Strong Men and Nosy Grandmothers

A familiar genre of urban complaint goes like this. Someone is screaming at no one on the train platform, and nobody nearby does anything. The complaint usually treats this as a policing problem, then arrives quickly at the question of whether the police are doing enough. What the complaint tends to miss is that formal policing was always a thin layer over a much thicker substrate of soft enforcement, performed by ordinary people, mostly without thinking about it.

That substrate is what I want to map. The shorthand I keep coming back to is that public space (the streets and trains and stores you walk into without an appointment) runs on two informal classes of enforcer. Call them the Strong Man and the Nosy Grandmother. Each occupies a social role with its own kind of friction, its own targets, its own costs. When both roles are present and active, the system mostly corrects itself. When one or both withdraw, the system tips rather than slowly degrades.

The Strong Man enforces by physical presence. He is the guy in line who turns around when the volume rises behind him, the bodega owner who comes out from behind the counter, the cluster of dads at the playground who walk over without speaking when something is off. His enforcement tool is the implicit threat of escalation, which mostly does not need to be used. Most antisocial behavior in public is opportunistic and probes for friction. A body that turns toward the probe, with the wrong kind of stillness, ends the probe.

The Nosy Grandmother enforces by social legibility. Her tool is reputation inside a dense local network where reputation still costs something. The teenager smoking in the stairwell is not afraid of her physically. He is afraid that she knows his mother.

The two roles are complementary because they cover different cases. The Strong Man handles strangers and the genuinely dangerous, the moments where physical presence is the only message that registers. The Nosy Grandmother handles members of the in-group, especially the early-stage drift of kids who haven't yet committed to anything. The first works on people whose stake in local reputation is zero. The second works on people whose stake is high. Between them, they cover most of what a healthy block needs covered.

Sociologists have circled this for decades. Jane Jacobs had her version, the "eyes on the street," where safety was a byproduct of mixed-use density and shopkeepers who knew their block. Robert Sampson and his collaborators later turned a related intuition into a measurable construct, collective efficacy, which captures how willing neighbors are to intervene in low-level disorder. In their Chicago data, it predicts violent crime rates better than poverty or racial composition.

The broken-windows literature has been argued about for forty years, but the part that holds up best is the soft-policing piece, where the operative variable is residents who already act as if the window matters. The original framing got distorted into a story about police, partly because police are easier to fund and study than the social fabric they sit on top of. State capacity is legible. Informal enforcement is hard to count.

Taking the soft layer seriously changes the question. The one worth asking is what conditions produce a block where the Strong Man turns around and the Nosy Grandmother opens the window. Some combination of density, tenure, overlapping networks, and a shared sense of who counts as a member. All four have been eroding.

The ideological piece deserves its own beat, because it is the part most often elided. Over the last twenty years, an enormous amount of cultural production has gone into delegitimizing the Strong Man specifically. The guy who turns around on the train is, in one prevalent reading, a vigilante, a busybody, a person who should mind his business, possibly a racist, in the worst case a future Bernhard Goetz or Daniel Penny. None of these readings is crazy on its face. Strong Men get things wrong, sometimes catastrophically.

Loose enforcement and zero enforcement are very different equilibria, and the gap between them is where most of the trouble lives.

The cultural script went from "use judgment" to "do nothing" without an intermediate stop. The Nosy Grandmother got hit by a related but distinct critique, which cast her as nosy in the pejorative sense, an enforcer of stifling norms, the antagonist of every coming-of-age novel about escaping the small town. The critique has a grain of truth. Communities with strong informal enforcement can be suffocating, cruel to deviants, hostile to the kid who doesn't fit. The alternative implicitly proposed, though, was zero enforcement rather than loose enforcement.

When the soft layer withdraws, the state is what's left, and the state is structurally bad at this work. The Strong Man on the train is a continuous, low-cost sensor with discretion. A police officer is a discrete, expensive response that has to be dispatched. By the time the situation has escalated enough to justify dispatching one, the cheap intervention window has already closed, and you are in the regime of arrests and citations and the slow theater of municipal response.

The encampment is the canonical example. Nobody plans an encampment. An encampment is what you get after eighteen months of nobody saying anything to the first guy with the tent, then another six months of nobody saying anything to the second, and then a phase change that's hard to reverse, because by then it's a small society with its own internal rules.

San Francisco spent most of 2024 trying to undo this with what the mayor called aggressive sweeps. The operation required interdepartmental coordination among police, public works, and homeless outreach teams, plus an enabling Supreme Court ruling. That is the cost of restoring by force what used to be maintained by glance.

Being precise about what the soft layer does matters here. Soft policing keeps the cost of correction near zero at the early stages of drift. A glance from someone who clocks what's happening, or a window that opens. These cost the enforcer almost nothing, and most of them succeed, because most antisocial behavior is testing the room. Deviance gets calibrated against what the room will tolerate, in real time, by people who have noticed the rules slipping and are not yet committed to anything worse.

The cleanest contemporary illustration is online moderation. A good Discord server runs on mods who are present, recognizable, and willing to apply small frictions early. A bad Discord server has absentee mods and chaos, or absentee mods until things get bad enough to require mass bans, which always feel disproportionate because they are. The platforms that have tried to outsource moderation entirely to automated systems run into the same problem cities do. Algorithmic moderation, like state intervention, can only act at the level of the discrete violation, never at the level of the developing vibe.

The vibe is the actual thing being policed, and the soft layer is the only mechanism that can read it at the right resolution. Some of the most orderly public spaces in American cities right now are privately owned. The lobby of a fancy hotel works because the hotel hires dedicated staff to play the Strong Man role explicitly, on payroll. The doorman is a Strong Man with a job description.

This works, but it scales badly, and it produces the strange contemporary geography in which the only places that feel orderly are the ones you have to pay to enter. The membrane around an Equinox or a members club is doing the work the membrane around a neighborhood used to do. You can think of the proliferation of paid third places as a kind of privatization of soft policing, sometimes deliberate, often accidental, almost always more expensive in aggregate than the thing it replaced.

The roles were also gendered and demographically narrow in ways that have since eroded.

The Strong Man role used to be played mostly by men of a particular sort, older and settled and embedded, and the cultural delegitimation of that figure has been thorough. The Nosy Grandmother role was played disproportionately by women who were home during the day, which is itself a vanishing demographic. Both roles were tied to specific life arrangements (single-earner households, long tenure, neighborhood-scale employment) that have eroded for reasons largely unrelated to enforcement. You can lament that or celebrate it. The enforcement consequences are downstream either way.

The more useful framing, I think, is to treat soft policing as a piece of social infrastructure rather than a personality trait. Infrastructure can be designed. It can also be neglected. A block where everyone is a renter on a one-year lease, working remotely, ordering groceries to the door, with no kids in the local school, is a block whose soft-policing infrastructure has been quietly disassembled. The residents may be perfectly nice people. They simply lack the structural conditions under which they would ever turn around on the train.

Which suggests, somewhat counterintuitively, that the policy levers most likely to restore the soft layer are not about enforcement at all. They are about the conditions of legibility and tenure that make enforcement come from neighbors rather than uniforms. Long leases. Owner-occupants. Local employers. Schools where parents from the same block know each other. Membership institutions, religious or otherwise, that produce dense and overlapping ties. These are the conditions under which a Nosy Grandmother is even possible, because they are the conditions under which she knows whose mother is whose.

The Strong Man and Nosy Grandmother analysis is descriptive, not normative. Both figures can be wonderful and both can be terrible. The same Strong Man who deters a mugging can also harass a kid for being the wrong color on the wrong block. The fact that the system fails when they withdraw doesn't mean the system worked perfectly when they were present.

Honest accounting requires noting that the historical baseline was uneven, that certain populations got more enforcement and others less, and that some of the contemporary discomfort with the Strong Man figure is downstream of completely reasonable observation about how that role was abused. But honest accounting also requires noting that having no soft layer is a worse equilibrium than having an imperfect one, and that the current strategy of substituting state capacity for neighborly intervention is producing outcomes nobody actually wants, including the people who advocated for it.

The design question that follows from this framing is how to build social forms that produce soft policing as a side effect, without requiring people to consciously sign up for the job. The historical examples were almost all accidents. Density, walkability, mixed-income housing, shared schools, religious congregations, sports leagues, the local Italian-American social club whose existence nobody could quite explain. Nobody designed these to produce informal enforcement. They produced it anyway, because they produced the underlying ties.

It is genuinely unclear whether any contemporary social form can produce the same byproduct on purpose. The run clubs and group chats and Discord servers and weekly dinners that have sprung up to fill the third-place void are doing meaningful social work, but they are mostly portable and elective and demographically thin. They don't have membranes around physical territory. A run club doesn't really care if someone is yelling on the corner two blocks from where it meets, because the corner is not its corner.

The forms that historically did produce territorial enforcement were ones where membership and geography were the same thing, the parish or the block association or the volunteer fire company. They were inefficient, locally exclusionary, embarrassing in their politics, and they produced safe streets anyway.

Replacing them with anything that doesn't have geographic skin in the game produces the outcomes we are currently producing. Cities are now looking for ways to manufacture, through programs and budgets and ordinances, what used to come for free from people who happened to live near each other and care. The candidates on the table (community ambassadors, neighborhood safety teams, business improvement districts paying for cleaning and patrol) are all partial. They produce a thin paid version of the Strong Man without restoring the Nosy Grandmother at all.

I don't have an answer to the obvious next question.

The question is whether any of the new geographically-anchored experiments (network state proposals, intentional residential communities, the small subset of remote-work towns that have managed to convert transients into actual neighbors) will produce informal enforcement at meaningful scale. Probably not, because most of them select for the demographic least likely to play either role. But the design space is at least open, in a way it hasn't been since the parish stopped being the default unit of urban life.

In the meantime, the cities visibly working on the problem are mostly doing it through state capacity, because state capacity is what they have. San Francisco's coordinated sweeps are one example, along with the various civic ambassador programs in midsize American downtowns. These interventions do meaningful work and will continue to be necessary as long as the soft layer remains thin. They will also remain expensive, brittle, politically contested, and chronically behind the curve, because they are operating downstream of the layer where the problem actually lives.

The unsexy implication is that civic order is mostly a function of the social ecology underneath it, and that the policy debate about enforcement has been having the wrong argument for about a generation. The argument that matters is between cities with working soft layers and cities without them. Once you've lost the soft layer, every available option will look bad, because every available option is operating on the wrong substrate.

Which leaves the open design problem of how to grow the substrate back, on purpose, in conditions that didn't produce it the first time. Harder, probably, than any of the current contenders acknowledge. The historical instances emerged from the texture of daily life in ways that may not be reverse-engineerable from a memo.