Sustainable Megacities

A megacity is conventionally a metropolitan agglomeration above ten million people. The definition is administrative convenience rather than measurement, and this brief argues that the convenience has done real damage: it has attached a policy programme, a research literature and a great deal of infrastructure finance to a unit whose boundary nobody can specify and whose share of human urban experience is much smaller than its share of attention.

The subject splits into four questions that are usually run together. What is a city, physically — urban form, density, and what density statistics actually measure when the denominator is an administrative choice. Where most urban people actually live — in informal settlements, and whether those are a housing failure to be cleared or a housing system to be upgraded from within. What a city can pay for — own-source revenue, borrowing capacity, and the tenor mismatch between an asset with a fifty-year life and a municipality that can borrow for seven. Who decides — the gap between the functional city, which is a commuting catchment, and the administrative city, which is a boundary drawn for other reasons and is almost always smaller and almost always plural.

Scope boundary. Instrumentation, sensing and municipal algorithms belong to smart cities, and appear here only where the evidence bears on governance quality. Dwelling design and construction belong to future housing systems. Intercity and intra-city rail belong to high-speed transit networks. Bulk water belongs to water infrastructure megaprojects. Purpose-built vertical settlement belongs to arcologies, underground cities and floating cities. This brief is about whether a very large city, as an object of government, can be made to work at the scale at which it actually functions.

Sourcing note, stated at the top because it shapes what follows. The research pack behind this brief could not reach UN DESA, UN-Habitat, the OECD or the World Bank. World Urbanization Prospects projections, informal-settlement population counts, metropolitan fragmentation counts and municipal creditworthiness data are therefore declared gaps in this brief, not figures written from memory. Each is named where it belongs, with what would settle it. The argument does not depend on any of them.

Established The functional extent of a city is an artefact of the travel-time threshold you choose, and the choice changes the answer by a factor of three. Cattaneo and colleagues delineated city-regions worldwide in Nature Cities in 2024, classifying 30,079 urban centres across 213 countries and territories into four tiers — town (20,000 to 50,000), small city (50,000 to 250,000), intermediate city (250,000 to one million) and large city (above one million) — and mapping their catchments by travel time. Established At a three-hour cutoff the world contains 1,403 primary city-regions. At a one-hour cutoff it contains 4,210. Established The authors describe the one-hour figure as the more indicative of commuting times. Established Nothing about the world changed between those two numbers; a single methodological parameter did. Frontier The consequence is blunt and it applies to this brief’s own title: any statement about how many metropolitan areas exist, or about the functional extent of a megacity, that does not name its threshold is not a measurement. It is a convention wearing the clothes of one.

Established The megacity is salient, not modal. The same study found that 3.2 billion people can physically reach multiple tiers of urban centre within one hour, and 4.7 billion within three; that 92 per cent of the global population lives within one hour of an urban centre of some size; and — the finding that should reorganise the subject — that among people living in or nearest to towns and small cities, twice as many have easier access to intermediate cities than to large cities. Established The authors state plainly that this underscores the crucial role of intermediate cities. Frontier Read against a policy literature, a finance pipeline and a research field organised around the largest agglomerations, this is a direct empirical challenge to the framing rather than a footnote to it. Speculative If urban attention and urban capital tracked population exposure rather than city size, the modal object of urban policy would be a city of a few hundred thousand people in a low- or middle-income country, and almost nothing retrieved for this brief suggests that is what happens. Frontier That mismatch is a live and under-argued position, and it is the reason a brief titled “Sustainable Megacities” should open by questioning its own unit.

Established Cities are governed at a scale at which they are not measured. The Global Urban Monitoring Framework comprises 77 indicators across five domains — society, economy, culture, environment, and governance and implementation — addressing four objectives, and was endorsed by the UN Statistical Commission in March 2022. Established Moorthy and colleagues report in npj Urban Sustainability that fewer than 20 per cent of all indicators are, on average, reported at the city level; most are reported nationally. Established The same paper finds that citizen science could contribute to 52 of the 77 indicators, about 68 per cent. Frontier A framework built to govern cities whose evidence base is more than four-fifths national is a measurement constraint that has quietly become an institutional one: you cannot hold an authority to account at a scale for which no numbers are published. Established That paper also carries, second-hand and without a base year, the familiar figures that urban areas cover three per cent of the terrestrial surface while accommodating 55 per cent of the world’s population, projected to reach 68 per cent by 2050. Frontier This brief quotes them only as what that paper carries, because the two halves of that sentence do not use the same definition of urban and because the primary source could not be obtained — see section 14.

Speculative The two numbers this subject most needs are the two this brief does not have, and saying so is part of the finding. Frontier On metropolitan fragmentation, the research pack could verify that the OECD publishes a figure whose title reads “Metropolitan areas can include hundreds of municipalities” and a second titled “Administrative fragmentation of metropolitan areas, 2014”, but could not retrieve the underlying values; oecd.org was unreachable. Established Note what the published title itself asserts: hundreds, not a number. Frontier On municipal finance, not one borrowing-capacity or credit-rating figure could be retrieved — every World Bank DOI attempted returned a 403 at the resolver — so this brief prints no creditworthiness number for any city. Speculative What survives is structural and does not need the missing cells: a metropolitan economy is one labour market, and it is administered by a plurality of jurisdictions each of which has a veto over revenue-sharing and a constituency-level reason to use it. Frontier The claim that municipal creditworthiness rather than planning capacity is the binding constraint on megacity infrastructure is widely asserted, plausible on its face, and — in the evidence assembled here — unadjudicated in either direction. It is stated as an open question, not resolved by assertion.

Established City-scale government does demonstrably achieve things, inside constraints it did not choose. Simpson and colleagues examined Cape Town’s Integrated Development Plan for 2022 to 2027 and found 15 of 49 programmes designated Climate Priority Programmes, in a city of approximately 4.7 million where youth unemployment among 15- to 24-year-olds reached 47.3 per cent in 2020 and approximately 32,400 people lost jobs during the 2015–2018 drought. Established They describe the plan as the primary basis on which the public hold local government accountable. Established In Bogotá, Álvarez-Rivadulla and colleagues studied the District Care System introduced in 2020, which delivers services at Care Blocks sited in underprivileged areas; ethnographic work found the policy gave marginalised women access to public institutions often for the first time and generated unanticipated caregiver support networks, while not resolving the underlying burden of unremunerated care. Established Raven and colleagues synthesised eight databases covering close to two thousand urban experiments into ten lessons across process, politics and impact. Frontier The capability is real. Established Its limits are also measured: Yazar, Haarstad and Elfving found that fewer than half of C40 cities with climate action plans emphasise procedural justice, with equity-pledge signatories, later joiners and Global South cities showing stronger integration and the remainder exhibiting tokenistic participation; and Liang, Han and Day, on 58 in-depth interviews in Beijing, found NGOs strategically downplaying distributional inequities in order to keep operating.

Established The physical record at megacity scale is harsher than the governance record. Saito and colleagues sampled stream communities across the Metropolitan Region of São Paulo, described at source as a megacity with over 22 million inhabitants, and found biodiversity loss that was strongly non-random with respect to body mass: selective loss of very small, moderately large and large taxa compressed size ranges by four orders of magnitude relative to rainforest streams, food-web complexity collapsed from networks with over 150 trophic links to as few as three, and energy flow fell by a factor of 1,446. Established Ng and colleagues analysed land surface temperature and intra-urban heat island intensity across eleven Southeast Asian capitals from 1990 to 2024, with land-cover classification above 97 per cent accuracy, and report significant warming of 0.350 °C per year in Vientiane, 0.235 in Kuala Lumpur and 0.215 in Bangkok, alongside major urban expansion and vegetation and water loss; the authors flag their exclusion of socio-economic and governance variables as a limitation. Established Khromova and colleagues assessed four nature-based solution types across Barcelona, Boston and Rotterdam and found heat risk the most spatially autocorrelated hazard, green roofs and permeable pavements consistently feasible in dense heat-prone neighbourhoods, and rain gardens and urban parks constrained by land availability and underground infrastructure. Frontier That last constraint is worth pausing on: some of the binding limits on urban adaptation are literally buried, and the records of what is buried are worse than the records of what is built.

Established Instrumentation has been audited, and the audit is not flattering. Moghaddam and Cao analysed 28 implementations selected from 157 AI deployments across six domains between 2015 and 2024. Established ShotSpotter: 97 per cent acoustic accuracy yielding 9.1 per cent crime-fighting effectiveness. Established Chicago’s predictive policing list: 16.3 per cent of listed individuals confirmed as gang members, against commanders’ belief of 95 per cent. Established COMPAS: overall accuracy around 60 to 70 per cent, with a 44.9 per cent false-positive rate for Black defendants against 23.5 per cent for white defendants. Frontier High metric accuracy is compatible with near-total policy failure, and the paper’s argument is that it is sometimes the mechanism of it. Established On the predictive side, the best retrieved spatial urban-growth model — Contento and colleagues’ probabilistic hexagonal cellular automaton, calibrated on 2001, 2006 and 2011 and validated on a held-out set of 200 cells — achieves 81 per cent average accuracy on a single metropolitan area at a five-year horizon, with neighbourhood influence dominating in low-density areas and employment and income dominating in highly urbanised zones. Frontier Eighty-one per cent, one metro, five years, is the honest state of the art, and it is the ceiling against which every confident 2050 urban projection should be read.

Frontier Are megacities governable at their functional scale at all? Three lines of evidence bear on it and they do not point the same way. Against: the functional object is not stable — Cattaneo’s 1,403 against 4,210 means “governing at functional scale” has no fixed referent until somebody picks a threshold. Against: the administrative scale sits far below the functional one, and fewer than a fifth of urban indicators are published at city level, so the unit is neither jurisdictionally nor statistically visible. For: city government demonstrably delivers — Cape Town’s climate priority programmes, Bogotá’s care system, two thousand documented experiments. Speculative What would have to be true for functional-scale governance: a delineation chosen and frozen, accepting that no natural threshold exists; fiscal authority coextensive with it; statistical reporting at that unit; and democratic legitimacy at a scale that currently has no electorate. Frontier The ordering matters and it is counter-intuitive: the measurement problems are the cheap ones and are partly solved; the fiscal and legitimacy problems are neither.

Frontier Are informal settlements a solution rather than a problem? This is the brief’s most consequential open question and the honest answer is that the evidence assembled here cannot close it. The strongest located argument for the affirmative is Van Oostrum and Shafique’s case for plurality in applying building regulation to slum upgrading — regulating from within rather than replacing — joined by Acioly’s street-led citywide upgrading, which connects informal and formal city through area-based planning rather than clearance. Frontier The counterweight is equally serious: Bardhan and Pan interrogate the mass rehabilitation-housing model in India, Goytia treats the pandemic as a stress test of informal settlements, and Ngom sets out the public health burden in sub-Saharan African settlements. Speculative Declared honestly: every item in that paragraph is a bibliographic record whose text this brief could not obtain. No population number, no effect size and no direction of effect is claimed from any of them, and the impact-evaluation literature that would settle it — Jaitman’s review, Pérez Casas’s Nicaragua evaluation — sits in the same condition. Speculative For “solution” to hold, three things must be true: in-situ upgrading must beat resettlement on measured welfare; tenure security must be separable from physical upgrading, so that the cheap intervention can be delivered without the expensive one; and a settlement’s own building practices must be regulatable without being formalised out of affordability. Frontier The third is precisely the contested claim. The first is answerable and has not been answered here.

Frontier Is urban density a measurement or an artefact? Density is population divided by an area whose boundary is a choice, and the choice moves the count of city-regions threefold. Speculative The tractable fix is a single delineation applied across every compared city, which is a convention rather than a discovery, and is the single most fixable item in this whole subject. Frontier The definitional inconsistency between national criteria is the subject of a 2025 UN DESA chapter whose text this brief could not obtain and which should be the first thing a better-resourced pass retrieves.

Frontier Is urban climate adaptation land-constrained rather than finance-constrained? Khromova’s three-city result says the parks and rain gardens are stopped by land availability and buried infrastructure, while the interventions that fit — roofs and pavements — are feasible in exactly the dense hot places that need them. Speculative If that generalises, the binding constraint shifts to subsurface infrastructure records and subsurface rights: unglamorous, cheap, and almost entirely unstudied.

Speculative Should a megacity be governed as a metabolic system with enforceable interior ecological thresholds? The São Paulo result supplies, for the first time in this literature, a measurable interior state variable that is a flow rather than a stock: trophic complexity and energy throughput, down 1,446-fold. Handwave A regime that regulated a city against a floor on interior energy flow would require the metric to be computable routinely and cheaply, a legal instrument that attaches to a flow rather than to a parcel — which essentially no property system contains — and a jurisdiction covering the whole catchment. Handwave It is coherent, nobody is building it, and it is the most interesting thing in the subject.

Speculative Is care infrastructure the missing municipal utility? Bogotá’s Care Blocks are an existence proof that a genuinely new municipal function can be created and sited. Frontier Generalisation requires demonstrated fiscal sustainability and transfer outside Latin American municipal traditions, and neither has been tested. Frontier And does urban experimentation compound? Two thousand experiments in eight separate databases yielding ten distilled lessons is an enormous denominator for a small output, and the existence of eight databases rather than one is itself evidence that outcomes are not measured commensurably enough to accumulate.

Frontier The workback target is a megacity governed at its functional extent: one accountable authority whose boundary matches the daily-life catchment, which can tax, borrow and spend across it, and whose statistics are reported at that unit. Seven links, in order.

Established L1 — freeze a delineation. The object is threshold-dependent by a factor of three; the one-hour commuting-indicative cutoff is the defensible default. Frontier Binds institutionally, but cheaply: this is a convention, not a discovery, and the science is finished. Established L2 — produce statistics at that unit. Today under a fifth of urban indicators are reported at city level. Measure progress as the share of the 77 framework indicators published at the delineated functional unit; citizen science is a demonstrated partial route to 52 of them. Frontier L3 — harmonise the national definitions of urban. Also cheap. Also not done. Speculative L4 — make fiscal authority coextensive with the delineation, by amalgamation, by a metropolitan tier with independent revenue, or by binding revenue-sharing. Speculative L5 — let that tier borrow, which needs own-source revenue, a credit history and a subnational debt framework; measure it as own-source revenue share, credit rating, and access to long-tenor local-currency debt. Speculative L6 — legitimacy at the new scale, on which nothing retrieved here speaks directly. Established L7 — substantive capability, which already exists: Cape Town and Bogotá are the proof.

Frontier L4 is the binding link, and it is institutional rather than scientific, technical or even financial. L1 to L3 are conventions and statistical labour — unglamorous, tractable, undone. L5 is genuinely open and this brief has no numbers for it. L6 is a real political unknown, and scaling up a unit that already under-delivers procedural justice at its present size is not self-evidently an improvement. Speculative But the chain stops at L4, because L4 requires existing jurisdictions to surrender revenue, and nothing in the retrieved literature describes a mechanism by which that happens absent a higher tier of government imposing it. Frontier Any programme that frames megacity governance as a planning problem, a data problem or a technology problem is aiming at a link that is not binding.

Speculative One structural observation about the shape of the chain. The links get harder monotonically and they get cheaper monotonically in the opposite direction: L1 to L3 cost almost nothing and are undone, while L4 to L6 cost almost nothing in money and are undone because they cost power. Frontier A field that spends its effort on the expensive-looking end — sensing, modelling, master planning — is behaving as though the constraint were resources, when the retrieved evidence says it is authority. Speculative The workback plan is therefore unusually short and unusually unpalatable, and that combination is a reasonable diagnostic that it is pointing at the right thing.

Frontier Five results this brief waits on, none of them a physics result. First, the in-situ upgrading impact-evaluation base with effect sizes: does upgrading in place beat resettlement on health, income, security and housing deficit, and over what horizon. The literature exists and was not obtainable here. Frontier Second, municipal borrowing capacity and creditworthiness for large low- and middle-income-country cities: own-source revenue shares, rating distributions, achievable tenor. Without this the L5 link cannot even be characterised, let alone crossed. Established Third, metropolitan fragmentation counts for named functional urban areas — how many municipalities actually sit inside the Paris, São Paulo or Manila catchments. The data objects are published; the values were unreachable.

Frontier Fourth, a definitional settlement: how national criteria shape urbanisation statistics, and what a harmonised delineation does to the headline urban share. Speculative Fifth, out-of-sample transfer for spatial growth models — whether an 81-per-cent single-metro result survives being moved to another metro and another decade. Frontier Note what is not on this list: no materials science, no new sensing modality, no computational advance. Speculative Every one of the five is a study somebody could commission this year with existing methods, and the reason none has been done at the scale required is that no institution owns the question — the metropolitan unit that would commission it is the thing whose absence is the subject of the brief. Frontier A sixth dependency sits behind all of them and is rarely named: a subsurface asset register of usable quality, because the adaptation options that survive the land constraint are the ones whose feasibility depends on knowing what is already underground. Frontier This brief also depends on results produced elsewhere on this site: on future taxation models for the own-source revenue question, on future federalism for the tier-creation question, and on future capital markets for whether long-tenor local-currency municipal debt can exist where it currently does not.

Speculative Six measurements that would move this subject, ordered by how cheap they are.

Speculative 1. Freeze the threshold and publish the consequence. Adopt the one-hour cutoff as a reporting convention for a defined set of city-regions, publish the resulting boundaries, and report how many statistical series change materially when recomputed on them. Cost: near zero. It converts a definitional argument into a testable claim about how much the definition matters. Speculative 2. An indicator-coverage audit at the functional unit. For twenty named city-regions, count how many of the 77 framework indicators are published at the functional boundary, at the core-city boundary, and nationally only. The current figure — under a fifth at city level — is a global average; the distribution is unknown and would tell you where the reporting deficit actually sits.

Frontier 3. The upgrading trial that separates tenure from bricks. A settlement-level design with four arms — tenure security alone, physical upgrading alone, both, neither — measured on health, school retention, household investment and income over at least five years. Speculative This is the single most valuable experiment in the subject, because the entire “informality as solution” position rests on the claim that tenure is separable from and prior to physical upgrading, and that claim has never been isolated. Frontier It is ethically demanding and operationally hard, and matched-comparison designs may be the realistic form.

Speculative 4. A municipal debt panel. Own-source revenue share, rating, and achieved tenor for fifty large cities, published annually. Frontier 5. Out-of-sample transfer of the growth models. Take the calibrated hexagonal cellular automaton, apply it unchanged to three metros it was not fitted on, and report the accuracy drop. A large drop retires long-horizon urban projection as a planning input. Speculative 6. Routine trophic monitoring. Establish whether food-web energy flow can be estimated in an urban catchment cheaply enough for annual reporting, which is the precondition for treating a city’s ecological interior as a regulated variable rather than an object of study.

Frontier What none of these require. No new instrument, no new material, no new computational method, and in four of the six cases no new fieldwork — only the reprocessing and republication of series that already exist at a boundary that does not. Speculative The exception is the upgrading trial, which is expensive, slow and the only one that would settle a question the subject has argued about for fifty years.

Established Almost nothing on the physical side of this subject is unsolved. Metro systems, bulk water conveyance, wastewater treatment, district energy and grid supply at megacity scale are all in commercial service somewhere, at the required capacity, at known cost. Frontier What is not solved is delivering them into a fragmented polity at an affordable cost of capital, and the failure mode is financial and procedural rather than technical.

Established Where genuine engineering constraints do bind, they are unglamorous. Nature-based solutions in three studied cities were limited by land availability and underground infrastructure, not by design or intent; green roofs and permeable pavements were consistently feasible in dense heat-prone neighbourhoods while parks and rain gardens were not. Frontier The implied requirement is a subsurface asset register of a quality most cities do not possess — where the pipes, cables, culverts and foundations are, to what tolerance, with what ownership. Speculative That register is a public good, it is cheap relative to what it unlocks, and its absence silently forecloses a whole class of adaptation.

Established Heat is the load that is measurably rising: eleven Southeast Asian capitals show significant intra-urban warming over 1990 to 2024, the fastest at 0.350 °C per year. Frontier Cooling load, water demand and mortality all track it, and the engineering response — surfaces, shading, canopy, ventilation corridors — is well characterised and land-hungry, which returns the problem to land, which returns it to tenure and to who may build what where. Speculative Sensing and municipal analytics are covered in smart cities; the finding relevant here is that instrumentation is not a substitute for the institutional links, and can degrade them.

Frontier The one engineering requirement that is genuinely unmet is a construction and procurement pipeline sized to a metropolitan capital programme rather than a municipal one. Speculative A transit or trunk-water programme at functional-city extent implies sustained delivery capacity over decades, and the delivery record for programmes of that size is the subject of megaproject governance rather than of engineering. Speculative Industrialised and robotic construction methods, treated in automated construction systems, are the plausible route to the volumes implied by in-situ upgrading at scale, and they interact with the tenure constraint in an awkward way: standardised industrial output presumes a regularised parcel, which is exactly what an upgraded settlement does not have.

Frontier The closest neighbour is smart cities, and the relationship is cautionary rather than complementary. The audited record of urban AI — 97 per cent acoustic accuracy against 9.1 per cent effectiveness, a list believed 95 per cent accurate that was 16.3 per cent accurate, a risk instrument with nearly double the false-positive rate for Black defendants — is the strongest available evidence that instrumenting a city is not the same as governing it. Established The same audit records community resistance that succeeded, in Toronto and Detroit, and counter-examples that worked: Amsterdam’s algorithm register, Seoul’s participatory waste management, bias mitigation at New York City Health and Hospitals.

Frontier Future housing systems supplies the dwelling; this brief supplies the question of whether the occupant has a defensible interest in the ground under it. Frontier High-speed transit networks and water infrastructure megaprojects are the two capital programmes whose tenor most exceeds municipal borrowing capacity, which makes them the natural test cases for the finance link. Frontier Megaproject governance carries the delivery record; infrastructure resilience carries the failure modes.

Speculative Three further adjacencies are less obvious and more interesting. Biodiversity restoration shares the São Paulo result, which is simultaneously an urban measurement and an ecological one, and is the empirical bridge between the two subjects. Frontier Climate migration planning is the substitute good when a city cannot be adapted, and the intermediate-city finding suggests the destinations will more often be mid-sized than mega. Speculative And the private-city literature — developer-provided urban governance as a substitute for municipal capacity in low-capacity states — is the live alternative to the whole institutional programme this brief argues for, which is a reason to name it rather than to ignore it. Frontier It is also the honest test of this brief’s thesis: if the institutional links cannot be crossed, private urban governance is what fills the space, and it fills it without the accountability channel that the whole argument for a public metropolitan tier rests on. See also institutional design and civic technology.

Frontier Three institutional preconditions carry this subject, and none is technical. The first is a metropolitan authority matching the functional city. Today the functional unit is a commuting catchment and the administrative unit is a plurality of municipalities; the OECD publishes a figure titled “Metropolitan areas can include hundreds of municipalities”, and even that title asserts only “hundreds”. Speculative There are three known routes — amalgamation, a metropolitan tier with its own revenue, and binding revenue-sharing — and each requires existing jurisdictions to give something up. Frontier This is why the constraint is institutional and not administrative: the obstacle is not that nobody has drawn the boundary, it is that every actor inside the boundary has a veto.

Speculative The second is borrowing at infrastructure tenor. A metro line, a trunk sewer or a treatment plant has a life measured in decades; a municipality that can only borrow short is forced either to refinance repeatedly at whatever rate prevails, or to fund from current revenue, or not to build. Frontier The preconditions are stated in the municipal-finance literature — own-source revenue, a credit history, a regulatory framework for subnational debt, and the intermediation that turns those into a market — but this brief could obtain none of that literature’s content and therefore states no capacity figure for any city. Speculative What can be said without the data is that this is a tenor problem before it is a cost-of-capital problem, and that the two are routinely conflated in the policy discussion.

Frontier The third is land tenure reform that permits upgrading in place. If the occupant of an informal dwelling has no interest a court will recognise, three things follow: the household will not invest in the structure, the utility cannot lawfully connect it, and the municipality’s cheapest option is clearance. Speculative The frontier proposal is regulatory pluralism — applying building regulation from within the settlement rather than importing a formal code the settlement cannot afford to satisfy. Frontier Whether that works is genuinely unsettled, and this brief has records rather than findings on it. Speculative What is not unsettled is the direction of the constraint: nothing else in the upgrading toolkit functions without a recognisable interest in the ground.

Established The distributional facts in this subject are measured and they are unflattering. Fewer than half of C40 cities with climate action plans emphasise procedural justice, and cities without it exhibit tokenistic modes of participation. Established Cities that signed the equity pledge, that joined later, and that sit in the Global South show stronger procedural-justice integration — which is the opposite of the assumption most Northern readers bring to the question. Frontier Civil society under constrained governance adapts by trading claims: in Beijing, NGOs frame distributive equity as less urgent and strategically downplay distributional inequities in order to retain procedural access.

Frontier That trade is the ethical crux of the whole subject. A metropolitan tier large enough to finance infrastructure is also further from the neighbourhood that bears the construction, the displacement and the tariff. Speculative Scaling up a unit that already under-delivers procedural justice at its current size is not obviously an improvement, and a brief arguing for the metropolitan authority is obliged to say so. Speculative What would make it defensible is an accountability channel that does not depend on the complainant’s continued operating licence — which is exactly what the Beijing finding shows is missing.

Established On instrumentation, the numbers are the argument: a 44.9 per cent false-positive rate for Black defendants against 23.5 per cent for white defendants, inside a system with respectable overall accuracy, is a demonstration that aggregate performance conceals distributed harm. Frontier On clearance versus upgrading, the ethical asymmetry is stark and under-argued: clearance destroys a functioning social and economic network that took decades to form, and the evidence base that would justify it against in-situ upgrading is precisely the evidence base nobody has assembled at scale. Speculative Absent that evidence, the burden of proof should sit with clearance, and in most cities it does not. Frontier That inversion is not an oversight: clearance produces a legible parcel, a financeable asset and a photograph, and upgrading in place produces none of the three.

Frontier If the intermediate-city finding is right, the century’s settlement question is not what the megacity becomes but what four thousand mid-sized cities become. Twice as many people near towns and small cities have easier access to intermediate cities than to large ones, and 92 per cent of humanity lives within an hour of an urban centre of some size. Speculative A civilisation that concentrates its urban capital, its research attention and its institutional invention on the top of the size distribution is optimising the tail.

Speculative The megacity nonetheless remains the sharpest test of whether large-scale collective provision is possible at all, because it is where the fragmentation, the fiscal mismatch and the legitimacy gap are simultaneously worst. Frontier If a jurisdiction cannot be constructed that matches a twenty-million-person labour market, that is a general statement about the size of polity humans can build deliberately, not a fact about cities. Handwave The optimistic reading is that this is a solved problem historically — national states were assembled out of exactly this kind of fragmentation, repeatedly, and usually by a higher tier imposing it. The pessimistic reading is the same sentence.

Handwave At the far end, the São Paulo measurement suggests a different civilizational frame entirely: a city as a metabolic object with an interior state that can be quantified and, in principle, held above a floor. Handwave A civilisation that governed its cities that way would be doing something no polity has attempted, and the obstacle is not measurement but the absence of any legal instrument that attaches to a flow. Speculative The ordinary version of the same idea is already half-built — catchment authorities, air-quality zones, waste and water balances — and what it lacks is a single accountable holder of the balance sheet, which is the metropolitan authority again, arriving from a different direction. See planetary stewardship for the same problem at global scale, and industrial ecology for the accounting method it would need.

These horizons track institutional and fiscal instruments, not construction. The construction is available now.

• 10 yr: Frontier A harmonised delineation convention becomes normal in comparative urban statistics, because it is cheap and the sensitivity result is now published and hard to ignore. Frontier The share of the 77 monitoring indicators reported at city level rises from under a fifth, slowly, with citizen science filling part of the gap in cities that have organised volunteers and not in cities that have not. Speculative Two or three more metropolitan coordination bodies are created with planning functions and no independent revenue, which is the form that is politically achievable and the form that does not cross the binding link. Frontier At least one large city-region publishes a proper subsurface asset register and discovers it unlocks adaptation nobody had costed.
• 25 yr: Speculative If the fiscal link is ever crossed, it happens in this window and it happens because a national government imposes it, not because municipalities agree. Speculative The plausible first instances are unitary states with strong central authority and a single dominant metropolitan economy, where the higher tier already has the power and only needs the motive. Speculative Long-tenor local-currency municipal debt becomes available to a second tier of cities beyond the handful that already have it; the constraint that binds is domestic institutional investors with matching liabilities, not foreign appetite. Frontier The in-situ upgrading question is answered empirically by then, one way or the other, because the trial is not expensive and the political demand for the answer is rising.
• 50 yr: Speculative Either a functional-scale metropolitan authority exists in several countries and is unremarkable, or the answer is settled negatively and cities are governed permanently below their functional scale by coordination rather than jurisdiction. Speculative This brief’s judgement is that the second is more likely, because the veto structure is stable and the coordinating form is good enough to relieve the pressure without solving the problem. Handwave The interesting minority outcome is that the intermediate-city finding wins the argument and urban institutional invention moves down the size distribution, where fragmentation is milder and the whole problem is easier.
• 100 / 250+ yr: Handwave Metropolitan jurisdictions at functional scale become the ordinary unit of subnational government, and the twentieth-century municipality reads the way the parish reads now. Handwave Cities carry enforceable interior ecological thresholds expressed as flows, which requires a property and regulatory system that does not currently exist in any legal tradition. Handwave Or the whole frame dissolves, because remote participation in urban labour markets decouples access from proximity and the commuting catchment — the object this entire brief is organised around — stops being the thing that defines a city.

• Depends on Depends on almost no unbuilt technology. The construction, water, transit and energy systems required at megacity scale are in commercial service. What it depends on instead is a set of results nobody is producing: an impact-evaluation base for in-situ upgrading with tenure and physical works varied separately; municipal borrowing-capacity and creditworthiness data for large low- and middle-income-country cities; metropolitan fragmentation counts for named functional urban areas; a definitional harmonisation of national urban criteria; and out-of-sample transfer tests for spatial growth models. It also depends on results this site treats elsewhere: revenue instruments in future taxation models, tier creation in future federalism, and long-tenor local-currency debt markets in future capital markets.
• Requires (not on this map) All three are institutional or financial, and that is the brief’s argument. The first is a market that mostly does not exist: long-tenor local-currency debt available to a city, which requires own-source revenue, a credit history and a subnational debt framework before any of the intermediation can be built. It is a tenor problem before it is a cost-of-capital problem, and the two are routinely conflated. The second is the cheapest and the most consequential: an interest in the ground that a court will recognise, without which the household will not invest, the utility cannot lawfully connect, and clearance remains the municipality’s least-cost option. The frontier proposal is regulatory pluralism — building regulation applied from within the settlement rather than imported wholesale — and whether it works is unsettled. The third is where the chain stops: a jurisdiction coextensive with the commuting catchment, achievable by amalgamation, by a metropolitan tier with independent revenue, or by binding revenue-sharing, each of which requires existing municipalities to surrender revenue. Nothing in the retrieved literature describes a mechanism by which that happens absent a higher tier imposing it. A programme aimed at planning capacity or at instrumentation is aimed at a link that is not binding.
• Enables Enables the only class of infrastructure programme that matches a metropolitan labour market rather than a fragment of it, which is the precondition for transit, trunk water and district energy at the scale at which they are economic. Enables household investment in informal housing, which is by a wide margin the largest under-used capital formation channel in urban economies, and which is unlocked by a recognisable interest in the ground rather than by a subsidy. Enables statistical visibility at the unit where outcomes actually occur, without which no accountability at that scale is possible.
• Adjacent Sits alongside smart cities, which is the substitute good most often reached for and the one with the worst audited record when it is used as a substitute rather than a complement. Sits alongside future housing systems, which supplies the dwelling this brief supplies the tenure for, and alongside high-speed transit networks and water infrastructure megaprojects, which are the two capital programmes whose asset life most exceeds municipal borrowing tenor and are therefore the natural test cases. Shares its central measurement problem with planetary stewardship: both need a legal instrument that attaches to a flow rather than to a parcel, and neither has one. Sits against the private-city literature, which is the live alternative that appears wherever municipal capacity does not.

Established “Fifty-five per cent of people live in cities, and 68 per cent will by 2050.” These figures reach this brief only as a second-hand citation inside a citizen-science methods paper, with no base year attached. They rest on national definitions of urban that differ between countries — which is the entire subject of a 2025 UN DESA chapter whose text could not be obtained here. Frontier The number is not so much wrong as not a single measurement: it is an aggregation of roughly two hundred different national definitions. Speculative Quote it with the definition attached or not at all. This brief prints no urbanisation projection of its own, and the absence is deliberate.

Established “Urban areas are only three per cent of land, so cities are efficient.” Both halves of that famous sentence were retrieved as stated, in adjacent clauses of the same paper — and they do not use the same definition of urban. Frontier The three per cent is a land-cover, built-up-area measure. The 55 per cent is an administrative, national-criteria measure. Established Dividing one by the other to obtain impact per hectare, or a global urban density, combines two different universes and produces a number about nothing. Speculative This is the cleanest example in the subject of a defect that survives because both halves are individually true.

Established “There are N metropolitan areas in the world.” There are 1,403 primary city-regions at a three-hour travel cutoff and 4,210 at a one-hour cutoff, a factor of 3.0, from a single study using a single method. Frontier Any citation of a city-region count without its threshold is meaningless, and the one-hour figure is the one the authors describe as more indicative of commuting times. Established The same defect governs density: density is population over an area whose boundary is a choice, so a ranking that mixes cities measured on their built-up core with cities measured on their metropolitan county is comparing different objects. Frontier The enthusiast form is “City X is denser than City Y, therefore density policy works”, where X was measured on its core and Y on its county.

Established “Megacities are where urbanisation is happening.” Among people living in or nearest to towns and small cities, twice as many have easier access to intermediate cities than to large ones. Frontier The megacity is salient, not modal, and a subject organised around it — including this one — is organised around the tail of a distribution. Speculative The corollary nobody draws: the institutional inventions this brief argues for would be far easier at intermediate-city scale, where fragmentation is milder and the fiscal gap smaller, and there is no obvious reason they are being attempted at the hard end first.

Established “Smart-city instrumentation improves outcomes.” ShotSpotter: 97 per cent acoustic accuracy, 9.1 per cent crime-fighting effectiveness. Chicago’s predictive list: 16.3 per cent confirmed gang members against a believed 95 per cent. Frontier High metric accuracy is compatible with near-total policy failure, and the audit’s claim is that it is sometimes the mechanism — the optimised metric substitutes for the policy objective and then defends it. Speculative The correct inference is not that instrumentation fails but that procurement and objective-setting are the binding constraint, which is again institutional. Established The same audit records counter-examples that worked, so the pessimistic generalisation is also unsupported.

Established “We can project urban populations to 2050 with confidence.” The best retrieved spatial growth model achieves 81 per cent average accuracy on a held-out set of 200 cells, on one metropolitan area, at a five-year horizon. Frontier Multi-decadal global projections are scenario exercises and should be read as such. Speculative The enthusiast form is quoting a 2050 urban population figure to three significant figures; the honest form carries the uncertainty as a first-class object.

Frontier “Informal settlements just need upgrading budgets.” The frontier argument located for this brief is about regulatory pluralism — applying building regulation from within the settlement rather than importing a formal code — not about spending level. Speculative Framing it as a budget problem presumes the formal regulatory model is correct and merely underfunded, which is the thing in dispute. Speculative The mirror-image error, common among enthusiasts for informality, is treating settlements as self-sufficient systems that only need to be left alone; the public-health and pandemic literature is the counterweight, and this brief has records of it rather than findings from it.

Established “Green infrastructure is limited by political will.” In three cities studied systematically, parks and rain gardens were constrained by land availability and underground infrastructure. Frontier Some constraints are literally buried, and a city that does not know where its own pipes are cannot plan around them. Established “Cities are the level where climate justice happens.” Fewer than half of C40 cities with climate plans emphasise procedural justice, and NGOs in at least one megacity strategically drop distributive claims to stay operational. Frontier Cities are a level at which climate justice is attempted, with a measured success rate below half, which is a different claim.

Frontier A methodological misconception worth stating plainly, because this subject is governed by machine-assembled assessment reports. The research behind this brief caught its own bibliographic instrument substituting one document for another: a Crossref query about protected-area targets returned a result set entirely about organ cryopreservation. Established The result was discarded, but the failure was silent — no error, no low-confidence signal. Speculative Three other independent research passes in the same session hit the same class of failure, including a publisher URL serving a different article than the DOI requested. Frontier This is a real fact about how citations become corrupted in assembled scholarship, and it is the reason every DOI in the reading list below should be independently resolved before being relied on. Speculative The corollary matters too: “could not locate” is not “does not exist”. A source unreachable through one gate is unverified in that pass, not fabricated.

Speculative Finally, the misconception this brief is most at risk of creating. Arguing that the binding constraint is institutional can be read as arguing that it is therefore easy — a matter of political will, a decision somebody could simply take. Frontier It is not. Amalgamating jurisdictions, creating a revenue-raising metropolitan tier and reforming land tenure are among the hardest things a state does, they redistribute power rather than money, and they have failure records at least as long as their success records. Speculative Institutional is a statement about where the constraint sits, not about how yielding it is.