The normative layer under the inequality debate, and it starts with an uncomfortable fact: eliminating inequality is achievable, has been achieved, and is universally rejected. Levelling by destroying the top works — Mao, Pol Pot, and the reverse-dominance hierarchies of hunter-gatherer bands all did it, and produced equality. Since essentially nobody who campaigns against inequality wants that, the stated goal is not the operative goal. What is actually being tracked is a curve rather than a target: some inequality is functional (it prices capability and rewards taking risks), too much is destabilising (it buys unrest), and the real question is where on that curve we are and which direction to move — which is a far harder claim to make and a far easier one to argue honestly. This page establishes the subtraction (the stated reason underdetermines the objective) and deliberately declines the substitution (naming the hidden real motive), because that guard is what separates this from conspiracy framing.
Links: Economics, Positional Goods (the descriptive twin — that page says where inequality is unavoidable, this one asks what it is for), Subsistence vs. Participation, Absolutes and Differentials (the parent guard — subtraction not substitution, and the “criterion satisfied while policy continues” test this page runs), Risk and Entrepreneurship (why profit rewards correct bets), Variance Is Not Luck, Civilizational Cycles — structural-institutional (Turchin: instability as the cost of the upper end), Econ Nerds — Three Myths about Inequality (origin), Value and Profit
Thesis page (portable). Opens the normative layer the vault had been missing — it holds a great deal on how to measure and argue about inequality and had almost nothing on whether any of it warrants action. This is the first pass, not the finished treatment.
Chris: “The claims in the video that one can eliminate inequality by shaving off the top has historical basis, and thus shows eliminating inequality is not the real goal.”
The argument in full:
This is the same instrument Absolutes and Differentials runs on war justifications: a criterion satisfied while the policy continues was always a proxy. Here the criterion is satisfiable on demand and is refused, which is the stronger version — you do not even have to wait for events to run the test.
What survives is a subtraction. The goal is not “less dispersion in the distribution,” because the cheapest route to less dispersion is on the table and nobody takes it. Something else is being tracked, and the honest next step is to ask what — not to announce it.
Do not convert this into “so what they really want is power.” That is the substitution the parent page forbids, and it re-commits the original error — swapping one simple stated motive for one simple hidden motive, sign flipped. The people who reject levelling-by-destruction reject it for good reasons they will state if asked: it kills people, it destroys the productive capacity that made the surplus, and it substitutes a worse hierarchy for a better one. Those are real objections, sincerely held.
The finding is not that anyone is lying. It is that the objective as stated does not pick out what they are defending, so the argument is being conducted in the wrong terms by everyone in it — which is the diagnosis, and is more useful than an accusation.
Replace the target with a curve. Chris’s formulation: “some of it is actually good as it creates incentives and rewards capability, too much and it causes unrest.”
The functional end. Dispersion in outcomes is how an economy prices capability and effort, and how it pays for risk. If the surgeon and the roofer and the failed founder all end at the same place, the signal that routes people to hard, scarce, or risky work is gone. Risk and Entrepreneurship is the vault’s statement of the underlying point: profit is the reward for correct bets under uncertainty, and a bet with no differential payoff is not a bet. Some inequality is not a regrettable side-effect of a working system — it is the system working, and it is the residue of exactly the process that produced the growth that makes redistribution possible at all.
The destabilising end. Past some level, dispersion buys unrest: legitimacy erodes, positional competition intensifies, and surplus elites who expected to be at the top become counter-elites (Turchin). Note the awkward datum recorded on that page — the 2023 PLOS ONE result that inequality alone out-predicted elite overproduction as a predictor of instability. That cuts toward this section: whatever the mechanism, the level appears to carry real predictive weight for unrest.
So the shape is an optimum, not a monotone. Both endpoints are bad and for unrelated reasons — one destroys the incentive to produce, the other destroys the consent to be governed. This is why the debate resists one-liners in both directions: “inequality is bad” and “inequality doesn’t matter” are each a claim that the curve is monotone, and both are wrong.
The honest consequence, which neither coalition likes: the interesting question is not whether inequality is bad but which side of the optimum we are on, and by how much — and that is a quantitative claim requiring a measure, which returns you to PSZ vs. Auten–Splinter and the ~40% of income invisible to tax returns. The normative question is hostage to the measurement question, which is a genuinely uncomfortable place to end up and is probably the true state of affairs.
Having subtracted, the candidates for what people mean when they say “inequality”:
| Candidate | Would explain | Test that would confirm it |
|---|---|---|
| The floor is too low | why levelling-by-destruction is rejected — it doesn’t raise the floor | concern should track absolute deprivation, not dispersion |
| The process is unfair — gains came from rents, not production | why particular fortunes draw fire and others don’t | concern should track source of wealth, not amount |
| Positional goods are unaffordable | why the complaint concentrates in housing, healthcare, admission | concern should track the restricted basket, not the Gini |
| Status anxiety / expectation gap | the timing mismatch (concern rising as the trend flattened) | concern should track expectations, not conditions |
These are not exclusive and probably all contribute. The one thing the page will not do is pick one and declare it the real motive — that is §1’s guard. What it does claim is that each of these is a better-specified objective than “reduce inequality,” and each implies a different policy, which is why the substitution matters practically and not just epistemically.
Note that candidate 3 is the vault’s own live hypothesis and therefore the one to be most sceptical of.
Portable beyond this topic:
The achievable-and-refused test. If a stated goal has a known, available, demonstrated method, and the people holding the goal refuse the method, the stated goal is not the operative one. Ask what the refusal protects — that is closer to the real objective than anything in the stated position.
It is cheap to run and it is not an accusation: the refusal is usually principled, and the principle is the thing you were looking for. Its failure mode is over-application — some goals are refused-in-method for straightforward moral reasons that leave the goal intact (one may want fewer murders and still refuse to achieve it by mass preventive detention). The test establishes that the stated objective is underspecified, not that it is insincere, and the difference is the whole discipline.