The Loyalty Test

Court depositions unsealed this week from a lawsuit against the Department of Government Efficiency contain a sentence that deserves careful study. In March 2025, a DOGE associate named Justin Fox emailed Michael McDonald, the acting head of the National Endowment for the Humanities, with a deadline and a demand. At the end of the message, Fox wrote: "We're getting pressure from the top on this and we'd prefer that you remain on our side but let us know if you're no longer interested."

McDonald was already cooperating. He had aligned his political views with the administration and actively sought to work alongside DOGE. He was, by any reasonable measure, already on their side. The email was not informational. It was a test.

The Mechanism

A loyalty test is not an inquiry about loyalty. It is a demonstration of the power to question it. The test works not by gathering information but by inducing a specific psychological state in the recipient: the sudden awareness that their status is conditional, revocable, and currently under review.

Fox's email contains three components that are worth separating:

The pressure attribution. "We're getting pressure from the top" accomplishes two things. It invokes a higher, unnamed authority whose expectations cannot be questioned or appealed. And it positions Fox not as the source of the demand but as a fellow traveler passing along something he himself cannot control. This is deniability architecture. The threat arrives wrapped in sympathy.

The stated preference. "We'd prefer that you remain on our side" converts compliance from an obligation into a favor. McDonald is not being ordered to cooperate. He is being given the opportunity to be preferred. This framing inverts the power relationship: the compliant party feels they are doing the demanding party a service, rather than submitting to a demand.

The opt-out clause. "Let us know if you're no longer interested" creates the illusion of genuine choice. This is the most sophisticated element. A direct threat leaves the target feeling coerced and potentially motivated to resist or document the coercion. An opt-out clause makes the whole exchange feel voluntary. Choosing to comply feels like an independent decision, not a capitulation. The opt-out exists to ensure the target never frames the interaction as pressure.

Why It Worked

McDonald, by his own testimony, had already signaled his alignment. DOGE knew this. The email was not sent because his loyalty was in doubt. It was sent because testing loyalty, even when it is not in question, achieves something distinct from confirming loyalty: it establishes that loyalty is a category that can be revoked.

In the days after the email, NEH terminated approximately 1,400 grants and issued layoff notices to 116 employees, roughly two-thirds of its workforce. McDonald signed off on all of it. Whether he would have done so without the email is unknowable. What is clear is that the email preceded the action, and the action was total.

The structural logic is this: an institution that terminates grants in numbers exceeding its entire prior history is not executing a judgment. It is demonstrating that its judgment has been suspended. The loyalty test accelerates this process by short-circuiting deliberation. When a target is managing conditional status, they are not evaluating requests on their merits. They are managing the relationship.

The Anatomy of a Structured Threat

This pattern recurs across contexts far removed from government: in executive suites when a board chair contacts a CEO ahead of a contested vote, in partnerships when a dominant investor mentions "alignment" before a term sheet, in social hierarchies when a host tells a guest they hope they're "comfortable" with an arrangement that benefits only the host.

The last signal is the most reliable. Loyalty tests are most frequently administered to the already-loyal. Their purpose is not to screen for defectors. Their purpose is to deepen compliance by making the target aware that compliance is being monitored and that its continuation is conditional. The test does not reveal who is loyal. It manufactures a higher grade of loyalty in those who already are.

Fox's deposition testimony, now partially unsealed after a brief court order to suppress it, revealed that he could not define DEI to the satisfaction of opposing counsel, that he characterized a grant documenting Reconstruction-era racial violence as insufficiently universal, and that he was 23 years old at the time of these decisions. None of this is relevant to the mechanics of the email. The email worked regardless of who sent it, because the mechanism is not personal. It is structural. The authority behind the preference is what matters, and that authority was real. The cooperation followed.

---

The loyalty test is not a dramatic instrument. It does not arrive as an ultimatum. It arrives as a preference, a hope, an expression of concern. The target often files it under "communication" rather than "coercion." That misclassification is the point.