Consider the Source: Project 2025 and the Ledger

There is a habit in political writing of treating a document as a symbol. Project 2025 has been a symbol for two years now, which means most of what has been written about it is about what it stands for rather than what it says. That is a loss, because the document is unusually specific, it is public, and it can be checked.

So let me do something unglamorous. Let me consider the source.

Not in the dismissive sense. In the literal one. Where did this document come from, what does it actually contain, what happened to it, and what did that do to the country. The answers are more interesting than the symbol, and one of them is a correction to something I had absorbed myself.

I. The source

Mandate for Leadership: The Conservative Promise is a book. Nine hundred and twenty pages, thirty chapters, one for each major federal department or agency. It was published in 2023 by The Heritage Foundation, which holds the copyright, and it is the first pillar of a larger project organized by Heritage called the 2025 Presidential Transition Project.¹ The man who directed that project, Paul Dans, signed the note that opens it, and he had been chief of staff at the Office of Personnel Management in the previous administration.²

The book is not the work of a fringe. Heritage describes the project as “more than 50 (and growing) of the nation’s leading conservative organizations joining forces,” and by the time the project wound down, its own site said it “was a product of more than 100 organizations.”³ The book itself says it “is the product of more than 400 scholars and policy experts from across the conservative movement and around the country.”⁴ The scale is the point: this is an institutional product, assembled over years, by people who expected to be hired to implement it.

The project was organized around four pillars, and the book names them. Pillar I is the volume itself, which “puts in one place a consensus view of how major federal agencies must be governed.” Pillar II is “a personnel database that allows candidates to build their own professional profiles and our coalition members to review and voice their recommendations.” Pillar III is “the Presidential Administration Academy, an online educational system taught by experts from our coalition.” Pillar IV is the playbook, where the project said it was “forming agency teams and drafting transition plans to move out upon the President’s utterance of ‘so help me God.’”⁵

Read Pillar II again. A private foundation built a database of vetted candidates for federal jobs, and its members were to review them. That is the machinery of a government-in-waiting, assembled by an organization that had not been elected to anything.

The money behind it is not disclosed. Heritage does not publish its donors, and neither did the project. Its tax filings show the scale of the operation: revenue of $133.9 million in the year ending December 2024, of which $106 million was contributions.⁶ Investigative reporting by DeSmog traced more than $122 million given since 2020 by foundations connected to six billionaire families, including Bradley, Coors, Koch, Scaife, Seid, and Uihlein, to organizations on the project’s advisory board.⁷ That is a reporter’s finding, not a Heritage disclosure, and I want to be precise about the difference. Heritage’s own filings do not name its funders.

And the document has a stated audience. The foreword says plainly that “one set of eyes reading these passages will be those of the 47th President of the United States,” and that “the entirety of this effort is to support the next conservative President, whoever he or she may be.” The goal is stated without ornament: “Our goal is to assemble an army of aligned, vetted, trained, and prepared conservatives to go to work on Day One to deconstruct the Administrative State.”⁸

That is the source. A private institution, funded by donors it does not name, wrote a governing program for an office it did not hold, and staffed a roster to carry it out.

II. What the document does not say

Here is the correction, and it is to me as much as to anyone.

Project 2025 is routinely described as a blueprint for the “unitary executive.” It is described as laying groundwork for presidential impoundment of appropriated funds. Both phrases get attached to it constantly. I had absorbed both.

Neither appears in the book. I extracted the full text and searched it. The string “unitar” appears zero times in 920 pages. So do “impound” and “impoundment.” The document contains no impoundment argument at all.⁹

This does not make the document harmless, and it does not mean the ideas are absent. What the book does say about executive power is aggressive enough on its own terms. Russ Vought, who wrote the chapter on the Executive Office of the President, argues that “in its opening words, Article II of the U.S. Constitution makes it abundantly clear that ‘[t]he executive power shall be vested in a President of the United States of America,’” and that “[t]hat enormous power is not vested in departments or agencies, in staff or administrative bodies, in nongovernmental organizations or other equities and interests close to the government.” He writes that “[t]he President must set and enforce a plan for the executive branch,” and that “[t]he modern conservative President’s task is to limit, control, and direct the executive branch on behalf of the American people.”¹⁰

That is a strong theory of presidential control. It is not the unitary executive theory by name, and the difference is not nothing: the document’s argument is about direction of the branch, which has a long pedigree, while the phrase people attach to it has a specific and contested constitutional meaning. Heritage also ran a separate “Presidential Transition Project” apparatus, and it is possible the phrase appears in material outside the book. But the book is the thing that was called the blueprint, and the book does not contain it.

One more. The “180-day playbook” is one of the most-repeated descriptions of the project, and it is real. But it is on the project’s website, not in the book. The archived project page said, “The fourth pillar of Project 2025 is our 180-day Transition Playbook and includes a comprehensive, concrete transition plan for each federal agency.”¹¹ The phrase “180-day” does not appear in the book at all. And the playbook itself was never published. Paul Dans described it to Semafor in 2023 as “a 180-day game plan of regulations and executive orders that a president could sign on day one,” but the document was described rather than released.¹²

Why does this matter enough to spend a section on? Because “consider the source” cuts both ways. If the argument is that this document is the origin of a governing program, then the document is the evidence, and the evidence should be read rather than recalled. A movement that inherited two phrases the text does not contain has been described, in part, by people who did not open it. That is true of critics and it was true of me.

III. How much of it happened

Now the harder question, because the honest answer is contested and the contest is instructive.

There is no single official count of Project 2025’s implementation, and the counts that exist are not comparable. That is the first thing to understand about the numbers.

The most rigorous effort is maintained by the Center for Progressive Reform and Governing for Impact, which compiled 532 executive-branch recommendations from the domestic chapters and mapped each to an action. Their February 2026 update reported that the administration “has initiated or completed 53 percent of Project 2025’s domestic administrative policy agenda in the 12 months following the president’s inauguration. In all, 283 of the 532 recommended actions identified in this tracker have been put into action.”¹³ Note the qualifier: domestic administrative agenda. That is a subset of the book, chosen because those are the recommendations an executive can act on.

A second independent effort, project2025.observer, tracks 321 objectives across 34 agencies and headlines “55%” overall progress, with 141 completed and 70 in progress. It arrives at 55% by weighting in-progress items at half, which is an inference from its chart arithmetic rather than a stated rule.¹⁴ A third, at p2025t.vercel.app, tracks a curated 66 “major” proposals and shows a 70% implementation gauge.¹⁵

Those denominators differ by a factor of eight. A 53% on 532 items and a 70% on 66 items are not the same phenomenon, and neither is a corroboration of the other. And the widely repeated 53% is one finding echoing: several secondary sites repeat the Center for Progressive Reform figure while attributing it to a different source, which produces the appearance of agreement where there is only repetition.¹⁶

Here is the part I did not expect. Heritage itself says the number is 53%. Bloomberg Law reported in June 2026 that Heritage, in a fundraising email, wrote that “[t]he Trump administration has implemented 53% of the policy proposals in Project 2025, according to the conservative group that authored the plan.”¹⁷

Two organizations on opposite sides of this fight, using different methods and different denominators, landed on the same figure. I want to be careful here rather than make a rhetorical meal of it. The agreement is striking, and it is at least partly coincidence, because the two counts are not measuring the same set. CPR’s 53% is 283 of 532 domestic administrative recommendations. Heritage’s 53% is its own characterization of its whole agenda, offered in a fundraising email with no methodology attached. What the convergence does establish is narrower and still significant: neither side’s account of the last year puts the number below half. If the plan’s own authors and its most careful critics both say roughly half, the argument about Project 2025 is no longer whether it is being implemented. It is over.

There is a third count worth citing, from outside the tracker ecosystem. The 19th reported at the end of 2025 that the administration had implemented “roughly half of the goals laid out in the document’s 920 pages.”¹⁸

All of these counts share a weakness: none has been independently recomputed from raw data, including by me. They are as-stated figures, and the freshest institutional one is seven months old. I am reporting the state of the measurement, not certifying it.

IV. The pattern in what happened

If you stop counting and start reading the record, a cleaner structure appears, and it is the most useful finding here.

Project 2025 proposed two kinds of things. Some it asked the executive to do, which a president can largely do alone. Others it asked Congress to do, which a president cannot. And the implementation record tracks that line almost mechanically.

Where the document asked for executive action, the action happened. Where it asked for legislation, the legislation mostly did not, and where the administration wanted the result anyway, it found a different route.

Start with the clearest case. The book said “[f]ederal education policy should be limited and, ultimately, the federal Department of Education should be eliminated.”¹⁹ But the same chapter acknowledged the limit: “In order to fully wind down the Department of Education, Congress must pass and the President must sign into law a Department of Education Reorganization Act (or Liquidating Authority Act).”²⁰ Congress has not passed one. The department still exists, still publishes, and is still running programs as of this week.

What happened instead is the interesting part. An executive order in March 2025 directed the Secretary to “take all necessary steps to facilitate the closure of the Department of Education” to the maximum extent permitted by law, and the department cut its own workforce nearly in half, from 4,133 to roughly 2,183.²¹ Then, rather than waiting for Congress, it transferred functions to other agencies through interagency agreements: civil-rights and student-privacy enforcement to the Department of Justice, workforce programs to Labor, Indian education to Interior, student-loan administration to Treasury, and a partnership with Health and Human Services on special education.²² Title I was not moved to HHS as the book proposed; the department instead issued guidance letting states spend up to three percent of Title I funds on “direct student services,” which is a school-choice flexibility the book would have approved of.²³

That is a partial implementation by a mechanism Project 2025 never named. The book said “eliminate.” The administration said “return to the states,” and did part of it administratively.

Immigration follows the same pattern. The book’s primary recommendation was for the President to “pursue legislation to dismantle the Department of Homeland Security.” That did not happen, and no bill came close.²⁴ What the book asked the executive to do instead, it did with force: “Prioritizing border security and immigration enforcement, including detention and deportation, is critical,” it said, and detention and deportation are at record scale.²⁵ ICE’s own data shows 319,980 removals in fiscal 2025, and 356,389 in fiscal 2026 through July 11, more than the full prior year inside nine and a half months.²⁶ Detention population reached 67,022 against a maximum capacity of 80,700, at 83 percent occupancy, and the non-detained docket with final removal orders stood at 2,120,058 people.²⁷ The 287(g) program, which deputizes local officers, grew to 2,559 agreements across 39 states and two territories.²⁸

The book asked for 100,000 detention beds. Funded and contracted capacity has not reached that number, which is the kind of gap a careful reader should hold onto rather than round away.²⁹

The pattern repeats in the regulatory agencies. The book said EPA’s “structure and mission should be greatly circumscribed,” and proposed eliminating the stand-alone enforcement office.³⁰ EPA cut 23 percent of its personnel in July 2025 and dismantled the Office of Research and Development, replacing it with a smaller office; staffing has fallen to about 12,700, the lowest level since the Reagan administration.³¹ Civil judicial enforcement filings by the Justice Department on EPA’s behalf fell to 16 in the first year, against 127 in Obama’s second-term first year, 86 in the first Trump year, and 67 in Biden’s first year, according to an analysis by the Environmental Integrity Project.³² EPA’s own report counters with a different measure, 2,127 civil cases concluded, which it calls the highest in nine years; the two figures measure different things, administrative conclusions and judicial filings, and both can be true.³³

And where the book wanted withdrawal from treaties, the executive simply withdrew. Project 2025 said “the next conservative Administration should withdraw the U.S. from the U.N. Framework Convention on Climate Change and the Paris Agreement.”³⁴ The notification was deposited in January 2025 and took effect for the United States on January 27, 2026.³⁵

Where Congress was required, some things still happened through Congress, by a different route than the book prescribed. The book wanted Medicaid’s open-ended federal match replaced with “a more balanced or blended match rate, block grants, aggregate caps, or per capita caps.”³⁶ That did not pass as written. What passed was a reconciliation law cutting federal Medicaid spending by an estimated $911 billion over a decade, with work requirements starting January 2027 in 44 states, and enrollment down 5 million, or six percent, between May 2025 and May 2026.³⁷

So the honest structure is this: roughly half the agenda, and the half that happened is disproportionately the half that a president could do alone. That is not a comment on Project 2025’s merits. It is a comment on where American governing power now sits. A 920-page plan written for a legislature-inclusive process was implemented, to a large degree, by the part of it that did not need the legislature.

V. The disavowal

The candidate the document was written for spent 2024 saying he had nothing to do with it.

On July 5, 2024, he posted to Truth Social: “I know nothing about Project 2025. I have no idea who is behind it. I disagree with some of the things they’re saying and some of the things they’re saying are absolutely ridiculous and abysmal.”³⁸ At the September 10, 2024 debate he said, “I have nothing to do with Project 2025. I haven’t read it. I don’t want to read it purposely. I’m not going to read it.”³⁹ His campaign issued a memo saying that “Project 2025 had nothing to do with the campaign, did not speak for the campaign, and should not be associated with the campaign or the President in any way.”⁴⁰

Then in October 2025, during a government shutdown fight, he wrote on Truth Social directing “Russ Vought, he of PROJECT 2025 Fame, to determine which of the many Democrat Agencies… he recommends to be cut.” AP and PBS characterized the post as a dramatic about-face.⁴¹

“Fame” is doing a lot of work in that sentence. It is the same Russ Vought who wrote the chapter arguing that “[t]he President must set and enforce a plan for the executive branch.” Project 2025’s chapter authors are not abstractions in a think-tank volume: Vought ran the Office of Management and Budget, Peter Navarro, who wrote the trade section, became a top trade adviser, and Brendan Carr, who wrote about the Federal Communications Commission, now runs it.⁴²

I am not arguing that the disavowal was a lie in the strict sense that requires me to know his mind. I am noting a simpler thing that is entirely on the record. In July 2024 he said he had no idea who was behind the document. In October 2025 he referred to its most senior author by the fame the document had given him. Between those two statements, roughly half the document became policy.

And the document predicted the relationship itself. Its foreword says one set of eyes reading it would be the 47th President’s, and that the effort supports “the next conservative President, whoever he or she may be.”⁸ That is not a boast about influence. It is a stated plan for a handoff, written by people who expected the handoff to happen whether or not it was acknowledged.

VI. Consider the source

Here is what I take from all of it.

The most important fact about Project 2025 is not that a conservative think tank wrote a conservative agenda. Organizations do that constantly, and most of their documents go unread and unimplemented. The important fact is the shape of what happened: a private foundation, funded by donors it does not name, wrote a 920-page governing program for an office it did not hold, assembled a roster of people to staff it, and after the candidate disavowed it, roughly half of it became the policy of the United States.

Both counts of that half come from camps that disagree about everything else. That is the strongest evidence available, and it is why the number has held up in a conversation that has been unusually resistant to evidence.

So consider the source, in both senses. Consider where this came from: not a movement in a hurry, but an institution with $134 million a year, a personnel database, a training academy, and four years of patience. And consider the document as a source, which means reading it, including the correction I had to make to myself. Two of the phrases most attached to it are not in it. The parts that are in it are worse than the phrases, and more specific, and harder to argue with, because they are written down and they were carried out.

The last thing worth saying is the least comfortable. If a private institution can write a governing program and see half of it enacted without ever having to defend it in an election, the question is not what Project 2025 wanted. The question is what the next such document will contain, who is writing it now, and whether anyone will read it before it is implemented rather than after.

I have no confidence that they will.