects the reality of annual hydrologic shortages across the Upper Division States.” That’s the Upper Basin’s own formal proposal acknowledging its own exposure.

A grand bargain. Mutual mandatory reductions in both basins, paired with a negotiated shield against Compact litigation for the life of the agreement.

A diagnosis failing in real time

Anne Castle and John Fleck’s own 2021 paper argued crisis conditions plus federal deadlines reliably force compromise. By early 2026, both were present in full, and no deal resulted. Castle’s own words, testing her prior thesis against reality: “we have all those underlying conditions, and we don’t have a compromise.”

Fleck’s own diagnosis is worth sitting with: “What we have seen in the last few years is a shift to a leadership that is made up of people who are solely looking out for the interests of their own community.” His own statement of the alternative is among the most direct in this whole research effort: “I would not be at all happy to win the litigation and see the Central Arizona Project shut down. I would see that as a failure even though my community’s water supply might be protected.”

The recommendation

Pursue Option A now, built as a hard aggregate cap with a fully tradable registry beneath it, and build it as shared infrastructure. Because the cap can be tied to trailing river flow rather than fixed by negotiation, relaxing it when conditions improve becomes arithmetic, not a fresh political fight nobody wants to start.

Five limits are worth stating plainly. It doesn’t replace emergency authority in an acute crisis, it doesn’t manufacture water that doesn’t exist, it doesn’t fully solve groundwater depletion, and it doesn’t resolve the Salton Sea’s exposure on its own. And the interbasin trading mechanism proposed here isn’t yet fully executable under current law — a real gap this paper names honestly rather than assumes away.

The full white paper has grown substantially again with this version. It now includes the full delivery-obligation debate on its own terms, and two current legal precedents affecting how the paper’s own proposals would work. It also gives a much fuller accounting of Upper Basin perspectives across negotiators, elected officials, water managers, and internal critics alike. It makes the case for the harder options as fairly as the case for the easier one, and flags plainly where this proposal rests on extension rather than precedent.

📄 Read the full white paper

This paper was produced through an iterative research and drafting process with Claude (Anthropic), working from primary sources — federal agency documents, court filings, peer-reviewed studies, congressional testimony, tribal vision statements, and named-attribution reporting — verified throughout. Comments and corrections welcome; further updates will follow as developments in the basin warrant them.

Beyond the Compact: A Market-Based Path Through the Colorado River Crisis

The Centaur Desk — Mon, 21 Sep 2026 19:29:12 GMT

Update note for returning readers: this is a fast, narrow correction, not a full revision. Version 1.2 stated the Compact’s Lee Ferry “tripwire” threshold as a flat 82.5 million acre-feet. An October 2025 paper co-authored by water law scholar Anne Castle and water manager Eric Kuhn shows the real number is a range, not a fixed figure. It also gives current Lake Powell numbers that have already moved since 1.2 was written. Getting this right mattered enough to fix immediately, rather than wait for the next full update. The rest of the backlog of new material is still being held for that later release.

In August 2026, the federal government did something it had never done before on the Colorado River: it imposed a ten-year water-management plan by order, because the seven states that share the river couldn’t agree on one themselves. Nevada sued within three days. Arizona is expected to follow. A separate legal deadline, tied to the Compact’s own text, is expected to hit within weeks of this post. This is the story of how we got here, and a concrete proposal for a way out that doesn’t run entirely through the Supreme Court.

The arithmetic was wrong from the start, and not by accident

The conventional story is that the Compact’s framers innocently mistook an unusually wet stretch of years for the norm. The historical record doesn’t support that. Multiple federal hydrologists — including USGS scientist E.C. LaRue, who testified directly to Congress in 1925 — warned before and after ratification that the river couldn’t support the promised volume. Congress didn’t follow up. Every crisis since, including the one now unfolding, traces back to a choice made with eyes open, not an honest data error.

The government’s own modeling already knew the answer

A draft federal environmental review released in January 2026 tested five possible management approaches against 1,200 hydrology scenarios. The approach that ended up governing by default — because the seven states never reached their own agreement — was already projected to fail to prevent emergency reservoir crashes more than a third of the time. This wasn’t a surprise that emerged after the fact. It was known, on paper, months before it happened.

Command-and-control just failed, publicly — and a bigger deadline is coming

After the seven basin states missed a third straight round of negotiating deadlines, the Department of the Interior imposed a ten-year operating framework by federal order. It mandates deep, uncompensated cuts on the three Lower Basin states — Nevada, Arizona, and California — while imposing no mandatory reduction on the four Upper Basin states.

Nevada sued within three days. Arizona is expected to follow. But a bigger legal question sits underneath both.

The Compact requires the Upper Basin not let the river’s ten-year flow at Lee Ferry drop below 75 million acre-feet, plus its own half of the Mexico treaty obligation. Depending on how that share is calculated, the real threshold falls somewhere between 82.05 and 82.48 million acre-feet. Basin commissioners project that range will be reached by October 2026. Lake Powell has already fallen below the 3,500-foot elevation Reclamation has pledged to protect, with a projected low of 3,464.4 feet by March 2027. Anne Castle, the water law scholar much of this paper builds on, predicted back in February that litigation was coming “regardless” of which path the states chose.

This paper argues the current approach is failing for a structural reason, not just a political one. It tries to solve a voluntary-exchange problem — how to move a shrinking, variable resource to its highest-value uses — using a coercive, fixed-volume tool built for a river that no longer exists.

Curtailment-by-federal-order also punishes the water users who have already conserved the most. Southern Nevada recycles 85 percent of its indoor water and is still absorbing a 71 percent cut. The approach rewards nothing, and it generates litigation instead of resolution.

Not all “voluntary” conservation is the same

A 2004 study of the 2002 drought found mandatory restrictions cut water use by up to 56 percent, while appeals to civic duty achieved only about 12 percent. That’s often read as proof that voluntary approaches fail. Read correctly, it proves something narrower: asking people to conserve out of a sense of responsibility to others is a weak lever, because no individual’s sacrifice meaningfully moves the outcome.

Paying someone a real price for real conservation is a different mechanism entirely. Utah already pays farmers $390 per acre-foot to conserve. Phoenix’s seasonal pricing cut peak summer water use by 40 percent and nearly eliminated grass lawns. The proposal below rests on payment, not persuasion — a mandatory floor for the baseline everyone must hit, and a market on top that pays people to do better.

This is a commons problem, and commons problems have a known fix

Roughly three-quarters of the river’s consumptive use goes to agriculture, priced in many districts at a small fraction of what municipal users pay. One recent study found $30.32 per acre-foot for agricultural irrigation against $512.01 for municipal supply, with a quarter of all diversions costing nothing at all. That gap isn’t evidence of failure. It’s evidence of enormous unrealized gains from trade, sitting untouched beneath a century of paper allocations.

What a working solution requires:

A mandatory floor plus a paid ceiling — a baseline reduction required of both basins, with a tradable registry on top that pays for going further. A flat mandate alone gives no one a reason to do better than the minimum.
Verified measurement — a basin-wide registry of tradable conservation credits, measured by satellite evapotranspiration data already in official use by the Upper Colorado River Commission.
A real joint authority — a single federal-state commission with actual enforcement power and standing tribal representation, replacing today’s separate, non-binding commissions.
Anti-concentration rules — ownership caps and use-it-or-lease-it rules, learned from where Australia’s otherwise successful water market has drawn criticism.
Tribal water as a full right — an owned, tradable entitlement on each tribe’s own terms. Real precedent already exists: the Colorado River Indian Tribes are leasing conserved water to the Central Arizona Project right now.
Community and basic-needs protection — a standing mitigation fund for rural economies, and a protected minimum supply for every utility’s basic domestic needs.

Two paths, and a third worth naming

Option A — Preserve the Compact, build the market. A mandatory floor and a voluntary trading layer operate on top of the existing state allocations, requiring a narrower federal statute to permit interstate transfers. Faster, less politically explosive — but it leaves the underlying 17.5-million-acre-foot promise unrepaired, and does nothing about groundwater depletion, which now accounts for roughly two-thirds of the basin’s total water storage losses.

Option B — Replace the formula. Fixed entitlements are replaced with proportional shares of whatever the river actually delivers each year. This isn’t a foreign import — the 1948 Upper Basin Compact already allocates among four states by percentage for exactly this reason. The whole-basin version was actually attempted in March 2025, and it failed, not over the specific numbers, but because some states wouldn’t accept any binding delivery obligation at all.

A grand bargain. The same policy team behind the percentage-based idea has more recently proposed something broader: mutual mandatory reductions in both basins, paired with a negotiated shield against Compact litigation for the life of the agreement. This may be closer to their actual central recommendation than proportional shares alone.

A word on the basin’s tribes

Tribal water rights don’t sit on top of a state’s allocation — they draw from the same pool as every other user in that state, which is part of why they’ve been so often overlooked in this fight. This paper treats tribal entitlements as full, ownable rights, tradable on each tribe’s own terms rather than assumed away. That’s not theoretical: the Colorado River Indian Tribes are already leasing conserved water to the Central Arizona Project, building on a right first settled in 1964 — one of the oldest, strongest claims on the entire river.

Who actually wins and loses

The paper tests Option A against four illustrative water users spanning both basins and both ends of the priority system. All four come out ahead of where they stand today — the registry converts uncompensated cuts and idle conservation into paid, voluntary transactions.

But two structural exposures resurface even under a well-designed market. A rural economy can still hollow out if a large rights-holder sells conservation year after year. A lower-income utility can still be outbid by a wealthier one in an acute shortage. That’s why the mitigation fund and the basic-needs carve-out are treated as integral to the registry, not optional add-ons.

The recommendation

Pursue Option A now, built specifically as a mandatory floor paired with a voluntary registry above it, and build it deliberately as shared infrastructure — the measurement systems, the joint commission, the tribal and community-protection mechanisms. A number of credentialed water law scholars already argue that nothing short of a grand-bargain settlement, or full proportional restructuring, will hold up against a river that keeps shrinking. If they’re right, this is the same infrastructure either path would need.

Three limits are worth stating plainly. Neither option replaces emergency operational authority in an acute crisis. Neither manufactures water that doesn’t exist — river science says the basin needs a 13 to 20 percent cut in total use, and no amount of clever trading substitutes for that. And neither touches groundwater, which is quietly doing as much damage as the reservoirs everyone is watching.

The full white paper lays out the hundred-year legal history behind the current crisis in more detail than this summary allows. That includes newly uncovered congressional testimony from 1925, the specific mechanics of the compact provision now nearing its breaking point, and the federal government’s own modeled performance data for the alternatives it considered. It walks through the Australian, Chilean, and New Zealand precedents this proposal draws on and warns against, and works through the legal provisions that determine which paths are actually achievable. It makes the case for the harder options as fairly as the case for the easier one, and flags plainly where this proposal rests on extension rather than precedent.

📄 Read the full white paper

This paper was produced through an iterative research and drafting process with Claude (Anthropic), working from primary sources — federal agency documents, court filings, peer-reviewed studies, congressional testimony, and named-attribution reporting — verified throughout. Comments and corrections welcome; further updates will follow as developments in the basin warrant them.

Beyond the Compact: A Market-Based Path Through the Colorado River Crisis

The Centaur Desk — Thu, 17 Sep 2026 20:18:48 GMT

Update note for returning readers: this version reflects several weeks of additional primary-source research. That includes newly surfaced 1925 congressional testimony, the specific Compact provision now nearing a legal breaking point, the federal government’s own January 2026 modeling of the alternatives, and a substantially expanded tribal water-rights section. If you read version 1.1, this is a real revision, not a minor edit.

The arithmetic was wrong from the start, and not by accident

The government’s own modeling already knew the answer

Command-and-control just failed, publicly — and a bigger deadline is coming

Nevada sued within three days. Arizona is expected to follow. But a bigger legal question sits underneath both. The Compact requires the Upper Basin not let the river’s ten-year flow at Lee Ferry drop below 75 million acre-feet, plus its own half of the Mexico treaty obligation — a real threshold of 82.5 million acre-feet. Basin commissioners project that threshold will be breached by October 2026. Anne Castle, the water law scholar much of this paper builds on, predicted back in February that litigation was coming “regardless” of which path the states chose.

Not all “voluntary” conservation is the same

This is a commons problem, and commons problems have a known fix

What a working solution requires:

A mandatory floor plus a paid ceiling — a baseline reduction required of both basins, with a tradable registry on top that pays for going further. A flat mandate alone gives no one a reason to do better than the minimum.
Verified measurement — a basin-wide registry of tradable conservation credits, measured by satellite evapotranspiration data already in official use by the Upper Colorado River Commission.
A real joint authority — a single federal-state commission with actual enforcement power and standing tribal representation, replacing today’s separate, non-binding commissions.
Anti-concentration rules — ownership caps and use-it-or-lease-it rules, learned from where Australia’s otherwise successful water market has drawn criticism.
Tribal water as a full right — an owned, tradable entitlement on each tribe’s own terms. Real precedent already exists: the Colorado River Indian Tribes are leasing conserved water to the Central Arizona Project right now.
Community and basic-needs protection — a standing mitigation fund for rural economies, and a protected minimum supply for every utility’s basic domestic needs.

Two paths, and a third worth naming

A word on the basin’s tribes

Who actually wins and loses

The recommendation

📄 Read the full white paper

This paper was produced through an iterative research and drafting process with Claude (Anthropic), working from primary sources — federal agency documents, court filings, peer-reviewed studies, congressional testimony, and named-attribution reporting — verified throughout. Comments and corrections welcome; further updates will follow as developments in the basin warrant them.

Michigan Is 46th. Michiganders Think We're 25th.

The Centaur Desk — Wed, 16 Sep 2026 15:43:09 GMT

This is the summary. The full white paper (V1.1)— with sources, bill numbers, and the evidence behind every claim — is available here:

📄 Download the full white paper: Look About You (PDF, 23 pages, v1.1)

Updated from v1.0 — see the note at the end.

Earlier this year, pollsters asked six hundred Michigan voters a simple question: out of fifty states, where does Michigan rank?

The answers came back between 25th and 29th. Average. And nearly identical across party lines.

We ranked 33rd in 2017.

The record

Education. 45th. One in four fourth graders reads proficiently, the same share as in 1998. September’s state test showed 38.2 percent of third graders proficient — a new low in that exam’s twelve-year history. And a law penalizing districts that use unapproved reading curricula cannot be enforced, because the Education Department has no authority to ask what curriculum districts use.

Jobs. 44th in unemployment at 5.0 percent, when Indiana and Wisconsin stood at 3.3.

Roads. 47th in pavement condition, on above-median funding. The state stopped collecting engineering-grade road data in 2019 to save money.

Before anyone reaches for the weather

Minnesota ranked third this year. South Dakota, North Dakota and Nebraska joined it in the top five.

Whatever is happening to Michigan is not happening to states that share our winters.

The fact that makes it fixable

Michiganders aged 18 to 29 were offered a choice: a job here, or a similar job at similar pay somewhere else.

Eighty-nine percent chose Michigan over California. Eighty-seven percent over Chicago. Eighty percent over Texas. Those margins held across every party.

Only 64 percent expect to still be here in ten years. The ones with bachelor’s degrees are the least likely to stay.

This is not a state people are fleeing. It is a state people are leaving reluctantly, for work.

They are not asking for amenities. They are asking for a place to live and a job that pays for it.

Twelve changes. None of them original.

This is the paper’s central claim: every item below is already law somewhere else, passed by legislatures of both parties, with results anyone can look up. Michigan is not being asked to invent anything.

Beat the reading deadline. July’s science-of-reading law was right. It doesn’t take effect until 2032. A child starting kindergarten this fall reaches fifth grade before her teacher must know how to teach her. Mississippi did it in two to three years and now outreads us.
Report school results honestly. The state test and the national test differ by thirteen points on the same children. And a law penalizing districts that use unapproved reading curricula cannot be enforced — the Education Department has no authority to ask what curriculum districts use.
Stop taxing improvement. Rehab an empty storefront and your taxes rise. Leave it rotting and they don’t. That rule is in a 1994 implementing statute, not the constitution. The Legislature wrote it. The Legislature can unwrite it.
Let people build. Michigan is short somewhere between 97,000 and 190,000 homes — but a unit count is the wrong measure. In 2025 this state passed the point of having more residents over 65 than children under 18, and its median home was built in 1973 for a family with children in it. The bipartisan Housing Readiness Package has sat in committee since spring. Local officials say it preempts them, and their strongest objection — that the pipes can’t carry it — deserves an answer.
Honor other states’ licenses. We license about 180 occupations. A family moves here, the job is real, and the spouse can’t work for six months. More than twenty states have fixed this.
Let clinicians practice to their training. Michigan’s nurse practitioner rules are the most restrictive in the Midwest. Some counties have fewer than ten primary care physicians per hundred thousand people. More than forty states have fixed this.
Make quitting legal again. Detroit became the auto capital because men kept leaving to start rivals. In 1985 Michigan made non-compete agreements enforceable, and inventor mobility fell about a fifth.
Stop writing off the adults. Michigan already spent thirteen years educating the people now working below capacity, and they never left. Of Michiganders who completed apprenticeships in 2022, 94 percent were working here a year later, earning a median $80,700.
Pay the utility for reliability, not for building. Roughly half the cost of rebuilding Michigan’s grid is guaranteed monopoly return — under either technology path. The state’s own utility regulator now says realigning that incentive may be the single biggest affordability opportunity available.
Stop buying strangers. A Bay County plant took a state grant, $6 million in incentives and a federal loan facility up to $544 million. It opened to a governor’s ribbon-cutting and a presidential visit. In July, all 140 jobs ended.
Keep Michigan companies Michigan-owned. The firms that quietly lead world markets from small Michigan towns were mostly founded between 1945 and 1985. Their owners are retiring now. Colorado and Iowa made it easier to sell a company to its employees. Michigan has done nothing.
Publish the scorecard. Michigan has no standing, state-produced accounting of where it ranks. That’s why every such number arrives from a magazine or a think tank and can be dismissed as somebody’s methodology. A ranking the state publishes about itself cannot be.

What’s deliberately not here

Reading is not conservative. Housing is not liberal. A power bill is not a political party.

Asked whether Michigan’s two parties should compromise to solve the state’s problems, residents said yes by 91 percent to 4.

What to ask

To any candidate, of either party, at any level. Accept no answer without a number and a date.

Which of the twelve will you deliver, and when?

One more number

In 2025, for the first time since record-keeping began, more Americans moved into Michigan than out.

The margin was under two thousand people. A rounding error, and a signal.

None of the twelve changes above requires a tax increase, a constitutional amendment, or a rescue from Washington. Every one is already law somewhere else.

Michigan isn’t being asked to invent a future. It’s being asked to get out of the way of one.

📄 Download the full white paper (PDF, 23 pages, v1.1 — September 2026)

What changed in version 1.1

If you downloaded version 1.0, here is what is new — and what is not.

Item 4, on housing, is substantially expanded. Everything else in the paper is unchanged. No figures elsewhere were revised and no claims were withdrawn.

The addition is a demographic argument the first version missed entirely. In 2025 Michigan passed a threshold the country will not reach until roughly 2029: we now have more residents over 65 than children under 18. Seventeen states have crossed. Texas and Utah may not until after 2050.

That changes what a housing shortage even means. Harvard’s Joint Center projects that households headed by someone 65 or older will account for roughly 92 percent of all net household growth nationally between 2025 and 2035. Meanwhile fewer than 4 percent of American homes have single-floor living, a no-step entrance, and doorways wide enough to pass through.

So Michigan holds a large stock of three-bedroom houses built for families with children — median year built, 1973 — in a state producing fewer of those families every year.

Which produces a paradox that explains this state better than any shortage figure. A community can have a surplus of large family houses and a severe shortage of housing that fits its actual residents, at the same time.

A widow can sell her four-bedroom house, but the only alternatives require leaving the town where her church and her doctor are. So she stays. And the young family that would have bought it cannot find an entry-level home, because the people in those homes have nowhere to go.

The ladder stops working at both ends.

The practical consequence is that one set of buildings serves both generations. A cottage is a downsizing option for a retired couple and a first home for a young one. An accessory dwelling houses an aging parent or an adult child. A duplex gives an older owner rental income and a younger household a way into a good neighborhood.

Which is worth saying to anyone opposing these reforms on neighborhood-character grounds. The duplex is not for strangers. It is for the widow on your street who cannot manage the stairs and does not want to leave.

Version 1.1 also adds two things to Item 4’s ask: legalizing missing-middle housing in existing neighborhoods, and pairing zoning reform with a path to fund water and sewer capacity — because that last objection, raised by local officials at a committee hearing in May, is the one criticism of the pending legislation that cannot be waved away.

This framing draws on Mike Hathorne’s work on what he calls the Great Housing Reversal.

Beyond the Compact: A Market-Based Path Through the Colorado River Crisis

The Centaur Desk — Sun, 13 Sep 2026 17:22:32 GMT

In August 2026, the federal government did something it had never done before on the Colorado River: it imposed a ten-year water-management plan by order, because the seven states that share the river couldn’t agree on one themselves. Nevada sued within three days. Arizona is expected to follow. This is the story of how we got here, and a concrete proposal for a way out that doesn’t run through the Supreme Court.

The arithmetic was wrong from the start

The Colorado River Compact of 1922 promised seven states and Mexico a combined 16.5 million acre-feet of water a year. The river’s long-run historical average is closer to 15 million acre-feet. Over the past quarter-century of drought, it has delivered far less than that.

The Compact was oversubscribed the day it was signed. Every crisis since — including the one now unfolding — traces back to that original arithmetic error, rather than to any single state’s bad faith.

Command-and-control just failed, publicly

Nevada sued within three days. Arizona is expected to follow. Legal scholars consider the case likely to reach the Supreme Court.

Curtailment-by-federal-order punishes the water users who have already conserved the most. Southern Nevada recycles 85 percent of its indoor water and is still absorbing a 71 percent cut. The approach rewards nothing, and it generates litigation instead of resolution.

It also assumes the river’s decline is one-directional, when the honest hydrological picture includes real variability in both directions. A system with no mechanism to relax when snowpack is strong is a system permanently locked into its worst year.

This is a commons problem, and commons problems have a known fix

Replace top-down allocation with clearly defined, tradable rights, and let voluntary exchange do the work of moving water to its highest-value use.

What a working solution requires:

Verified measurement — a basin-wide registry of tradable conservation credits, measured by satellite evapotranspiration data rather than self-reported diversions.
A real joint authority — a single federal-state commission with actual enforcement power, replacing today’s separate, non-binding Upper and Lower Basin commissions.
Anti-concentration rules — ownership caps and use-it-or-lease-it rules, learned from where Australia’s otherwise successful water market has drawn criticism.
Tribal and environmental rights — treated as owned, tradable entitlements from the outset, not residual claims cut first in a shortage.
Community protection — a standing mitigation fund, financed by a small transaction levy, to protect rural economies when farmland conservation is sold to cities.

Two paths, not one prescription

Option A — Preserve the Compact, build the market. A voluntary trading layer on top of the existing state allocations, requiring a narrower federal statute to permit interstate transfers. Faster, less politically explosive — but it leaves the underlying 16.5-million-acre-foot promise unrepaired beneath the market.

Option B — Replace the formula. Following the model Australia adopted after its own river crisis, fixed entitlements are replaced with proportional shares of whatever the river actually delivers each year. This fixes the over-allocation problem at its root and works automatically whether flows fall or recover — but it means reopening the Compact’s core formula, the harder legal and political lift. Australia’s own experience shows it can also produce a fourfold rise in water values that concentrates wealth among early rights-holders.

Who actually wins and loses

The paper tests Option A against four illustrative water users spanning both basins and both ends of the priority system. They are a senior Upper Basin rancher, a growing Upper Basin municipal utility, a senior Lower Basin agricultural operation, and a junior Lower Basin municipal utility dependent on Central Arizona Project water. All four come out ahead of where they stand today.

But two structural exposures resurface even under a well-designed market. A rural economy can still hollow out if a large rights-holder sells conservation year after year. A lower-income utility can still be outbid by a wealthier one in an acute shortage. Neither is a flaw unique to this design — both are the standard critique of market-based commons solutions generally. That’s why the mitigation fund and the basic-needs carve-out are treated as integral to the registry, not optional add-ons.

The recommendation

Pursue Option A now, and build it deliberately as shared infrastructure — the measurement systems, the joint commission, the tribal and community-protection mechanisms. A number of credentialed water law scholars already argue that nothing short of restructuring the Compact’s core formula will hold up against a river that keeps shrinking. If they’re right, this is the same infrastructure Option B would need on the day it becomes politically possible.

Neither option replaces emergency operational authority in an acute crisis, and neither manufactures water that doesn’t exist. What they can do is ensure that the water the basin does have moves to where it’s worth the most, voluntarily — before the courts decide the question for all seven states at once.

The full white paper lays out the hundred-year legal history behind the current crisis. It walks through the Australian, Chilean, and New Zealand precedents this proposal draws on and warns against, and works through the Boulder Canyon Project Act provisions that determine which of these two paths is legally achievable and how. It makes the case for the harder option as fairly as the case for the easier one, and flags plainly where this proposal rests on extension rather than precedent.

📄 Read the full white paper

This paper was produced through an iterative research and drafting process with Claude (Anthropic), working from primary sources — federal agency documents, court filings, peer-reviewed studies, and named-attribution reporting — verified throughout. Comments and corrections welcome; a version 1.2 will follow as new developments in the basin warrant it.

Michigan Is 46th. Michiganders Think We're 25th.

The Centaur Desk — Fri, 11 Sep 2026 15:17:26 GMT

This is the summary. The full white paper(V1.0) — with sources, bill numbers, and the evidence behind every claim — is available here:

📄 Download the full white paper: Look About You (PDF, 20 pages, v1.0)

Earlier this year, pollsters asked six hundred Michigan voters a simple question: out of fifty states, where does Michigan rank?

The answers came back between 25th and 29th. Average. And nearly identical across party lines.

We ranked 33rd in 2017.

The record

Education. 45th. One in four fourth graders reads proficiently, the same share as in 1998. September’s state test showed 38.2 percent of third graders proficient — a new low in that exam’s twelve-year history. And a law penalizing districts that use unapproved reading curricula cannot be enforced, because the Education Department has no authority to ask what curriculum districts use.

Jobs. 44th in unemployment at 5.0 percent, when Indiana and Wisconsin stood at 3.3.

Roads. 47th in pavement condition, on above-median funding. The state stopped collecting engineering-grade road data in 2019 to save money.

Before anyone reaches for the weather

Minnesota ranked third this year. South Dakota, North Dakota and Nebraska joined it in the top five.

Whatever is happening to Michigan is not happening to states that share our winters.

The fact that makes it fixable

Michiganders aged 18 to 29 were offered a choice: a job here, or a similar job at similar pay somewhere else.

Eighty-nine percent chose Michigan over California. Eighty-seven percent over Chicago. Eighty percent over Texas. Those margins held across every party.

Only 64 percent expect to still be here in ten years. The ones with bachelor’s degrees are the least likely to stay.

This is not a state people are fleeing. It is a state people are leaving reluctantly, for work.

They are not asking for amenities. They are asking for a place to live and a job that pays for it.

Twelve changes. None of them original.

Beat the reading deadline. July’s science-of-reading law was right. It doesn’t take effect until 2032. A child starting kindergarten this fall reaches fifth grade before her teacher must know how to teach her. Mississippi did it in two to three years and now outreads us.
Report school results honestly. The state test and the national test differ by thirteen points on the same children. And a law penalizing districts that use unapproved reading curricula cannot be enforced — the Education Department has no authority to ask what curriculum districts use.
Stop taxing improvement. Rehab an empty storefront and your taxes rise. Leave it rotting and they don’t. That rule is in a 1994 implementing statute, not the constitution. The Legislature wrote it. The Legislature can unwrite it.
Let people build. Michigan is short somewhere between 97,000 and 190,000 homes, depending on whose estimate you use. The bipartisan Housing Readiness Package — HB 5529–5532 and 5581–5585 — was filed in February and endorsed by the governor, the House speaker, and business, environmental and housing groups alike. It has sat in committee since spring. Local officials say it preempts them, and their strongest objection — that the pipes can’t carry it — deserves an answer.
Honor other states’ licenses. We license about 180 occupations. A family moves here, the job is real, and the spouse can’t work for six months. More than twenty states have fixed this.
Let clinicians practice to their training. Michigan’s nurse practitioner rules are the most restrictive in the Midwest. Some counties have fewer than ten primary care physicians per hundred thousand people. More than forty states have fixed this.
Make quitting legal again. Detroit became the auto capital because men kept leaving to start rivals. In 1985 Michigan made non-compete agreements enforceable, and inventor mobility fell about a fifth.
Stop writing off the adults. Michigan already spent thirteen years educating the people now working below capacity, and they never left. Of Michiganders who completed apprenticeships in 2022, 94 percent were working here a year later, earning a median $80,700.
Pay the utility for reliability, not for building. Roughly half the cost of rebuilding Michigan’s grid is guaranteed monopoly return — under either technology path. The state’s own utility regulator now says realigning that incentive may be the single biggest affordability opportunity available.
Stop buying strangers. A Bay County plant took a state grant, $6 million in incentives and a federal loan facility up to $544 million. It opened to a governor’s ribbon-cutting and a presidential visit. In July, all 140 jobs ended.
Keep Michigan companies Michigan-owned. The firms that quietly lead world markets from small Michigan towns were mostly founded between 1945 and 1985. Their owners are retiring now. Colorado and Iowa made it easier to sell a company to its employees. Michigan has done nothing.
Publish the scorecard. Michigan has no standing, state-produced accounting of where it ranks. That’s why every such number arrives from a magazine or a think tank and can be dismissed as somebody’s methodology. A ranking the state publishes about itself cannot be.

What’s deliberately not here

Reading is not conservative. Housing is not liberal. A power bill is not a political party.

Asked whether Michigan’s two parties should compromise to solve the state’s problems, residents said yes by 91 percent to 4.

What to ask

To any candidate, of either party, at any level. Accept no answer without a number and a date.

Which of the twelve will you deliver, and when?

One more number

In 2025, for the first time since record-keeping began, more Americans moved into Michigan than out.

The margin was under two thousand people. A rounding error, and a signal.

None of the twelve changes above requires a tax increase, a constitutional amendment, or a rescue from Washington. Every one is already law somewhere else.

Michigan isn’t being asked to invent a future. It’s being asked to get out of the way of one.

📄 Download the full white paper (PDF, 20 pages, v1.0 — September 2026)

Corrections welcome. This paper is versioned, and errors brought to my attention will be fixed in the next release.

The Great Decoupling, Version 2.0: Two NYC Housing Markets, Not One

The Centaur Desk — Fri, 04 Sep 2026 01:48:31 GMT

Version 1.0 of this proposal went up a week ago. This is Version 2.0, and it changes something I should have gotten right the first time.

The earlier version was ownership-forward. Its rallying cry was “Own Your City.” Read plainly, it treated renting as the thing you escape rather than a legitimate way to live in this city.

That was a bias, and a paper arguing against government preference in housing has no business building one in. So this version describes a different destination: two functioning markets, rental and ownership, with no thumb on the scale between them beyond what a household actually prefers. It also does something the first version didn’t — it names the distortions favoring ownership, not just the ones favoring incumbent renters, including a state tax law and an assessment cap most New Yorkers have never heard of.

The new line: Choose Your Home. Don’t Inherit It.

What follows is the executive summary. The full 62-page white paper is linked at the bottom, with every claim traceable to a public source, and it’s still a living document — versioned, dated, and corrected in the open when something turns out to be wrong.

A Standing Offer to a City Running Out of Time

Eighty years of rent regulation have produced a 1.4 percent vacancy rate that falls below 0.4 percent for apartments renting under $1,100 a month, a shelter system that logged 194,531 individual users in a single year — the most in its history — and median asking rents of $3,616 per month. The cure being applied is the disease, administered at higher dosage.

The Great Decoupling is offered as an answer. Not a counter-proposal that quibbles at the margins. Not a partisan response to a progressive mayor. A comprehensive alternative built on a different theory of what housing markets are, how they fail, and what has actually worked in the cities and countries that have tried it. It is a standing offer to this city — available now for those with the political courage to take it, and ready for the majority of New Yorkers who may yet demand it.

Where it agrees, and where it departs

The Decoupling agrees with Mayor Mamdani on the diagnosis: housing in New York City is a moral emergency. Where it departs — specifically, and on the evidence — is on the cure.

Eighty years of price freezes have not produced affordability. Eighty years of regulatory oversight have not produced quality. Eighty years of a public housing bureaucracy have not produced dignity, and the scale of need has always outpaced what direct government construction can deliver. And eighty years of freezing tenants in place have not preserved the city’s neighborhoods — they have embalmed them, while denying new communities the housing churn from which every great New York neighborhood was born.

What the evidence suggests can work — what a functioning market with proper incentives, transparent information, and portable purchasing power has done in other cities, Tokyo above all, the largest metropolis on earth and among the most affordable — is to supply more, price honestly, and subsidize people rather than addresses.

The destination: two markets, not one

The destination is two functioning markets, not one. A rental market good enough that choosing to rent is a decision rather than a defeat. An ownership market open to anyone who wants in. And no thumb on the scale between them — including the thumbs this city currently presses on both sides.

That last commitment cuts in an uncomfortable direction. The distortions favoring rental incumbency are well known: rent regulation, address-based subsidy, succession rights. The distortions favoring ownership are less discussed. State law assesses cooperatives and condominiums as if they were rentals, so rental buildings pay more than three times the effective tax rate on co-ops — a penalty passed through in rent. Assessment caps lock long-tenured owners in place. The Decoupling removes what New York can remove, and names the federal biases it cannot.

Eight interlocking actions

Each essential. None sufficient alone.

1. Eliminate all rent stabilization and rent control in the private market. This ends the eighty-year experiment whose failure the current crisis documents — sequenced so liberalization demonstrably leads and decontrol is released by verified supply triggers, not the calendar.

2. Execute a federal asset swap delivering clear, unencumbered title to all 335 NYCHA developments, permanently severing public housing’s dependency on HUD funding.

3. Auction NYCHA properties competitively, funding a tiered portable voucher program calibrated to every household configuration and built to shrink — subsidies that fall automatically as supply expands, incomes rise, and renters become owners.

4. Replace failed government housing inspection with a private rating system that makes quality competition the market’s enforcement mechanism.

5. Liberate the construction market from the regulatory maze — zoning, permitting, environmental review, prevailing wage, landmarks, building codes — and open every fast channel to new homes: any building convertible to housing as-of-right, small apartment buildings and accessory units legal on every residential lot, development rights tradable citywide.

7. Address infrastructure — water, sewer, power, sanitation, transit — as a coordinated precondition of construction through an Infrastructure Acceleration Fund and pre-cleared auction sites.

8. Replace the governance architecture built to perpetuate the old system with purpose-built institutions that implement the Decoupling and then dissolve — leaving a functioning market, not a new bureaucracy, and carrying NYCHA’s workforce across the transition with hiring preferences, untouched pensions, and retraining.

How it pays for itself

The Decoupling is funded by the system it replaces. Auction proceeds from NYCHA properties fund the voucher program and the Infrastructure Acceleration Fund. Restored property tax revenues from privatized developments replace eliminated federal subsidies. Reduced social service costs as concentrated poverty disperses offset transition expenses.

The Decoupling does not require a $22 billion taxpayer commitment. It generates the capital to fund itself from assets the government already owns but has never been able to manage.

The comparison to the current approach is not close. Independent analysis puts the government’s own construction plan at $671,000 to $807,000 in public money per affordable unit — a price tag capital is already fleeing. One specialized affordable-housing lender describes pulling back from New York “drastically” as the current rules make deals unworkable. The Decoupling carries no comparable per-unit price tag, because it does not ask the government to build. It asks the government to stop preventing others from building.

The Decoupling does not ask New Yorkers to choose between affordability and markets. It asks them to notice that the system promising affordability has delivered a 1.4 percent vacancy rate — and that the system which actually delivers affordability has never been tried here.

Who this is for

The persuadable majority for this proposal is already present in this city. It is the market-rate renter paying inflated prices because a million stabilized units are permanently off the market. It is the Black or Hispanic New Yorker facing a homeownership rate of 17 to 26 percent after eight decades of affordability programs.

It is the small landlord watching their building become economically unviable under a rent freeze specifically designed to produce the conditions that justify seizure. It is the NYCHA resident who has heard government promises for thirty years while their elevator has been broken for six months. It is the young professional who cannot save a down payment while paying market rent in a city that claims to be solving its housing crisis.

Even the city’s own tenant advocate agrees. New York’s Public Advocate has named NYCHA the city’s single worst landlord every year since 2018, under three mayoral administrations of three different political orientations, finding it carries more open repair violations than every private landlord on his watchlist combined. “If it were a private company, NYCHA would be the worst of all,” he has said. When the city’s own tenant champion reaches the same conclusion as this paper, the conclusion is no longer a matter of ideology.

None of these people are ideological opponents of progressive housing policy. They are people whose lived experience already speaks to the case the Decoupling makes, whether they recognize it or not. The task ahead is to document what the current approach produces as it unfolds, and to keep this alternative ready. Because the Decoupling is more than an exit from what has failed. It is a map toward what demonstrably works — abundance where building is legal, two tenures worth choosing between, neighborhoods chosen rather than assigned, proven where it has been tried, and funded by what it replaces.

On the evidence of June 25, 2026, that documentation is already underway.

Choose Your Home. Don’t Inherit It.

Read the full white paper — Version 2.0, 62 pages, PDF: The Great Decoupling V2.0

What’s new in this version: a section on what a genuinely good rental market looks like, drawing on Germany’s rent-index system and Zurich’s cooperative housing sector; a new closing part naming the tax distortions that favor ownership in New York and explaining how the paper removes them; and a rewritten conclusion built around choice rather than ownership as the goal. It also absorbs a serious empirical challenge to the paper’s supply projections rather than avoiding it, because that’s the discipline this document is supposed to hold itself to.

A note on method. This paper was researched and drafted through directed collaboration between the author and Claude, Anthropic’s AI. Every material claim traces to a public source. The paper is versioned and dated; as evidence accumulates it will be revised, and corrections will be noted at each revision. Challenges and corrections are welcome — that is what the comments are for.

The Great Decoupling: Rebuilding New York Housing on Ownership, Not Dependency

The Centaur Desk — Wed, 26 Aug 2026 15:43:30 GMT

Last post made the case in one page, without a single number. This is the argument with the numbers in it.

What follows is the executive summary of a 57-page white paper on New York City’s housing crisis — the diagnosis, the eight interlocking actions, and how they pay for themselves. The full paper is linked at the bottom, with every claim traceable to a public source.

A word on where this came from. I’m a retired geophysicist, not a housing economist. I started by asking Claude, Anthropic’s AI, a simple question about rent control in New York. Months of directed research later, this is what came out: a comprehensive proposal built by checking each claim against primary sources — the Furman Center, the federal monitor’s quarterly reports, the city’s own planning documents, Rent Guidelines Board data, and peer-reviewed research from Stanford, UC Irvine, and elsewhere. Where the research contradicted what I expected, the paper changed. Where a figure turned out to be stale or wrong, it was corrected and the correction noted.

I make no claim to be the first person to notice New York has a housing problem. The claim is narrower: that the pieces of a working solution already exist, scattered across cities that have solved parts of this, and nobody had yet assembled them into one coherent map. That’s what this is.

The paper is versioned and dated — this is Version 1.0, August 25, 2026 — because it’s a living document. As evidence accumulates, it gets revised, and corrections are noted at each revision.

A Standing Offer to a City Running Out of Time

Eighty years of rent regulation have produced a 1.4 percent vacancy rate that falls below 0.4 percent for apartments renting under $1,100 a month, a shelter system that logged 194,531 individual users in a single year — the most in its history — and median asking rents of $3,616 per month. The cure being applied is the disease, administered at higher dosage.

Where it agrees, and where it departs

The Decoupling agrees with Mayor Mamdani on the diagnosis: housing in New York City is a moral emergency. Where it departs — specifically, and on the evidence — is on the cure.

Eight interlocking actions

Each essential. None sufficient alone.

2. Execute a federal asset swap delivering clear, unencumbered title to all 335 NYCHA developments, permanently severing public housing’s dependency on HUD funding.

4. Replace failed government housing inspection with a private rating system that makes quality competition the market’s enforcement mechanism.

7. Address infrastructure — water, sewer, power, sanitation, transit — as a coordinated precondition of construction through an Infrastructure Acceleration Fund and pre-cleared auction sites.

How it pays for itself

The Decoupling does not require a $22 billion taxpayer commitment. It generates the capital to fund itself from assets the government already owns but has never been able to manage.

The Decoupling does not ask New Yorkers to choose between affordability and markets. It asks them to notice that the system promising affordability has delivered a 1.4 percent vacancy rate — and that the system which actually delivers affordability has never been tried here.

Who this is for

Even the city’s own tenant advocate agrees. New York’s Public Advocate has named NYCHA the city’s single worst landlord every year since 2018, under three mayoral administrations of three different political orientations, finding it carries more open repair violations than every private landlord on his watchlist combined. “If it were a private company, NYCHA would be the worst of all,” he has said. When the city’s own tenant champion reaches the same conclusion as this paper, the conclusion is no longer a matter of ideology.

None of these people are ideological opponents of progressive housing policy. They are people whose lived experience already speaks to the case the Decoupling makes, whether they recognize it or not. The task ahead is to document what the current approach produces as it unfolds, and to keep this alternative ready. Because the Decoupling is more than an exit from what has failed. It is a map toward what demonstrably works — abundance where building is legal, ownership where people live, neighborhoods chosen rather than assigned, proven where it has been tried, and funded by what it replaces.

On the evidence of June 25, 2026, that documentation is already underway.

Own Your City. The Great Decoupling.

Read the full white paper — Version 1.0, 57 pages, PDF: The Great Decoupling

The complete paper covers all eight actions in detail, with the evidence behind each: Tokyo’s zoning regime and Austin’s 16 percent rent decline, Stanford and UC Irvine research on rent control and inclusionary mandates, the city’s own Fair Housing Growth Strategy, the federal monitor’s NYCHA reports, and the lending data showing what capital has already concluded about rent-stabilized buildings. It also includes a note on method — how the paper was built, what errors were found and corrected along the way, and what remains uncertain.

A note on method. This paper was researched and drafted through directed collaboration between the author and Claude, Anthropic’s AI. Every material claim traces to a public source. The paper is versioned and dated; as evidence accumulates it will be revised, and corrections will be noted at each revision. Challenges and corrections are welcome — that is what the comments are for.

The City New York Can Become

The Centaur Desk — Tue, 25 Aug 2026 01:54:49 GMT

This is the first post from The Centaur Desk on New York City’s housing crisis.

What follows is not an argument. It’s a story — the case for what New York’s housing could become, told in one page, without a single statistic or footnote. Some things are worth saying plainly before they’re proven.

The proof is coming. A full white paper — built over several months through directed research with Claude, Anthropic’s AI, and checked against primary sources at every step — follows in a later post, with every claim traceable to where it came from. That’s the method behind everything on this desk: human judgment sets the questions and the standard; machine research does the work of finding, checking, and integrating what’s already known.

Start here. The case follows.

The City New York Can Become

New York is not a city in decline. It is a city under restraint — the greatest concentration of talent, experience, ambition, and hard-won reputation ever assembled in one place, running at a fraction of its capacity. Eight million people from every nation on earth, carrying skills no other city can match, live inside five boroughs — and inside a housing system that punishes the moves that build lives: earning more, growing a family, starting over, putting down roots. The energy has not gone anywhere. It is compressed. And compressed energy, released, is the most powerful force in economics.

The People Are the Engine

What made New York the capital of the world was never a plan handed down from above. It was people organizing themselves — the immigrant who opened a shop, the family that bought a brownstone and rented the garden floor, the block that became a neighborhood, the neighborhood that became an industry the world had not seen. The neighborhood was the machinery of all of it — a small engine of welcome, credit, and culture, turning strangers into citizens — and the city’s greatness was never downtown: it was the sum of its neighborhoods. New Yorkers built it from the bottom up, one voluntary bargain at a time. That engine has never stopped running. It has only been throttled.

What Went Wrong

For eighty years, housing policy has treated New Yorkers as dependents to be managed rather than owners to be trusted. Each program was built in good faith, and each delivered the same inheritance: less ownership, more dependency, and a longer line. The machinery built to shelter people now stands between them and the city they built. This is not a story of villains. It is an old idea reaching its logical end — and a city that has outgrown it. The Decoupling is not only the exit from that idea. It is the map toward what works.

The Promise: Own Your City

The Decoupling makes one promise: shelter, decoupled from dependency. The tenant gains the first chance to buy the home she rents. The project resident becomes a shareholder in the building he was told to be grateful for. The voucher becomes a bridge instead of a ceiling. The homeowner’s basement becomes a lawful home and a second income. The builder gets a generation of work. When people own, they invest; when they invest, they build; when they build, they stay — and the city compounds.

Improving residents’ lives is not the byproduct of unleashing the economic engine. It is the ignition. And no plan can pour a neighborhood, no agency can assign one — a city can only stop preventing them. The Decoupling stops preventing them.

The Horizon: The Second Half of the Century

The second half of the twenty-first century will belong to the cities that trust their people. New York’s claim to remain the leading city in the world rests on its people, and on how they organize themselves when no one stands in the way. The city’s oldest promise — if you can make it here, you can make it anywhere — was never about surviving New York. It was about what New York makes of those who choose it. The Decoupling completes the promise: make it here, and own it here.

New York does not need to be rescued. It needs to be released.

A note on method. This is vision, not evidence — the argument comes next. Everything on The Centaur Desk is researched and drafted through directed collaboration between the author and Claude, Anthropic’s AI, with every material claim in the forthcoming white paper checked against primary sources before publication. Where something turns out to be wrong, it will be corrected here, in the open. That’s the standard this desk holds itself to on every wicked problem it takes on.

Checking the Story I Named This After

Sun, 23 Aug 2026 17:15:16 GMT

Before publishing anything here, I checked the story this publication is named after.

It did not survive intact. That seems like the right way to start.

The story as usually told

In 2005, an online chess tournament let anyone enter in any configuration — humans alone, engines alone, or humans working with engines.

It was won by two amateurs from New Hampshire running ordinary desktop computers. They beat grandmasters with better hardware and better chess.

The lesson drawn from this, endlessly, in business books and conference keynotes: the pairing beats either half. Man plus machine defeats man or machine alone.

I had that sentence in my draft About page. Then I went looking for where it came from.

What the primary source says

ChessBase reported the result the week it happened, in June 2005. The tournament was the PAL/CSS Freestyle event, played on the Playchess server.

The winning team was Steven Cramton, rated 1685 USCF, and Zackary Stephen, rated 1398 USCF. They ran three machines — an AMD 3200+ and two Pentiums — and used four engines: Fritz, Shredder, Junior, and Chess Tiger. No grandmasters or international masters were on their team. They won the final 2.5–1.5.

So the core of the story holds. Better than holds, actually.

“Amateurs” undersells it. A 1398 rating is a novice. These were not strong players with good tools; they were weak players with good process. The advantage came from knowing which engine to trust in which position, and when to overrule all of them.

That is a more interesting finding than the version in the keynotes.

What does not hold

Three things.

The timeline. Deep Blue beat Kasparov in 1997. Kasparov’s “advanced chess” — one human, one engine — came a year later, against Topalov at León. The freestyle format, where any combination could enter, arrived with the PAL/CSS tournaments running from 2005 to 2008. Popular retellings compress all of this into a single moment. My draft did too.

A name. One widely-used AI-generated reference source gives the winner as “Zachary Bekkedahl.” It also claims a 7/8 score against the Hydra supercomputer and a performance rating above 3100. The contemporaneous report supports none of this. The surname does not appear anywhere in the primary sources. It was invented — confidently, plausibly, in the exact register of a real fact.

The conclusion. This is the serious one. Kasparov was still asserting centaur superiority in 2017, telling Tyler Cowen there was no doubt a human with engines beat the best engine alone. He was probably right when he said it.

He is not right now. Engines improved to the point where the human contribution went negative — overriding a modern engine is almost always an error. The last meaningful centaur rankings date to around 2014. Where the old human job survives at all, selecting among engines that disagree, the selecting is now done by another program.

The punchline of the story I named this publication after has an expiration date, and it has passed.

Why keep the name

Because research is not chess.

Chess has an evaluation function. Every position resolves to a number, every move is checkable, and once the machine calculates better than you do, your judgment is overhead.

There is no such function for whether Michigan’s road funding formula is the binding constraint on its infrastructure, or whether decontrolling New York’s rental market before supply arrives is a policy or a catastrophe. Nothing scores those. Nobody can run the tablebase.

What the machine gives me is enormous: drafting speed, source retrieval, structural proposals, consistency across a hundred pages, counterarguments I would not have thought of. What it cannot give me is the decision about what is true.

And the 2005 lesson survives its own obsolescence. Cramton and Stephen won on process — on a disciplined method for when to trust the machine and when to override it. That question is dead in chess and entirely alive everywhere else.

The standard

I found the invented surname the only way anyone finds them. I opened the primary source.

Everything published here is drafted by a large language model working under my direction. That is disclosed on the About page in detail, including the failure modes. This post is one of them, caught in the wild, on the first thing I tried to write.

The rule I work under: every factual claim traces to a document I can open myself. Not a summary of the document. Not a confident restatement. The document.

That rule is slow. It killed a sentence I liked about man and machine beating either alone. It will kill others.

Before any of that, this seemed worth establishing: when the evidence does not support the story, the story goes.

Including when it is the story on the masthead.

Corrections welcome and will be made in public.