Chapter Twenty-Two ended on a question: whether commitment can land on the people who take it up, rather than remaining in the place everyone sees and no one is responsible for. In the Chengjiang records, what catches commitment is always concrete people in concrete positions — the care groups, the sweepers of the banyan plaza, the four group admins of the mini-program. This book has repeatedly observed their attrition, and just as repeatedly the regeneration of cooperative capacity: the crisis network of Chapter Seventeen, the plaza of Chapter Nineteen, the taking-back mechanism of Chapter Twenty-Two, each re-established locally. Placing these records side by side reveals one and the same unaddressed question: the practices that catch others — what do they themselves stand on. Treating "the local" as the cure and "institutions" as the root of the disease is one and the same laziness; what this chapter asks about is the relation of support between the two scales — what local practice needs the larger scale to provide, and what it can offer that the larger scale cannot.
Suppose that after the mini-program disturbance subsided, the city established a fictional "Neighborhood Support Program": venue fee waivers, activity liability insurance, care and first-aid training, and an emergency referral hotline, open to application by care networks, mutual-aid groups, and venue self-management organizations. This chapter uses three years of fictional records of this program as a thought experiment; the persons, institutions, and data correspond to no actual policy or research finding. The object of analysis is not whether some particular scheme is right or wrong, but the mechanisms between scales: how support actually happens, and under what conditions it deforms into withdrawal, incorporation, and bundling.
The Mutual Dependence of the Local and the Larger Scale
The care groups need electricity and servers; the ground of the banyan plaza belongs to some property registry; the conduct of the volunteers falls under some set of liability rules. "Purely local spontaneous mutual aid" is a narrative that folds dependence away — it treats electricity, venues, insurance, and permits as background and calls only the part where people deal with one another the community. The first step of the analysis is therefore the same as Chapter Eighteen's method of unfolding "cultural depth" back into labor: unfold "the local" back into a string of concrete conditions, and then look at who supplies these conditions.
In everyday language "community" and "institution" stand opposed, as if one side were made of personal feeling and the other of documents. The analysis of this book has by now dismantled this opposition: Chapter Eight showed that professional services can be fairer than acquaintances; Chapter Seventeen showed that intermediary organizations absorb, platforms rearrange, families close off. What we call the larger scale is not some colossus but another set of bearing structures — rules of rights, insurance pools, cross-district knowledge, professional backup. It can do what the local cannot, and cannot do what the local can.
This chapter does not ask which scale is the better one. It asks: does an arrangement make the local capacity to take on stronger or weaker, and does it make the responsibility of the larger scale clearer or blurrier. By this measure, both "devolution" and "recentralization" can mean the same thing — the weakening of capacity; and both can mean repair. Scale itself guarantees nothing for either side.
What Local Practice Offers That the Larger Scale Cannot
The medicine-delivery records from Qiaonan in Chapter Eight supplied the contrast: the platform knew the prescription had lapsed; Auntie Chen discovered the anomaly from two days without going downstairs. Specific knowledge is produced by repeated contact; it attaches to familiarity with this person, this rhythm, and cannot be fully standardized — because the purpose of standardization is precisely to reduce dependence on repeated contact. The larger scale can collect cases, but it cannot replace this knowledge on site. This is the first irreplaceable of local practice.
The storm records of Chapter Seventeen showed that what first caught people were the building-group chats and the greengrocer's credit, not the command center convened ad hoc. Trust accumulates slowly through everyday performance and cannot be procured on the spot in a crisis; and when institutions at the moment of crisis draw on this stock, they often decline to acknowledge that it exists. The second irreplaceable of local practice is already-existing trust and its speed of response — Chapter Twenty-Two showed that this speed can be consumed by the illusion of visibility, but it itself can still only be produced locally.
Chapter One pointed out that any entrance's statistics take the already-accepted as their measure; the people pushed back so many times that they stopped reporting are not in the frame; and the request that thirty-eight people saw in Chapter Twenty-Two was finally caught by a social worker working offline. The third irreplaceable of local relations is a line of sight that covers those who do not enter procedures — not because the local is warmer of heart, but because it has repeated contact with these people's lives. The report has no column for the silent; this is a gap that no statistics at any scale can fill.
Only with the counterexamples written in full does the list avoid becoming a hymn. Chapter Twenty showed that mutual aid absorbs fluctuation but not crises of simultaneous tightening; Chapter Eight showed that familiarity also misjudges, that Auntie Chen might read a family visit as an anomaly; the signaling costs of Chapter One and the acquaintances' gaze of Chapter Nineteen show that local networks will exclude newcomers and surveil their own members. Each of the three irreplaceables of local capacity carries its own mode of failure along with it; treating them as a universal cure is precisely the starting point of the first deformation to be dealt with below.
Which Larger-Scale Conditions Local Practice Needs
The first year of the record should be written honestly, with the program's real gains. Once liability insurance was in place, two care groups dared for the first time to take on elders with severe impairment — until then the risk had been silently borne by the coordinators in person; the training gave three long-term volunteers first-aid certification; over the winter the hotline took in eleven cases that the local level had explicitly reported it could not hold; the venue waivers gave the banyan plaza's self-management arrangement formal standing for the first time. These gains are not withdrawn on account of the critique that follows, just as Chapter Eight did not withdraw the real value of convenience: support can genuinely happen, and precisely because of this, its deformations deserve to be seen one by one. The four classes of conditions that follow are precisely the gaps that the first year's gains pointed toward.
Chapter Eighteen established that "praise cannot replace accounting"; here we move the point over to the institutional side. Venues must be paid for, coordination takes time, and when something goes wrong someone must bear the responsibility — these costs recur continuously. A local practice without stable funding does not lack costs; the costs are merely subsidized by the silence of those who bear them: the plaza is swept in passing by two elders, the group admin's reviews are done late at night. The function of the funding condition is not to reward warm-heartedness but to stop treating warm-heartedness as a budget.
Liability for accidents in volunteer activity, venue permits, and care disputes need to be specified, within concrete institutions, as borne by whom. Unclear liability can place unforeseeable burdens on individuals and can reduce the willingness to keep participating, but it does not follow from this that local practice has only two possible states, "gray operation" and "personal liability." For different matters the risk may be carried by an organization, a contract, insurance, or a public institution, and whether each applies needs to be judged separately. This chapter's requirement is that the conditions of responsibility and support be specified before a task is taken on: goodwill cannot substitute for the distribution of risk, nor can the claim that one dispute necessarily leads to wholesale withdrawal be treated as an already-proven law.
Care skills, first-aid procedures, methods of mediation can be trained, and should be transmitted. The larger scale happens to hold precisely this verifiable knowledge. The shape of the transmission determines the consequence: knowledge that enters as tools teaches people how to do things and leaves the judgment of whom to help, and how far, to the scene; knowledge that enters as hierarchy demotes the local to the terminal end of execution. Chapter Seventeen analyzed the mechanism by which professional language isolates the public, and its inverse holds here just the same — every step of knowledge's descent either carries hierarchy or does not, and this is the fork between support and takeover within one and the same training course.
The knowledge condition has a further layer that is often left out: the experience of handling exceptions is scattered everywhere, and a method that one locality's care network worked out by trial, another locality often starts again from zero on. The larger scale can do exactly what the local cannot — compile the scattered experience into verifiable methods and send it back, being itself not a headquarters but an exchange layer. In Chapter Seventeen's records, the peer network discovered shortages of multiple medications before the hierarchy did, which shows that horizontal transmission is faster than vertical aggregation; the function of the exchange layer is to turn this occasional horizontality into institutionally available access, without compiling the methods into commands.
Any locality's capacity is limited. The key scale condition is not making the local bigger but making failure escalatable: a channel for referral, a position where professionals take over, reinforcement for a time. The existence of backup is in itself protection for the local — a practice without backup can only express its limits through refusal or through exhaustion, and both of these will be read as problems of character. Chapter Eight argued that the interface must allow responsibility to return; here the same logic climbs one level higher: what the local cannot hold must have somewhere to go.
Who Keeps Doing the Work After the Funding Withdraws
What devolution conceals most easily is the devolution of responsibility. Some task previously borne by public funds — visiting those who live alone, maintaining venues, emergency contact — is relabeled "community self-management," the budget disappears along with it, and the talk is of respecting the grassroots and stimulating vitality. The mechanism is a move in the bookkeeping: the same task is shifted from the public ledger onto an invisible ledger; the beneficiaries are the tightened budget and the institution freed from accountability; the bearer is the person expected to catch it. This process requires no one's malice; a decision-maker sincerely convinced that community is better can complete it.
Chapter Eighteen dealt with the community conscripting its own members' time in the name of tradition — the costs pressed onto those inside the community with the fewest options. What is dealt with here is the other side: the institution shedding public responsibility in the name of the local — the costs pressed onto the local level as a whole. The two can also compound, as when an agency initiates a festival while the labor falls on the community; the cost transfers discussed in Chapter Eighteen thereby enter into scale relations. What this chapter supplies is the other half of the account: when responsibility leaves the budget sheet, who disappears from the budget sheet. The two mechanisms are adjacent but distinct: one asks who pays inside the community, the other asks who exits on the public side.
To judge whether a "handing over to the community" is a return of authority or a withdrawal, look not at the talk but at what is handed over. When responsibility is transferred, are funding, legal standing, training, and backup simultaneously available? A transfer with the conditions synchronized may be a genuine return of authority — there are indeed matters better decided by the local itself, and Chapter Seventeen's analysis of the overconcentration of decision-making still holds. A transfer of responsibility without the transfer of conditions is withdrawal, however respectful its wording toward the local. This threshold depends on no conjecture about motives; the account itself is the evidence.
"Society is resilient" can become a circular argument: the community caught it, which proves the withdrawal right; the community failed to catch it, which proves the community still needs tempering — any outcome whatsoever guarantees the cut. The end of Chapter Seventeen already stated this danger plainly: post-crisis social resilience may be nothing but private attrition, with public institutions still handing care and verification over to unpaid relations, only the language of the reports having changed. The mature form of the resilience narrative renames exhaustion as strength.
The counterexample must be kept: some transfers really do expand the range of options of those involved — the locality gains decision-making power it did not previously have, and those who bear the work gain rotation and resources. The criterion is the same one Chapter Eighteen used in the dispute between renewal and preservation: whether the change increases or diminishes the range of options of those involved. When after a transfer the locality's capacity to say "yes" and "no" grows stronger, it is a return of authority; when it weakens while the tasks still get done, it is withdrawal — that the work still gets done is precisely why withdrawal is the hardest to detect.
Local Practice Treated as Free Infrastructure
The second severe case begins with an emergency response plan. The plan wrote the volunteer network into the response sequence; the activation during some crisis succeeded, and from then on activation became the default — no posts, no pay, only the name of "partnership." Each step of the slide is reasonable: this time truly was urgent, and they truly do know the ground. Looking back after three years, a considerable portion of emergency response has become standing duty resting on unpaid labor, and this standing duty appears in no staffing table whatsoever.
This kind of incorporation need not seize the locality's decision-making power and can even hold the local in the highest respect — it need only decline to compensate the local's labor. When a gap in public services is filled with volunteer hours, the wage of a formal position converts into a certificate of appreciation. The record of the four group admins' three hundred nights in Chapter Twenty-Two repeats here at institutional scale: the platform expresses thanks sincerely, the agency cooperates sincerely, and the account stands all the same. Chapter Eighteen's phrase, "recognition cannot replace compensation," applies to the institutional side word for word.
When the only payment is honor and coverage, the incorporation goes one layer deeper. The commended volunteer acquires a public image of goodwill, and this image raises the shame price of refusal — when the next request arrives, saying "not this time" no longer costs a private discomfort but the betrayal of a public image. Visibility could be recognition; as a substitute for compensation, visibility becomes a locking device. The way to identify it is the same as in the previous section: observe whether the commendations always appear where bearing is demanded, or where conditions are given.
Condition Bundling in Funding
The program's second year introduced assessment: service headcounts, activity sessions, presentable outcomes. Indicators are a management tool, but the consequence is a transfer of the power of definition — what counts as success is decided by the funder's categories, and local practice begins to reshape itself around the measurable part. Chapter Twenty-Two already showed how ratings crowd out help that does not get photographed; here the same mechanism moves from the platform to funding: the unmeasurable core of a care network — trust, the recognition of exceptions, relations that survive broken promises — is precisely the part for which the report has no columns.
Project-based funding carries a further cost that is often left out of the accounting: the writing labor of applications, documentation, audits, and final reports. These hours are borne entirely by the local, and borne unevenly — organizations that can write outcompete organizations that can care, writing ability thereby becomes a condition of survival, and the practices most in need of support and least skilled at paperwork are screened out in advance. Chapter Eighteen's analysis of archival labor is isomorphic here: the output is used by everyone while the process is not counted; when the next project begins, the experience starts again from zero with the loss of personnel.
Chapter Seventeen pointed out that resource dependence changes an organization's agenda; here the mechanism is unfolded: the funder need not prohibit any topic at all — it need only make some topics easy to approve and some topics hard to get through review. After a number of years, the organization's topic composition has been shaped by the profile of the funding, while the organization still believes itself to be choosing of its own accord — this is the deepest form of condition bundling: it alters no single decision, only the cost table of decisions.
Repair does not rely on asking funders for leniency. There are four workable directions: the base part unconditional — detailed below; project conditions tied to the matter and negotiable, rather than packaged as a whole; outcomes portable, accumulating no proprietary data or credit for any single funder; exit with a transition, the end of funding not meaning the cutoff of practice. None of the four fixes amounts; what they change is structure: they lower the price of saying no.
The Boundary of Standards and Support
The first three deformations all criticized the larger scale; now the positive side of the boundary is written in full. The legitimacy of local difference extends exactly as far as the rights baseline. If "respecting local character" means that somewhere may withhold the basic protections owed to every person, then this is not pluralism but laissez-faire — multiple scales do not mean multiple layers of exemption. The larger scale's least alienable function is to set a floor below which no locality may fall. This does not conflict with the preceding critique: what was criticized was never the standards themselves, but standards leaving the position of the baseline in order to take over the method.
The distinction thereby becomes the key: a standard that writes "must not fall below" is a baseline, and what it protects is the weakest; a standard that writes "must be done thus" is a method, and what it concentrates is decision-making power. Baselines can be unified at a nationwide scale; methods should be left to the local — because methods depend precisely on the specific knowledge discussed above. The failure of funding condition bundling is in essence a baseline misinstalled as a method: prescribing by unified category the practice of success, rather than drawing the floor of failure.
There is a second reason the baseline is needed: the local excludes from within. The signaling costs of Chapter One and the acquaintances' gaze of Chapter Nineteen both show that handing everything over to the local means handing those whom the local excludes over to the very networks that exclude them. External verifiability is therefore necessary. But verification is not takeover: those who are excluded need an external channel of complaint that does not pass through the local network, and like practices in different localities need to be able to check against one another horizontally — the cross-group comparison proposed in Chapter Seventeen is here a third shape, between "takeover from above" and "no verification at all."
Assigning to the local what suits unification — baselines, entitlements, basic protection — creates local exemption; drawing up to the center what suits the local — the recognition of exceptions, the choice of methods — creates rules detached from the scene. The two mismatches are symmetrical errors, and they supply each other with justifications: the experience of excessive control from above argues for wholesale devolution, and rumors of local abuse argue for wholesale recentralization. The analysis refuses the pull of both directions and returns to the functional list: baselines by standard, methods left to the local.
The Shape of Support Without Capture
Support should be layered. Venues, base funding, legal standing, emergency backup belong to the unconditional layer — they are the infrastructure that makes local practice possible, and they should not be bundled with any project goal; a care network need not perform outcomes in order to obtain liability insurance. Specific projects may carry conditions, and the conditions must be relevant to the matter: funding a training may require what is to be taught, but should not also require reporting who attended. The unconditional layer protects the local's persistence; the conditional layer constrains the matter, not the organization.
A genuine return of authority does more than give money; it also lets the local decide how to use it. Funding that gives money without decision power is still management, merely converting the cost of management into paperwork; conversely, granting decision power without conditions wraps withdrawal in the packaging of a return of authority. To judge the quality of a support arrangement, look at whether resources, conditions, and decision power move together — all three in step is support; only responsibility moving is withdrawal; only money moving while decision power stays put is purchase.
Whether backup is real is written not in the plan but in the record. The test is that at least one escalation has actually been caught: what the local referred up was done, and afterward the local's role was acknowledged rather than confiscated — the matter was taken over, and the practice of referral itself was not replaced. A backup channel through which no escalation has ever passed has the same standing as a fire extinguisher never used; and a channel in which every escalation ends in takeover teaches the local never to use it.
How Local Experience Returns to Institutions
Accepting support does not mean accepting takeover, and this boundary is maintained by both sides together. A locality should be able to refuse a grant without losing the unconditional layer — having declined this year's project, it should still be able to obtain insurance and venue next year. This is at the same time a structural constraint on funders: if refusal means being out, the so-called voluntary application is obedience renewed by the month.
If support flows only toward "capable leaders," it will intensify rather than relieve concentration inside the local — the administrator's visibility power in Chapter Seventeen, the "cannot do without her" position in Chapter Eighteen, will each be amplified by one more layer through targeted funding. Support for an organization should therefore include conditions for the organization's own health: rotation of the coordinating roles, the possibility of handover, entry paths for newcomers. This is not interfering in internal affairs; it is declining to turn public resources into supply for an internal oligarchy.
A local practice that receives public funds should be verifiable: where the money went, who is bearing what, what the outcomes are. Verification is at the same time constrained by Chapter Twenty-Two's record minimization — each record must name the specific function it serves, duplication prevention, auditability, dispute tracing, and a record that cannot name its function is merely raw material for profiling. The line between verifiable and abusable is the same line that the platform records use.
A Visibility Audit of Scale Relations
Gather the mechanisms into verifiable questions. For any relation between a locality and a larger scale: when responsibility was transferred, did the conditions and the decision power follow; for whom are the indicators defining success; on whom is the labor of application and reporting booked, and who subsidizes it; has the backup channel actually taken one over; does the standard write a baseline or a method; and when the local said "no" once, what did it lose. The six questions give no total score; they only make the trade-offs between scales available for discussion — the same method as Chapter Eighteen's labor audit, Chapter Nineteen's questions about venues, Chapter Twenty-Two's questions about the platform.
The audit presupposes neither that the local nor that institutions are the more trustworthy. It may show that some task should indeed be taken up, that some task should indeed be devolved, and that some task should not exist at all. Reading the audit's result as "the local is simply good" or "the professional is simply good" is to refold the list into a stance — precisely the kind of folding that the chapters have criticized.
If a local practice cannot continue, its ending likewise requires a handover: who takes over the unfinished care, how the already-committed funds are handled, whether the participants can take part in the arrangement. Budget proportions are decided by concrete institutions; this chapter's requirement is that changes in resources and in responsibility must appear together.
Conditions of Uptake and the Boundary of Support
Institutional withdrawal appears in the guise of "returning authority to the local"; condition bundling appears in the guise of "professional support" — corresponding one to one with Chapter Eighteen's warm conscription and progress freeze, Chapter Nineteen's order discipline and vitality displacement: the one contracts in the name of respect, the other defines in the name of help. Identifying them relies not on conjectures about motive but on tracking which way the conditions flow — whether funding, legal standing, decision power, and backup move together with responsibility, and whether the indicators write baselines or methods. The more moving the talk, the more necessary these questions become.
Gather the conditions of uptake from the preceding chapters. Conditions transfer together with responsibility — transferring responsibility without conditions is judged withdrawal; support is layered, with an unconditional base layer and project conditions relevant to the matter; decision power moves downward together with resources; standards draw baselines and do not prescribe methods; backup is real and has passed at least one test; the labor of application and reporting is counted into cost; visibility's recognition does not replace compensation; the local's refusal triggers no punishment; support for an organization includes conditions for its internal rotation. These conditions fix no amounts, proportions, or schedules; they are the yardstick for judging whether any given scale relation has crossed its boundary.
The three questions at the end of Chapter Twenty-Two — whether people can stay, whether they can try once more, whether commitment can land on people — must, once the local has caught them, be asked once again between scales: when one locality cannot hold, does escalation have somewhere to go; after escalation, is the local taken over or confiscated; when the larger scale exits, does it admit the withdrawal or pretend that the local is still holding. The next chapter no longer draws up an institutional inventory of common persistence, but lets these questions enter one day in the streets and lanes of Chengjiang, to see how six lives interrupt one another and still keep meeting. The closing here is a single sentence: a community's capacity to persist lies not in handing everything over to the highest scale or to the nearest neighbors, but in every handover transferring the conditions at the same time; whether the relation of support between the local and the larger scale holds is finally visible in whether the most ordinary request for help can be caught in the place nearest to it, taken over when it cannot be caught, and whether that nearest place is still there after the taking over.