Robert Conquest’s 3 Laws of Politics (GPT-5.6 redux)

---

Robert Conquest's three celebrated laws of politics are usually given as follows:

1. Everyone is conservative about what he knows best.

2. Any organisation not explicitly right-wing sooner or later becomes left-wing.

3. The simplest way to explain the behaviour of any bureaucratic organisation is to assume that it is controlled by a cabal of its enemies.

When I quoted these on this blog in 2014, I offered three glosses as to what Conquest might have meant:

1. Head over heart where it's important.

2. Eventual capture by the many idealists with a retributive agenda.

3. Regulatory capture.

Let's ask GPT-5.6 to review my response at the time.

---

Looking at them again twelve years later, I think all three of your responses deserve revision. Conquest's jokes are better than your explanations of them. Beneath the political provocation lie three rather general propositions about knowledge, institutions and incentives.

The first law is not really about choosing the head over the heart. It is about expertise. The more closely one understands a complicated system, the more one discovers that its apparently irrational features often embody constraints invisible to outsiders. Radical simplification is easiest when somebody else has to deal with the consequences.

The network engineer knows why the ugly legacy system cannot simply be ripped out. The surgeon knows why the irritating procedure exists. The programmer has learnt what usually happens when somebody proposes rewriting the twenty-year-old application from scratch. Expertise does not necessarily make people politically conservative. It makes them conservative about interventions in systems whose failure modes they particularly understand.

Conquest's first law is therefore a kind of informational Burkeanism: knowledge breeds respect for somewhat esoteric constraints. We are most susceptible to schemes of radical reconstruction precisely where we know little.

The second law is more interesting than Conquest's partisan formulation suggests. Organisations do not have opinions in quite the same sense as people do. They acquire opinions because particular kinds of people have stronger incentives to shape them.

A hospital, university, museum, charity or professional association contains many people principally interested in medicine, scholarship, paintings, charitable work or their profession. Such people may have perfectly definite political views, but they did not join the organisation in order to make it express them. Alongside them are people with much stronger views about what the institution itself ought publicly to stand for. They write policies, join committees, redefine missions and supply the moral vocabulary in which institutional decisions are discussed.

Those with intense preferences repeatedly defeat those with weak ones. The latter may disagree, but usually not enough to spend Tuesday evening arguing about the wording of the diversity statement.

That makes Conquest's second law less a law of Left and Right than a special case of Mancur Olson's general thesis. Organised minorities with strong incentives beat diffuse majorities with weak incentives. In the cultural circumstances of modern Western institutions, the energetic institutionalisers have very often been progressive. Under a different social equilibrium the ideological direction could conceivably be different.

So I would now rewrite the second law as: unowned institutions drift towards the ideology of those most determined to run them.

Your old explanation of the third law as regulatory capture was simply too narrow. Regulatory capture describes the familiar case in which an industry succeeds in bending its regulator towards the interests of the regulated. Conquest is making a larger and rather nastier joke.

Institutions can behave as though their enemies had taken control of them without anybody intending this outcome at all.

A bureaucracy rewards behaviour actually visible to the bureaucracy. Staff learn to satisfy procedures, targets, budgets, audit requirements and career incentives. Managers minimise risks to themselves. Departments defend their territory. Metrics substitute for purposes. The people who become most successful within the organisation are increasingly those who have mastered the organisation rather than necessarily those most committed to whatever it was originally created to achieve.

No conspiracy is required. Indeed that is the point of Conquest's joking insight. Ordinary institutional incentives can produce outcomes so perverse that sabotage seems the more charitable explanation.

This is the territory of principal-agent problems, Parkinson, Goodhart and Pournelle's Iron Law of Bureaucracy rather than regulatory capture. The organisation gradually becomes one of the things it exists to maintain.

So my revised glosses would be:

1. Knowledge breeds respect for constraints.

2. Unowned institutions drift towards the ideology of those most determined to run them.

3. Bureaucracies eventually optimise for themselves rather than their ostensible purpose.

Stripped of some of Conquest's conservative mischief, the laws actually become stronger. The first concerns information, the second collective action and the third principal-agent failure.

The explicitly political content is somewhat incidental while the institutional pessimism is certainly not.

---