"No, We Didn't": DOGE's Own Staffers Just Admitted the Whole Experiment Failed

The Senate Permanent Subcommittee on Investigations released a report last year documenting $21.7 billion in waste directly attributable to DOGE's activities. The breakdown is instructive. The largest single item: $14.8 billion paid to approximately 200,000 federal workers under the Deferred Resignation Program — the "fork in the road" buyout scheme — who were paid for months not to work. Another $6.1 billion went to the more than 100,000 employees involuntarily separated, many of whom sat on administrative leave for weeks collecting paychecks while their positions remained vacant. The government spent well over $20 billion to reduce headcount while simultaneously paying the heads it was reducing. [2]

It gets more uncomfortable from there. The Brookings Institution's Hamilton Project found federal spending climbed nearly 6 percent over the period DOGE was most active. The Cato Institute — not exactly a bastion of big-government advocacy — calculated that firing 10 percent of the federal workforce would produce only about $40 billion in savings, because most federal spending is not salaries. The Partnership for Public Service estimated that the combined cost of firing, placing workers on leave, and subsequently rehiring some of them ran taxpayers roughly $135 billion. Musk's own Office of Personnel Management director Scott Kupor acknowledged earlier this month that workforce reductions had, in his word, been "overdone." [2]

I think it was more important to reduce the federal deficit from $2 trillion to close to zero.

— Nate Cavanaugh, DOGE staffer, in deposition — immediately before admitting the deficit was not reduced

The People Behind the Promise

Justin Fox and Nate Cavanaugh, the two DOGE operatives at the center of the deposition videos, were both in their 20s when they arrived at the National Endowment for the Humanities. Neither had worked in government before. Neither had a background in grant administration, the humanities, or federal budget management. Their combined qualifications for the job amounted to some experience in tech startups and venture finance. Their salaries ran between $120,000 and $150,000 annually, funded by the taxpayers whose dollars they were ostensibly there to protect. [1]

Their methodology, as revealed in the depositions and court filings, was to feed grant descriptions into ChatGPT with a prompt asking the AI to identify DEI content. Grants that surfaced hits on terms like "DEI," "LGBTQ," "Black," or "equity" were flagged for cancellation. Fox acknowledged under questioning that he could not define DEI when pressed. When an attorney walked him through specific grants — a documentary on the 1873 Colfax massacre, where dozens of Black men were killed by former Confederates — Fox agreed the grant qualified as DEI and should be cut. No appeal process was established. Termination letters went out under the agency director's signature, which Fox and Cavanaugh had used without explicit approval. More than 1,400 grant recipients learned by form letter that their projects were over. [1]

A Conservative Case for Honest Accounting

Conservatism has a core commitment that distinguishes it from the alternatives: results matter. Not intentions, not rhetoric, not the ideological alignment of the project in question — results. A government program that costs more than it saves is a bad program regardless of which party runs it, which billionaire champions it, or how gratifying it is to watch the bureaucracy squirm. This standard is supposed to be our north star. It is not a standard that can be selectively applied only when Democrats hold the pen. [3]

One reads with some interest the recent enthusiasm in political commentary for applying rigorous accountability standards to government institutions. The impulse is correct and always welcome. It would carry more persuasive weight if that same analytical rigor were turned on DOGE with the same intensity it is applied elsewhere. When a program promises $2 trillion in savings and delivers a documented $21.7 billion in new waste, the honest conservative response is not to defend the program — it is to say, clearly, that it failed. And to ask what a serious effort at government efficiency would actually require. [2][3]

What Serious Efficiency Actually Requires

Real government efficiency is not fast. It does not fit into a DOGE social media post. It requires people who understand the programs they are reviewing, a methodology that holds up to legal scrutiny, and a willingness to distinguish between spending that is genuinely wasteful and spending that merely involves words the administration dislikes. It requires acknowledging that most federal spending is not salaries, so cutting salaries doesn't cut spending — it just shifts costs to contractors who charge more. It requires, at minimum, a plan for what happens when you fire the IRS auditors responsible for collecting the revenue that funds the rest of government. A Yale Budget Lab analysis estimated that IRS workforce reductions will cost $198 billion in lost revenue over a decade. That is not efficiency. That is arithmetic in reverse. [2]

None of this means the project of making government leaner and more accountable is wrong. It isn't. The federal bureaucracy has real inefficiencies, real redundancies, and real programs that have long outlived their purpose. A serious effort to identify and eliminate them — one built on competence rather than speed, staffed by people with relevant expertise, and measured against actual outcomes — would be worth doing and worth defending. What happened with DOGE was not that effort. What happened was that two young men with no government experience were handed enormous power, used an AI chatbot to make hundred-million-dollar decisions, and then sat under oath and acknowledged they hadn't reduced the deficit by a single dollar. The honest conservative verdict is that taxpayers deserved better than that. And the honest conservative commitment is to say so, regardless of who is in the room. [1][2]