When Trump appointed Elon Musk to head the Department of Government Efficiency, both supporters and critics immediately began telling stories that seemed obviously false. The administration claimed DOGE would save trillions (impossible given its limited scope and two-year timeline). Critics portrayed it as a dictatorial power grab (despite having no actual enforcement authority beyond publishing recommendations).

I found myself getting increasingly frustrated with this smoke screen. If everyone's lying about what DOGE is, then what is it actually doing? And why?

The Musk factor adds another layer of confusion. Is he the central character in this story, or a flashy distraction from the real governance changes happening? If he is the story, what's his actual agenda? The technocratic efficiency narrative doesn't quite align with targeting high-profile aid programs like USAID, but the "right-wing revenge" narrative doesn't explain his data-driven focus.

As I dug deeper, I noticed that this pattern - creating a parallel inspection authority outside normal bureaucratic channels - has historical precedents. And those precedents can tell us something important about what's happening now.

The People's Tyrant: Revolutionary Dictatorships

In states with democratic or republican traditions such as ancient Athens or Rome, sometimes the aristocratic element (wealthy families with traditions of excellence and public service) becomes an entrenched oligarchy, using its outsized participation in key governance bottlenecks to extract resources for themselves, and combining their control over formal governance mechanisms with their capacity to purchase influence outside of formal governance mechanisms, to block more democratic attempts at reform. In such cases, sometimes the dispossessed majority would support a single figure (with some preexisting power base and credibility) to serve as in effect a temporary chief executive, to use a combination of formal power and military intimidation to force reforms - frequently land redistribution and the restoration of civil rights to poorer citizens - that were otherwise infeasible.

Julius Caesar is an unambiguous example. For generations, a clique of wealthy Romans had used their power to block land reforms and protect their privileged and profitable access to state lands, through their control of the Senate, the political weight of their patronage networks, and through outright assassinating reformers. (See What is a republic? A Roman aristocratic perspective.) Caesar came from an old and prestigious but relatively economically marginal Roman family, acquired influence through military command,¹ and eventually brought his expeditionary army home to occupy Rome and install himself as dictator (a title traditionally invoked for military emergencies), to enact the sorts of reforms the aristocracy had blocked. Ultimately persistent opposition forced him to fight a civil war and declare himself dictator for life.

Athenians used the word “tyrant” to refer to approximately the same function Julius Caesar served, and it seems that figures like Peisistratus served a similar role to him. Wealthy and dramatic aristocrats such as Plato didn’t approve of tyranny, but they did recognize it as aligned with and emerging from democracy, and Plato treated tyrants as worth engaging with to promote reform. Tyrants themselves frequently made use of personal bodyguards and engaged in overt violence within the city to stay in power. Frequently they chose, whether reluctantly or happily, to make use of people whose willingness to commit such violence was motivated more by getting to feel powerful than by a principled civic commitment to reforming the government.

Information Control and Institutional Resistance

Insofar as an institution is committed to accurately informing and following the orders of its notional head, the person at its head can effectively govern alone. But if a large, complicated institution is committed to resisting certain orders, it can make it prohibitively difficult for its notional leader to know what’s going on well enough to understand which orders have or have not been carried out, or even which orders are possible to execute. In such situations, effective leadership depends on the possession of external sources of intelligence.

Alexis de Tocqueville’s book L'Ancien Régime et la Révolution describes a process in which the French state’s capacity to collect taxes originally depended on various local processors of information who could use their control of information to skim some taxes for themselves. The process of bureaucratization and legibilization that cut out the middleman by making tax collection more transparent and standardized culminated in the French Revolution, which forcefully repudiated the traditional rights and property claims of such intermediaries, mainly the aristocracy and clergy, in some cases by beheading them.

The Roman Catholic Church itself developed the Inquisition to check that local church institutions were complying with Roman directives. The Russian Czars developed an institution of secret police that outlived the Czardom.

Donald Trump, recently elected for a second Presidential term, has DOGE, run by Elon Musk, with a two-year expiration date, and without a license to kill.

The Politics of Administrative Visibility

As far as I can tell, the primary function of DOGE is to inspect the contents of various Federal agencies’ databases, and similar sorts of information-processing to check which goods and services are being purchased and on what basis. If the President tells you not to spend money on activity X, it’s relatively easy to simply tell him you stopped, or that you weren’t doing that in the first place, or that you have contractual obligations to disburse funds, and to generate executive summaries congruent with this story. If the President insists you give an outside audit team access to your databases, it’s quite a bit harder to quickly generate a plausible fake internal database that tells the story you want. So you’re more likely to simply give him access to the real database and accept the loss of information control.

When the chief executive can more directly inspect the spending of the organization over which they preside, they are likely to cut expenses that are:

• Genuine cases of waste and fraud that were tolerated to ease internal tensions.
• Genuine cases of waste and fraud that were vehicles for funding illicit activities. (E.g. the Iran-Contra deal.)
• Legitimate programs that the executive mistakenly believes to be illegitimate, due to limited trust and communication.
• Programs that the executive knows to be legitimate, but thinks they can score political points by cutting.
• Programs that everyone agrees are good, because it’s hard to find intermediates confrontational enough to make such cuts who don’t also maliciously or sadistically enjoy the confrontation.

This necessarily creates some ambiguity as to what the chief executive is trying to do, and what they’re doing by accident. However, the discourse on DOGE suffers from additional sources of ambiguity.

There’s a technocratic story in which DOGE is mainly being used to solve waste due to intermediation common in imperfectly centralized systems. Then there’s USAID.

Everyone's International to Someone: Foreign Aid, Foreign Influence

My Twitter feed is flooded with Effective Altruist friends posting anguished threads about DOGE's evident cuts to USAID programs like PEPFAR - interventions with reputedly stellar quality-adjusted-life-year-per-dollar metrics. These cuts suggest something beyond mere efficiency optimization is at work.

To be fair, it's entirely possible that PEPFAR is, in effect, a hostage taken by USAID; while Secretary of State Marco Rubio issued multiple waivers specifically exempting PEPFAR from a spending freeze, it appears that PEPFAR may be dependent on other USAID programs, such that it's difficult to cut one without the other.

Since the end of WWII and the beginning of the Cold War, the US State Department and CIA have built a “soft power” apparatus in which the US intervenes in the internal politics of other countries through overt and covert means such as propaganda, funding favored media and political organizations in target countries, and conditioning aid or other cooperation on compliance with US directives. Naturally, this soft power apparatus is used partly to maintain its own capacities to influence and coerce. USAID is a soft power institution.

During the Cold War, these soft power institutions were part of the way the corporate-capitalist US resisted the expansion of the more overtly centrally planned Soviet Union. After the fall of the Soviet Union, the US continued to use soft power to expand America’s political and cultural influence. From one perspective, this was a liberal power acting to further human rights for all people by expanding the reach of locally accountable government. From another, it was encroachment on the spheres of influence of rival powers like Russia and China.

While some on the American right have always been skeptical of the methods of such internationalist bureaucracies, before 2020 the Republican Party was willing to tolerate them, because of their contribution to America’s capacity to deter potential aggressors and otherwise deliver foreign policy victories to the credit of whichever party was in power. But during the COVID-19 state of emergency, these soft power institutions used their methods to impose a global censorship and propaganda regime that impinged on Americans’ ability to discuss the situation with each other.²

The libertarian and Jacksonian aspects of the American body politic found that situation intolerable. The outgoing Republican president was even banned from Twitter! So in 2022 Elon Musk bought Twitter, and in 2024, Donald Trump was reelected with a governing coalition committed to dismantling at least part of the soft power apparatus, and appointed a head of USAID who was already committed to dismantling it.³

Elon Musk takes an ambiguous position as to whether the technocratic or political story is the primary motive for DOGE. He claims to be focusing on the clearest most unambiguous examples of waste and fraud, and to be motivated by the need to reduce the government’s budget deficit. He claims that USAID was targeted because of the high observed rate of fraud and noncompliance. But he also tweeted that USAID is a leftist psyop, and told Joe Rogan that he’s avoiding some cuts because he’s afraid for his life.

This doesn’t seem like how he’d behave if the technocratic story were true. It seems consistent with a scenario in which Musk is dismantling an entrenched power structure and talking about the deficit to throw smoke. It also seems consistent with a scenario in which Musk is vibing with an incoherent set of right-wing memes, and trying to do and say whatever gets him favorable attention in the relevant scene, even when those don’t add up to a coherent program.

Either way, while USAID's hostages are mainly African children, the CIA actively assassinates adults, and has intervened on US citizens with programs like MKULTRA, so maybe that’s why USAID has been targeted and the CIA hasn’t.

1. Incidentally, it seems to me that while the function of US noncommissioned officers is mainly a military one, to concentrate decisionmaking authority in people with experience, expertise, and demonstrated good judgment, the function of the commissioned officer corps is similar to the function of the “cursus honorum” of Rome: to keep the military accustomed to follow the orders of members of the aristocracy. In their case, that was the sort of people with the means to seek elected ofice. In ours, the collegiate system, the successor of the Roman Catholic system for educating clergy. So the function of commissioned officers is to keep the army under the control of the (secular) church. ↩︎
2. See Joe Rogan’s interview with Mike Benz https://open.spotify.com/episode/2rXdCTkipx2Iu5dX1Gh0s5?si=T42n-H1-Q0yNxLms9qKWJg
   See also the Twitter Files (Wayback Archive, Icelandic Archive), and specifically:
   Trump’s ban: (Independent Archive, Wayback Archive, Icelandic Archive)
   COVID Censorship ↩︎
3. 50 Thoughts on DOGE (Wayback Archive, Icelandic Archive)
   Peter Marocco (Wayback Archive, Icelandic Archive)
   Politico: Contentious USAID Appointee (Wayback Archive, Icelandic Archive)
   Politico: USAID DIssent Memo (Wayback Archive, Icelandic Archive) ↩︎