Talking to my fellow libertarians about DOGE

I was listening to Reason‘s Just Asking Questions podcast episode about the Department of Government Efficiency (DOGE), and I thought host Liz Wolf was too willing to accept Elon Musk and DOGE at face value. I’d noticed a similar attitude at the Reason Roundtable last week. I get where they’re coming from. After years watching people suffer at the hands of government employees, Musk’s abuse of them feels like payback. But that doesn’t make DOGE wise or effective.

I think I need to have an argument with my fellow libertarians to explain that. It’s a complex subject, but I’ve boiled my rant down to a bunch of short points:

Not like a business. You’d think people who advocate “running the government like a business” would recognize that what Trump and Musk are doing is not how successful business leaders manage change. They don’t rush around firing people willy-nilly. Effective leaders have goals, and they have plans to reach those goals. More importantly, they communicate their plans and goals to everyone else in the organization, getting buy-in from as many people as possible and building a culture that supports the change. DOGE staffers are just swinging an axe at anything they dislike.

Twitter. Some people point to Musk’s management of Twitter as proof that he knows what he’s doing. After all, he also fired lots of people very quickly there, and “the website’s still up.” But that’s no great accomplishment: Modern cloud-based applications are engineered to keep running without human intervention. Heck, I haven’t touched my demo Spelling Bee site in almost a year, and it’s still up, so it’s no surprise that Twitter/X keeps going. But people have been leaving the platform, and those who remain seem to be spending less time on it. Twitter revenue continues to decline and it’s estimated that Twitter has lost 88% of it’s brand value and 79% of its net worth since Musk took over.

Power not success. Some people defend Musk’s management of Twitter by saying his goal was to achieve power, not financial success, and therefore his current role in the Trump administration shows him to have been a super-genius. Even if we accept that framing, the fact that Twitter worked out well for Musk doesn’t change the fact that Twitter is still collapsing. It would be a disaster to let him do the same thing to the government.

Information Problem. Those of us who prefer free markets to central planning like to point out that central planners can’t possible have enough information to efficiently control the activities of an entire economy. A similar information problem affects large organizations like corporations. Or the federal government. Elon Musk and a few dozen DOGE employees can’t possibly drop in and immediately understand large government departments with complex goals, tremendous amounts of data, and thousands of people with specialized subject matter expertise. When DOGE staffers make pronouncements about waste and fraud after just a few days, they are engaging in what F.A. Hayek would call “a pretense of exact knowledge that is likely to be false.”

Time Problem. One way around the information problem is to distribute the decision making, as the free market does through its pricing mechanisms. The organizational equivalent requires setting a clear vision, communicating that vision down through layers of management, and identifying, listening to, and empowering knowledgeable and effective people at all levels. Given how quickly DOGE flits from agency to agency, they obviously aren’t taking the time to do any of that.

Mistakes. DOGE is making a lot of mistakes. Musk’s claim that USAID was funding Politico to the tune of $8 million turned out to be multiple government offices subscribing to Politico’s professional services since 2016. The claim USAID spent $50 million on condoms for Gaza has no basis in fact. And lately IT experts have been explaining that Musk has been making false claims about Social Security data because he doesn’t have a clue how the technology works.^[1]Musk now says he suspects there are tens of millions of dead people in the social security database. I’m confident that if 30% of social security recipients were dead, it would have gone unnoticed until now. He’s misunderstanding his dataset.

Not the real world. Fighting fraud is hard. We’ve heard estimates that annual government fraud could amount to as much as $500 billion. But by its nature, successful fraud goes undetected, and so it’s impossible to measure. It’s also hard to detect. DOGE may uncover a few instances of fraud by trawling through databases, but most people who commit fraud are well aware of how government record keeping works, and they make sure the official numbers add up. Real fraud detection involves difficult and time-consuming investigations in the real world.^[2]E.g. If you suspect a corrupt official is paying a fake contractor for work that is never done, you’d probably have to check the contractor’s records to see if they paid employees to actually do the work and then check if the employees actually exist and if they received the payments. Then you’d have to check if the contracted deliverables exist and evaluate whether they look like the billed amount of work went into them. DOGE staffers are just cosplaying as fraud detectives, and they are unlikely to think of anything that experienced government fraud investigators wouldn’t know about.

Firing people. In pursuit of its goals, DOGE leans heavily on firing people.^[3]Probably not firing them in the legal sense, which requires showing cause, but definitely terminating their employment. Some people argue that firings are the only way to effect change, but that’s often an excuse for a lack of leadership. They clearly aren’t working to a plan. They said they were going to fire the wasteful and underperforming, but lately they are getting rid of all probationary employees, i.e. new employees with less job protection. They’ve gone from identifying inefficiencies to simply laying off those who are easiest to lay off. Avoiding the most effective action to do the easiest thing is always a sign of feckless decision making.

Sociopathy: When Musk tore down USAID, he tweeted out “We spent the weekend feeding USAID into the wood chipper.” He was putting thousands of people out of a job, and he was joking about it. Russell Vought, Trump’s head of the Office of Management and Budget has said “We want the bureaucrats to be traumatically affected… We want to put them in trauma.” This is not how normal people behave. Even if it is good policy to lay off thousands of people, you’re not supposed to enjoy doing it. These are sociopaths who enjoy hurting people. They are the worst people to empower.

Buyouts. Thousands of federal employees have been offered buyouts — financial incentives to resign — and so far about 77,000 of them have accepted the deal. As is typical with any Trump/Musk idea, the implementation has been chaotic. Although people accept buyout offers for a variety of reasons, the people most likely to accept are probably those most likely to easily find jobs elsewhere, i.e. high-value employees.

Return to office. Government employees who work from home have been ordered to return to office. This would make sense if there was a reason to believe employees working from home are malingering, but statements from the Trump administration have made it clear that the return-to-office order is an attempt to make employees miserable so they quit. In fact, many employees who have accepted the buyout are doing so to avoid having to return to the office. There’s no reason to believe that work-from-home employees who can’t return are actually poorer performers than those who can. It’s essentially terminating people at random.

Babies and Bathwater. The wholesale gutting of departments and agencies risks shutting down valuable government services along with the wasteful ones. Cuts to budgets have already disrupted drug trials, veterans programs, and world-wide disease prevention, not to mention the Department of Energy cuts that are disrupting the National Nuclear Security Administration’s handling of our nation’s nuclear weapons.

Too sudden. When Musk took over Twitter, one of his most outrageous acts was shutting down the Twitter API, the access points for hundreds of applications that interact with Twitter, without warning. DOGE (and Trump in general) has been doing much the same with government services, often shutting things down almost instantly, and just as developers using Twitter’s API had no time to adjust, government agencies are having difficulty find alternatives quickly.^[4]There are rumors of an NNSA team getting shut down in the middle of transporting a “physics package” — the nuclear parts of a nuclear warhead — leaving them to secure the materials as best they could while trying to find out what’s happening. None of this had to happen this way: The shutdowns could have been stretched out over a year or two, giving the affected agencies time to go through their planning process for a much smoother transition.

Destruction of capital. Employees bring more than just labor hours to a job, they bring knowledge. And quite often that knowledge is learned on the job and very specific to the job. E.g. how to administer a complex grant program or maintain a large software codebase. In economic terms, this knowledge is human capital, and like other forms of capital — machines, vehicle, factories, etc. — it increases productivity but is expensive to acquire. Terminating knowledgeable employees is therefore equivalent to destroying machines, vehicles, or entire factories. This is fine if you really don’t need the capital any more, but given the recklessness of the DOGE operation, much of this capital will likely have to be acquired again, at great expense in the form of training and practice.^[5]Unless they can re-hire the same people quickly enough.

Unkindness. Whenever Trump policies make people suffer — due to lost jobs or having grant-funded medical care cut off — it’s common to hear that “the cruelty is the point.” But while Trump does enjoy punishing people he doesn’t like, I think far more often the suffering is due to simple callousness. It’s not always the case that they want to hurt people. Sometimes they just don’t care. Either way is unkind.

Regime uncertainty. Trump and Musk are trying to change many things, all at once, without warning or time for the rest of us to prepare. Then sometimes they discover they’ve made mistakes and they walk parts of it back. And they do a terrible job of communicating any of this to the public. Even worse, they do all these things by unorthodox and possibly illegal means, and at this point there are about 80 active court cases with about 20 orders delaying, pausing, blocking, or overturning Trump administration policies. This is an unholy confusing mess, and that has consequences: People and businesses don’t know what will be happening in the future, and this uncertainty makes them cautious. It makes them hesitant to commit resources to activities with uncertain outcomes. And hesitancy spending money is how we get recessions.

Too little. In 2024, the federal government spent $1.7 trillion on Congressionally mandated healthcare, including Medicare, Medicaid, the Children’s Health Insurance Program (CHIP), and Affordable Care Act (ACA) subsidies. Another $1.5 trillion went to legally required Social Security payments. And $892 billion went to interest on the debt, which has to be paid. After that, the largest budgeted discretionary item is the hard-to-cut defense budget at $872 billion. Various benefits for veterans, federal retirees, and a variety of anti-poverty programs^[6]The refundable Earned Income Tax Credit and Child Tax Credit, unemployment insurance, supplemental security income for the elderly or disabled, the Supplemental Nutrition Assistance Program, school meals, low-income housing assistance, child care assistance, home energy assistance, aid for abused or neglected children, etc. add another $1 trillion. Education adds $345 billion. Other major components including transportation, natural resources, agriculture, science and medical research, law enforcement, and international spending, which add up to $244 billion, with another $345 billion in miscellaneous programs. That’s almost $7 trillion. On the other hand, DOGE has specified $6 billion in direct spending cuts. Between the buyouts and direct terminations, DOGE has presumptively cut 87,000 federal jobs and it looks like they will layoff 220,000 probationary workers. If we generously assume each job costs $150,000/year, that’s $46 billion. Assume for the sake of argument that they also succeed in throwing the entire $72 billion USAID budget into the wood chipper, and DOGE will currently save about $124 billion.^[7]I think all this was accurate when I wrote it, but be aware that things are happening fast. We’re going through a lot of pain and confusion to save 1.7% of the budget, an amount which taxpayers will barely feel on April 15th. Meanwhile Congress is currently putting together a spending bill with $300 billion in new funding…

Indirect inefficiency. Us libertarians talk a lot about “smaller government” but too many make the mistake of equating the size of government with the size of the government’s budget. That’s not the only burden of government. Every time the government limits us or controls us, it imposes a cost on society. Every permit you have to apply for, every impact study you have to perform, every law or regulation you have to follow, every victimless crime that you can commit^[8]Marijuana and ecstasy, gambling, sex work, high-flow showers and toilets, short-barreled rifles, certificates of need, plastic straws, zoning, rent control, sodomy, raw milk, unlicensed hair care, etc. — all of these are a burden, and for many people these burdens are far worse than paying taxes. DOGE terminating thousands of people does nothing to lift this burden: We still have to follow those laws and regulations. But where following those laws requires us to interact with the government — obtaining licenses, receiving inspections, filing reports, waiting for evaluations, getting permits, etc. — DOGE’s reduction of the government workforce could actually leave us worse off. Just as any business reducing its workforce makes customer service worse, firing government employees makes dealing with the government even more tedious and wasteful for the rest of us. Adding weeks, months, or years of delay will do nothing good for our economy. The government may be more “efficient,” but the burden will make us all less efficient.

Chains and crutches. The Cato Institute uses the following analogy: Government both puts people in chains and gives them crutches, and when you take the government apart, you should be careful to remove the chains first.^[9]Source for the crutches and chains analogy. I’ve been advocating for limited government for decades, and I can’t understand libertarians who think limiting government should start with USAID, the NIH, and the NOAA instead of CBP, ICE, HSI, DEA, and ATF. Or note that it is well within Trump’s executive power to end the scourge of federal civil forfeiture, yet neither he nor his libertarian supporters are talking about it. These people worry more about DEI than police abuse or victimless crimes.

Procedure. Some libertarians say that procedural objections to Musk and Trump’s DOGE activities ignore all the good that DOGE is doing. I’ve think I’ve offered plenty of arguments that DOGE is not, in fact, doing good, and is often quite harmful, but I would also like to defend the importance of procedure. It is the procedural aspects of government — elections, Congressional voting, passing laws, veto powers, legally required administrative procedures, open hearings, limits on police activity, independent courts, constitutional review — the whole system of checks and balances — that keeps the government in line. Basically, if libertarians want government to follow libertarian values, they will do that by enacting procedures that constrain government action. You can’t get libertopia without making the government follow the rules.

In summary, DOGE-style changes doomed Twitter and won’t help our government, but they may empower Musk even further. DOGE’s sociopathybro staff have too little information and too little time to understand what they are doing and they’re working only from government databases, so they’re missing real fraud, making mistakes, discarding the good with the bad, breaking things without warning, getting rid of experienced employees, and generally being unkind, all without doing much to relieve legal and regulatory burdens, while destroying government capabilities, creating economy-threatening uncertainty for small gains, making government less efficient, and setting us up for unconstrained authoritarian rule.