communicoupling · concept 18 of 26

The gap is in the
boundary, not the person.

Where does durable inequality between groups come from, when the people in them are, individually, the same? Charles Tilly's answer relocates the cause: not in the traits of persons but in relational mechanisms operating at the boundary between categories. Draw a line, run a few simple processes across it, and a persistent gap opens between people who differ in nothing.

We explain the gap between groups — in pay, in advancement, in wealth — by the qualities of their members: talent, effort, culture. Tilly showed you can generate a large, stable, self-reproducing inequality with populations that are statistically identical, purely from what happens at the line between a category "in" and a category "out". Exploitation, opportunity hoarding, emulation, and adaptation do the work.

The inequality is real; the trait difference is zero. Below is an honest agent-based model. Two categories of people carry the same underlying ability. Switch on the boundary mechanisms and watch a durable gap open between them — while a control readout keeps confirming the ability difference is exactly 0.00.

Live agent model · 40 in-group + 40 out-group · identical ability · four boundary mechanisms

Presets
Boundary mechanisms · toggle any live
True ability gap (in − out)
0.00
Simulation
Control · true ability difference
0.00
The two populations are drawn from the same ability distribution.
Outcome · wealth ratio (in : out)
1.00×
No mechanism running — identical people, identical outcomes.
in-group mean wealth out-group mean wealth round 0
IN-GROUP · category Amean 12.0
OUT-GROUP · category Bmean 12.0
Start with identical worlds: no mechanism is running, ability is equal, and the two lines sit on top of each other. Then switch on hoarding + exploitation and press Run.

The gap is at the boundary

Tilly's move: inequality is a relation, not a trait.

The ordinary explanation of a group gap is individual and internal: this group earns more because its members are, on average, more able, harder-working, better cultured. The gap lives inside people, as a property they carry. Tilly's Durable Inequality refuses this. He relocates the cause to the boundary — the categorical line itself — and to a small set of relational mechanisms that run across it. The people can be identical; the line does the work.

The model above is built to make that claim testable rather than rhetorical. Both categories draw their ability from one distribution; set the gap slider to 0.00 and the control readout will confirm, every round, that the true ability difference stays exactly zero. Nothing about a person predicts their group's fate. What predicts it is whether the boundary is switched on — whether exploitation is siphoning the out-group's product to the in-group, whether opportunity hoarding is reserving the good positions, whether emulation is copying the line into more and more contexts, and whether adaptation has grown routines around it. The gap you watch open is manufactured entirely at the boundary. Turn the mechanisms off and the populations are, again, the same.

What to try

Open it, spread it, and watch it stick.

01

Set abilities equal, then bite

Leave the gap at 0.00. Switch on hoarding and exploitation and Run. Within ~40 rounds the in-group line climbs, the out-group line sags, and the wealth ratio settles well above — a durable gap between people who differ in nothing.

02

Add emulation — it spreads

Now switch on emulation. The boundary is copied into more and more contexts: coverage climbs toward 100%, and the same two mechanisms now reach further, widening the gap without any change in ability. Inequality is contagious across organisations.

03

Add adaptation — it sticks

Switch on adaptation. Routines accrete around the boundary — an entrenchment stock that keeps rising while the line is live. The gap becomes load-bearing: everyone's daily arrangements now assume it.

04

Now turn the cause off — the ratchet

With adaptation on, switch exploitation and hoarding back off. The gap does not snap shut. Entrenchment decays only slowly, so the inequality persists for many rounds after its original cause is gone. Removing the boundary is far harder than installing it.

Identical people, unequal outcomes

"They earned it / they lacked it" reads the boundary as a trait.

Watch the finished run and the temptation is overwhelming: the in-group is visibly wealthier, so surely its members were more able, more diligent, more deserving. But you set their ability equal yourself. The wealth is downstream of the boundary, and inferring merit from wealth runs the arrow backwards — it treats an outcome of the mechanism as if it were the mechanism's cause. Position is being read as proof of the very quality that position was manufactured to withhold.

This is collider bias wearing a moral costume. Group and outcome are both wired to the boundary; conditioning on the outcome — "look how well they've done" — induces a phantom correlation between group membership and merit that was never in the ability data at all. The instrument holds ability fixed at 0.00 so you can see the phantom form. Every real inference of desert from a person's rung on the ladder risks exactly this move: mistaking where the boundary put them for what they are.

Why it endures

Categories are cheap to install and expensive to remove.

Durable inequality is durable because it is categorical, not individual. A category is a reusable device: once a society has a clean line — citizen/foreigner, boss/worker, member/non-member — it can be matched, over and over, to whatever advantage is going. Tilly's deepest move is that an exterior paired category (the unequal pair everyone can see) gets stapled to an interior one (a distinction inside an organisation), so that the boundary already lying around gets reused to sort people into positions. The advantage rides on the line, and the line is portable — it travels from firm to firm by emulation and settles in by adaptation.

Installing a boundary is a single categorical act; removing it means unwinding every routine, expectation, and paired-category match that has since grown around it. The gap outlives the intent, the generation, sometimes the original injustice — reproducing itself through structure long after anyone chose it. Inequality persists not because the disadvantaged keep failing a test, but because the line keeps being useful to whoever sits on its favourable side.

The mapping

Mechanism ↔ social life.

In the modelIn social life
a categorical boundaryThe line between an in-group and an out-group — the reusable device advantage gets matched to.
exploitationCapturing others' surplus: the product of the out-group's effort is siphoned to the in-group.
opportunity hoardingReserving a valuable niche for one's own — the good positions kept inside the category.
emulationThe boundary copied across organisations, so the same line applies in more and more contexts.
adaptationRoutines that entrench the line — daily arrangements built around it that make it sticky.
the durable gapPersistent inequality between identical people: the ratchet that outlives its own cause.

Where it tears

Three honest limits.

This shows sufficiency, not the whole story.

The model proves boundaries are enough to manufacture a durable gap with equal people — it does not prove that real gaps have no other source. Genuine differences exist and have histories of their own. Worse, boundaries were rarely drawn on blank ground: they were laid along pre-loaded lines of prior conquest, expropriation, and violence. Setting the ability slider to zero is a demonstration device, not a description of how any actual line was born.

"It's structural" can erase the agents.

Relocating cause from persons to boundaries is clarifying, but the boundary did not install itself. Specific people hoarded specific niches and specific institutions emulated them; naming only "the structure" can flatten those choices into weather, dissolving both responsibility and the particular injuries done to particular people. Tilly explains durability; he does not excuse the agents who built and profit from the line.

The numbers are hand-set toys.

The exploitation rate, the niche premium, the emulation and entrenchment speeds are all dials I chose to make the mechanisms legible. Nothing here is estimated from data. The claim is about the class of mechanism — that lines plus relational processes suffice to produce durable categorical inequality — not about any particular magnitude. Read the shape of the curves, not the decimals.