The Purge Reflex: Why Every Institution Eventually Starts Eating Its Own
Somewhere in your professional life, you've probably encountered the loyalty test. Maybe it was explicit — a political litmus question in an interview, a request to publicly endorse something you weren't sure about. Maybe it was subtle — a manager who wanted enthusiasm, not analysis. Maybe it was a company-wide email about "culture fit" that arrived three weeks before a round of layoffs that somehow only affected the people who'd raised concerns in the last all-hands meeting.
Here's what the historical record says about that: it's not new, it doesn't work, and the people running the test are usually the ones who should be worried.
The Han Dynasty Figured This Out — Then Forgot It
China's Han Dynasty, which ran from around 206 BC to 220 AD, was one of the most administratively sophisticated governments the ancient world produced. They built a civil service exam system, they developed centralized bureaucracy, and they thought seriously about how to staff an empire with competent people.
They also periodically lost their minds and purged their own bureaucracy.
The pattern was consistent: a new emperor or powerful faction would come to power, feel uncertain about the loyalty of existing officials, and begin requiring increasingly specific demonstrations of allegiance. Officials who had served well for years were suddenly suspect because they had served the previous regime. The test of loyalty became ideological purity rather than administrative performance. People who were genuinely good at governing got removed. People who were good at appearing loyal got promoted.
The result, documented across multiple dynasties and not just the Han, was that the institutions that ran the purges ended up staffed by people who had demonstrated one skill: surviving purges. That skill and the skill of competent governance overlap less than you'd hope.
This is not a subtle historical lesson. It's sitting right there in the primary sources, available to anyone who's ever run an empire and wanted to avoid this particular failure mode. And yet.
McCarthy's Math Problem
The Red Scare of the late 1940s and 1950s is probably the most documented American case study in institutional loyalty testing, and it's worth looking at it not as a political event but as an organizational behavior experiment at national scale.
The core logic of McCarthyism was this: if we can identify and remove disloyal people, the institution will be safe. It's a reasonable-sounding hypothesis. It's also wrong, and the wrongness is systematic rather than incidental.
Here's the mechanism. When an institution announces it is looking for disloyalty, it creates an immediate incentive for everyone to perform loyalty rather than to be loyal. These are different things. Performance of loyalty involves visible enthusiasm, public conformity, and — critically — the identification of others as less loyal than yourself. Actual loyalty involves honest engagement with institutional problems, including the willingness to say when something isn't working.
McCarthyism didn't find hidden communists who were sabotaging American institutions. It found people who were willing to accuse other people of being communists, which is a different skill set entirely. The State Department, the Army, Hollywood studios — institutions that went through aggressive loyalty purges in this period didn't emerge stronger. They emerged with their most independent thinkers gone and their most politically agile survivors in charge. That's not a security upgrade. That's a capability downgrade with extra steps.
The psychological research on this is pretty consistent: humans under surveillance and threat of punishment optimize for the appearance of compliance, not actual compliance. You get what you measure, and if you measure performed loyalty, you get a very good performance.
The "Culture Fit" Rebrand
Corporate America didn't invent loyalty testing — it just gave it a cleaner interface.
The phrase "culture fit" entered HR vocabulary in the 1980s and has been doing a lot of work ever since. At its best, it describes something real: whether a person's working style and values are compatible with a team's. At its worst — and organizational psychologists have documented this extensively — it functions as a loyalty test with plausible deniability. People who ask difficult questions, who push back on decisions, who raise concerns through official channels, get coded as poor culture fits. People who reflect leadership's preferences back to them get coded as strong culture fits.
The result is the same result you get in the Han Dynasty and in 1950s Washington: the institution gradually fills with people who are good at reading the room and bad at telling you what's actually in the room.
There's a term for this in organizational behavior research — it's called epistemic cowardice, and it's the predictable output of environments where honesty is punished and agreement is rewarded. The experimental data on psychological safety in teams — Google's Project Aristotle is the well-publicized example — shows that teams where members feel safe to disagree and raise problems dramatically outperform teams where they don't. This isn't counterintuitive. It's just consistently ignored by organizations that are in the grip of the purge reflex.
Why It Always Backfires
The loyalty test fails for a reason that's almost elegant in its simplicity: the people most likely to pass it are not the people most likely to be loyal.
Actual loyalty to an institution — the kind that involves honest feedback, genuine investment in outcomes, and the willingness to say uncomfortable things — requires a certain amount of security. People who feel secure in their position can afford to be honest. People who feel threatened optimize for survival. When you run a loyalty test, you threaten everyone. You select for the survivors. The survivors are, by definition, the people best at surviving loyalty tests, which means they are the people most willing to perform whatever the institution currently defines as loyalty, regardless of whether it's actually good for the institution.
Mao's Cultural Revolution purged the Chinese Communist Party's most experienced administrators and replaced them with ideologically enthusiastic teenagers. The result was a famine that killed tens of millions of people, because ideological enthusiasm is not a substitute for knowing how to run an agricultural system.
Enron had a performance review system called "rank and yank" that eliminated the bottom fifteen percent of employees annually. It created a culture where appearing to perform was more valuable than actually performing, where risk was hidden rather than disclosed, and where the people who raised concerns about accounting irregularities were the ones who didn't survive the next review cycle. The historical record of what happened next is available in any business school library.
The Lesson That Keeps Not Being Learned
Five thousand years of institutional history says the same thing in every language and every political system: organizations that can tolerate internal dissent outlast organizations that can't. The Roman Senate's ability to absorb criticism and debate — imperfect, corrupt, and chaotic as it was — lasted for centuries. The moment it became a rubber stamp, the institution became decorative.
The British civil service's tradition of "speaking truth to power" — the expectation that senior civil servants would tell ministers what they needed to hear rather than what they wanted to hear — is credited by historians as a significant factor in the stability of British governance across wildly different political administrations.
The institutions that survive are the ones that can hear bad news.
The institutions that run loyalty tests are, almost by definition, the ones that have decided they can't. And the historical record of what happens to institutions that can't hear bad news is sitting right there, five thousand years deep, waiting for someone to check it.
The data isn't hidden. It's just inconvenient.
