What happens when the Chainsaw Comes for the Guardrails
Nothing good...
640,000 dead. $244 billion lost.
That’s the current estimate of what DOGE accomplished in its ten months of operation. Not savings. Cost.
The deaths are global: children who stopped receiving malaria treatment, HIV patients cut off from antiretrovirals mid-course, families who lost access to food programs and clean water. The money is ours: lost productivity, rehiring costs, litigation, and IRS enforcement cuts that will bleed revenue for a decade.
We already had systems to find waste and fraud. Audits, evaluations, inspectors general, accountability offices with century-long track records. They were the guardrails.
This is what happens when you skip the evaluation and take a chainsaw to the guardrails instead.
The Human Cost
Brooke Nichols, an infectious disease modeler at Boston University, built a tracker. It updates in real time. As of November 2025, her model estimates that over 600,000 people have died as a result of USAID funding cuts. At least two-thirds of them were children.
The methodology has been peer-reviewed and published in the Journal of the International AIDS Society. It draws on known data about how many people PEPFAR and other USAID programs were serving, combined with published mortality rates when treatment stops. The UN Stop TB Partnership and Avenir Health contributed to the tuberculosis estimates. The tracker has been revised over 200 times as new data came in and colleagues offered feedback.
The numbers are projections, not body counts, and Republicans in Congress have disputed them. But the underlying logic is straightforward: when you cut off HIV treatment to millions of people, some of them die. When you end malaria prevention programs, some children die. When you stop funding clean water and food assistance, some families don’t make it. The question isn’t whether there’s a death toll. It’s how large.
Even conservative readings of the data point to catastrophe. PEPFAR alone was providing antiretroviral treatment in 54 countries. People who were stable on medication were cut off mid-course. That’s not a policy difference. It’s a death sentence for people whose virus will now develop resistance, who will now transmit to others, who will now die of something we knew how to prevent.
The World Food Programme projected a 40 percent decrease in funding. Diarrhea, pneumonia, malnutrition: the diseases that kill children when the infrastructure disappears. We had the infrastructure. We cut it.
I worked adjacent to some of these programs. I knew people who ran them, who spent years making sure the money actually reached the communities it was supposed to reach. They were terminated by email. The communities they served found out when the shipments stopped arriving.
The Financial Cost
DOGE claimed $214 billion in savings. But they were off by a couple decimal places. The actual number, when anyone bothered to check the receipts, was closer to $2 billion.
This wasn’t a dispute about methodology. It was arithmetic.
An $8 billion contract that was actually $8 million. Another decimal places problem. Someone added three extra zeros and nobody caught it. A USAID contract listed three times at $655 million each, tripling the claimed savings from a single line item. A Social Security contract entered as $232 million instead of $560,000. Contracts marked as terminated that federal databases showed were still active. Savings claimed on money that had already been spent years ago.
These aren’t rounding errors. These aren’t differences in accounting philosophy. This is data entry that wouldn’t pass a high school spreadsheet assignment.
NPR, CBS, Fortune, and the New York Times all ran the numbers. None of them matched what DOGE was claiming. When reporters pointed out the errors, DOGE quietly revised the website and introduced new errors. One analysis found at least 14 mistakes on the wall of receipts in a single day.
The organization that promised “maximum transparency” couldn’t manage basic subtraction.
And then there were the rehires.
In February, the administration fired more than 300 employees at the National Nuclear Security Administration, the agency that manages the country’s nuclear weapons stockpile. The next day, they tried to rehire them. Sources told CNN that the officials responsible for the decision did not seem to know what the agency did. But the human resources office had been cut out of the layoff process, so managers couldn’t figure out how to contact the people they’d just fired. Some couldn’t be reached. Some reconsidered whether to come back at all.
This happened over and over. The FDA had to rehire food safety scientists. The Department of Agriculture reversed plans to lay off 25 percent of staff at 58 facilities responsible for bird flu response. After HHS fired the entire team that runs the Low Income Home Energy Assistance Program, they had to bring back one employee for two and a half weeks just to run the formula that distributes $400 million to states.
Veteran federal employees reported being given 15-minute sessions to justify their jobs to 19- and 20-year-olds. An epidemiologist at the National Institute for Occupational Safety and Health was placed on leave and then called back. He told CNN there’s “still a lot of chaos, sort of throughout the federal workforce.”
Max Stier, CEO of the Partnership for Public Service, called it “a mosaic of incompetence.”
But the real damage isn’t the inflated savings or the botched firings. It’s the costs they didn’t count.
The Partnership for Public Service estimated $135 billion in lost productivity, paid leave for workers who were fired and then rehired, and administrative chaos from cuts made without understanding what the agencies actually did. IRS enforcement cuts are projected to cost over $500 billion in lost revenue over the next decade. Terminated contracts are being litigated, and many will require settlement payments. The Inspector General of the Federal Reserve found that cuts at the Consumer Financial Protection Bureau made its security program ineffective.
Add it up. One analysis found that DOGE’s net impact was a $244 billion loss. Not savings. Loss.
Elon Musk promised $2 trillion in cuts. Then $1 trillion. Then $150 billion. The final number, if you’re being generous, is maybe $2 billion in actual verified savings. If you’re being honest, it’s probably negative. In the hundreds of billions.
This is what efficiency looks like when the people in charge can’t do the math.
The Guardrails We Already Had
Here’s what makes this worse: we had systems for this. They were working.
USAID had annual audits. Impact evaluations. Inspector General reviews. Data-quality assessments. Third-party monitoring. The agency wasn’t perfect, and I’ve written about its flaws, but it wasn’t operating in the dark. There were people whose entire job was to find waste and fraud and fix it.
The Government Accountability Office has been doing this work since 1921. Over the past 15 years, GAO recommendations have generated $725 billion in savings and revenue increases. Last year alone, $67.5 billion. The return on investment: $123 for every dollar Congress gives them.
The Inspectors General that Trump fired in his first week had collectively identified over $50 billion in savings in fiscal year 2024. They were the fraud detectors. They were the ones who actually found the waste.
DOGE’s response was to fire them.
The USAID Inspector General, Paul Martin, released a report criticizing the aid freeze. The next day, he was terminated. The people doing the oversight were removed so the cuts could proceed without scrutiny.
GAO Comptroller General Gene Dodaro told Congress that DOGE barely contacted his office in the first month of operation. They had a 154-page roadmap for finding savings. DOGE ignored it. When GAO started investigating whether DOGE’s funding freezes violated the law, DOGE tried to “audit” GAO instead. They showed up at a congressional agency and attempted to assert authority they didn’t have.
They didn’t use the guardrails. They came for them.
Confirmation Bias as Methodology
Before any audit, before any review, before anyone checked a single contract, Elon Musk called USAID a “criminal organization.” He said it was “time for it to die.” Trump called the contracts fraud. No evidence. Just the conclusion.
This is textbook confirmation bias. You start with the answer, and then you look for evidence that supports it. You ignore everything that doesn’t fit. You fire the people who might tell you that you’re wrong.
Real evaluation asks different questions. What’s working? What’s not? How do we know? For whom? At what cost? Compared to what alternative?
DOGE asked one question: how do we justify the cuts we already decided to make?
When you go in knowing the answer, you don’t learn anything. You just confirm what you wanted to believe. You find the contracts that sound funny and put them on a website. You don’t ask whether the circumcision program was actually reducing HIV transmission by 60 percent. You don’t ask whether the monitoring contract was the thing preventing fraud in the first place. You don’t ask what happens to the people on the other end when the money stops.
And when you have power, your confirmation bias becomes policy. And your policy becomes 640,000 deaths.
What Evaluation Is Actually For
Evaluation exists to prevent exactly this.
It slows things down on purpose. It asks uncomfortable questions before the decision, not after. It builds in accountability to the people affected, not just the people in charge. It requires you to define what success looks like before you declare victory.
None of that happened here. DOGE moved fast because moving fast was the point. Speed forecloses deliberation. It prevents the questions that might complicate the narrative. It makes the cuts irreversible before anyone can measure the damage.
I wrote a few weeks ago about AI governance running the same extractive playbook as the development sector: taking without asking, declaring collaboration while concentrating power, framing extraction as partnership. DOGE is the domestic version. Same logic. Same refusal to ask who’s harmed. Same confidence that the people making the decisions don’t need to hear from the people affected by them.
The evaluation field has a term for what DOGE did. We call it “decision-based evidence making.” You decide what you want to do, and then you make the evidence fit. It’s the opposite of what evaluation is supposed to be. And it’s what happens when the people with the chainsaw decide the guardrails are in the way.
What’s Left
640,000 dead. $244 billion lost. An agency dismantled. An oversight system gutted. And no clear accounting of what was actually saved, if anything.
DOGE has now been quietly dissolved, eight months ahead of schedule. The principles, we’re told, will live on through other agencies. The deaths will too. The soft power, the HIV patients who lost access, the children who died of preventable diseases, the institutional knowledge that was fired. These aren’t coming back, at least not quickly.
I don’t know what rebuilding looks like from here. The funding won’t return with a signature. The trust won’t return at all. Governments around the world watched the United States abandon its commitments overnight. They’re not going to forget.
What I do know is this: the systems existed. The expertise existed. The data existed. We had ways to find waste and fraud that actually worked, that had track records, that returned more than they cost.
They chose the chainsaw anyway. And we’re all worse off because of it. Well, most of us are. Guess who isn’t.
Anthralytic is a strategy and evaluation studio using human expertise, data, and AI to help mission-driven teams clarify and amplify their impact.
