How the Rising City Index works

1 · The question

The grade answers one question: is this city building the future, or falling behind? It rates momentum, not wealth. Since v1.1 ten of the eleven scored measures describe change over a period, so a city that starts low can score high and one that stands still can score low. The one exception is access to public transport, a level, because no source yet measures its change for every city. A separate line, “where it stands” (2020–24 data), reports the starting level so the two are never confused.

Each measure had to pass four tests: (1) a primary, public source; (2) the same definition for all cities; (3) something a city can influence; (4) a clear direction of improvement.

Since v0.7 the grade is absolute: every measure is scored against a fixed benchmark, not against the other cities. A city’s score depends only on its own data, so adding cities to the set never moves an existing grade, and grades rise if every city improves.

2 · Eleven scored measures (and two for context), six forces, fixed benchmarks

Each of the six forces progrss tracks is scored from one to three measures. Each measure is scored from 0 to 100 against its benchmark. Measures marked context are shown on every card and as ranking tabs, but do not count in the grade: a level, a measure that covers under 80% of cities, or a period too distorted to score.

Every value is labelled Fact (counted by its source) or Estimate (modelled or derived); most measures here are Estimates. Built-up land growth is shown on each Planetary pressure tab as context, not scored.

3 · Scoring

1. Measure score. score = 100 × (value − floor) / (benchmark − floor), capped at 0 and 100. Changes are turned into a yearly rate (compound) or a per-decade amount first, so the lines stay valid when a later edition uses a different period.
2. Force score. The average of the force’s measure scores. A force is scored only with at least half of its measures (1 of 2, 2 of 3); a force resting on one of two measures is marked on the card.
3. Grade score. The average of the six force scores, out of 100. Each force counts equally. A city is graded with at least 5 of 6 forces; 1,029 of 1,032 are graded today (89 rest on 5 forces; a missing force is left out, not scored as zero). Ife, Nsukka and Ondo have no grade: with no 2020 broadband figure they keep only 4 forces. Since v0.9 the top two bands, AAA and AA, need all six forces scored, so a gap can never lift a city to the top: Hanchuan, Hengshui, Kaifeng and Putian (5 forces each) are held at A.
4. Band. Fixed thresholds, never curved to the results: AAA 85+ · AA 75+ · A 65+ · BBB 55+ · BB 45+ · B 35+ · CCC 25+ · CC 15+ · C below 15. Today no city reaches AAA; 3 reach AA (Suqian 77.1, Shangrao 75.8, Jinhua 75.4), all in China.
5. Breadth cap. A grade can sit at most two bands above the band of the city’s third-weakest force, so one or two strengths cannot carry a city. It lowers 27 grades today, among them Buenos Aires, Munich, Hiroshima and São Paulo.
6. The lines are part of the method. A benchmark changes only with a new method version and a note here, never to fit the results.

4 · Standing and light trend

Where it stands (2020–24 data) is a level, kept apart from the grade. Five levels are each scored 0–100 against a fixed line and averaged: human development 2020 (0 = 0.55, the UN line for low development; 100 = 0.95), fine-particle pollution 2021–23 (0 = 35 µg/m³, WHO interim target 1; 100 = 5 µg/m³, the WHO guideline), internet speed 2023 (0 = 10 Mbps, 100 = 250 Mbps, log scale), residents near public transport (0–100%), and research works per million residents 2024 (0 = 100, 100 = 10,000, log scale; this line has no outside anchor). 65 or more = high base, 35 to 65 = the middle, under 35 = low base. Since v0.8 this is absolute too: only a city’s own numbers move its label.

Light trend compares night-time light growth before and after 2022, using the only measure with several recent years for every city (2015, 2021, 2022, 2023, 2025). With mid = mean of 2021–23: early = (mid / 2015)1/7 − 1   late = (2025 / mid)1/3 − 1. “↑” when late − early is above +2 points a year, “↓” below −2, “→” in between. The card says in words whether light is growing or falling in each period. It describes two past periods; it is not a forecast.

5 · Robustness

We tested how much the result depends on our choices (script: tools/cities/check/rob7.py). The order of cities is stable; letters near a band line are not.

• Drop any one scored measure: Spearman rank correlation with the published scores 0.87 to 0.99; between 179 and 519 cities change grade, almost always by one band (at most 15 move two). The largest moves come from measures that are a force’s only one (transit access, speed growth). Without night-time light, 30 cities would fall below the 5-force minimum; without sun and wind, 16.
• Drop any one force: rank correlation 0.86 to 0.96; between 304 and 519 cities change grade, mostly by one band.
• Breadth cap: without it, 27 cities change; with a stricter one-band cap, 226 would.
• Near the line: 227 of 1,029 graded cities score within one point of a band line. Read their letter as “around”, not exact.
• Checks: the cities gate recomputes every measure score, force score, grade and cap from the inputs and the benchmark table on every build. The first 50 cities were also recomputed by hand by a separate reviewer; the 99 added on 5 November 2026 were sample-checked against their sources (air traffic for 12 cities, and spot checks of every other input; none wrong); so were the next four batches (90 to 160 values each recomputed, none wrong).

6 · Known limits

• Benchmarks are choices. Each line is anchored where possible (world pace for sun and wind, UN targets for transit access and compact growth, WHO for air) and explained in the table; greenness has no external anchor. Moving a line moves scores, which is why lines change only with a new version.
• Different areas and periods. Measures use the area their source publishes (urban centres, 25, 30 and 100 km circles, regions, airports, transit networks) and periods from 2010–20 to 2015–25; human development and income stop at 2020.
• Gaps. Air traffic growth covers 416 cities and transit opened 137, so both are context. Missing measures are left out of their force, not scored as zero. Electrifying has two measures since v1.0 (how much sun and wind was added, and how clean the new power is); Rewiring and Leapfrogging rest on one scored measure, and more are in preparation.
• Internet speed growth uses crowd-sourced tests (Ookla). 19 cities score 0 because their measured median speed fell from 2020 to 2023, most of them in Egypt; that may be sampling, not a real decline.
• Sun and wind near zero across much of the Global South. Against the world’s average pace, cities with little utility-scale solar and wind nearby score close to 0; plants near a city are not the power it uses, and rooftop solar is not counted. The clean share (v1.0) counts only plants built since 2015: a city surrounded by old coal but building only solar scores high, and the source sometimes lists one plant twice (near Mexico City it appears to count one solar farm twice).
• Population growth is counted twice in part: it scores in Demographic turn and is the denominator of compact growth, so fast-growing cities gain on one and can lose on the other.
• Gas flares light up the Niger Delta (Port Harcourt, Aba, Warri): night-time light there partly measures oil and gas activity.
• Night-time light is a proxy for activity, not GDP; LED street lighting dims rich cities in the satellite record, and conflict or blackouts cut light (Khartoum: war since April 2023), and recovery adds it back (Mosul’s 2015 baseline was under ISIS control).
• Research growth partly reflects indexing: more local journals entering OpenAlex inflates growth (Jakarta ×10, Baghdad ×8; Surabaya ×14, Medan ×23), which is why its line is set high. Iraq’s indexed output rose about ninefold nationally in 2015–24, so every scored Iraqi city carries a caution on its card. The 99 added cities come from an OpenAlex snapshot one day later than the first 50 (counts drift about 1–2% a day, in both years alike).
• Two sources for transit access: UN-Habitat’s official SDG 11.2.1 city figures come first (450 cities, mostly 2020; the nearest year where 2020 is not published). Where UN-Habitat publishes nothing, NASA SEDAC’s estimate of the same indicator is used (496 cities); it runs lower in the Global South. SEDAC values under 1% are left out as not plausible. Some figures are reported by national governments rather than modelled (for example Pakistan’s) and are marked on the card.
• City centres: circles are drawn around the UN’s population-weighted centre point. A few sit away from downtown (Surabaya, Cape Town and Conakry about 15–18 km); San Francisco (offshore) and Durban (35 km inland) are held back.
• Income per person is modelled (gridded from national and regional accounts) and its totals are stored to three significant figures, about ±0.1 point a year.
• Boundaries: in GHSL, Shenzhen shares one urban centre with Guangzhou, so several of its figures describe the wider Pearl River Delta.
• No causality. The grade describes direction of change. It does not say why a city moved or whether policy caused it.
• Licences: Global Energy Monitor, UN-Habitat and Ookla data are used for free publication; commercial licences are still to be confirmed, and nothing from them is sold.

7 · Living data: the progrss vault

The grade is not a one-off report. It is computed from the progrss vault, our own database of primary-source data, and is rebuilt as a new, checked edition when its sources publish something new.

Count from the vault, refreshed every hour · last update 5 Nov 2026, 23:41 UTC.

• Primary only. Every value is read from the publisher’s own file, API or satellite product (NASA, the EU’s Joint Research Centre, UN agencies, the World Bank, energy trackers, transit operators), never from a press article or a copy. Each one is stored with its publisher, edition, link, licence and a Fact or Estimate label.
• Always fed. Scheduled jobs reload the vault’s automated sources and sync it every hour; others (satellite tiles, operator reports) are read when a new release appears. A weekly scan flags new patterns for review.
• The grade moves with the data. When a source publishes a new release (a new year of night-time light, a new energy tracker, a new count of research papers) we load it, recompute every score and grade against the same benchmarks, re-run the independent check, and publish a new edition. Cities can rise or fall between editions; that is the point.
• As current as the sources allow. Each measure uses the latest year its publisher has released: night-time light, air traffic, transit and clean power run to 2025, research to 2024. Some official series lag, for example human development and income stop at 2020 and air quality at 2023, and every card shows the year so the lag is never hidden.
• Nothing silently rewritten. Each edition is dated and versioned. Old grades stay on record, so a city’s path can be followed edition to edition.

8 · Rules and reproducibility

• Independence: no city, company or government can pay for, submit data to, or review its grade before publication. Licensing the data does not change it.
• Primary sources only: every figure is read from the publisher’s own file or page, stored with its source link and data year.
• Versioned: this is method v1.1 (November 2026); the v1.0 grades are archived. Any change to measures, benchmarks or thresholds gets a new version number and a note here; past grades are not silently rewritten.
• v1.1 (5 Nov 2026, night): the grade now rates change, not level. (1) Rewiring scores internet speed growth 2020–23, counted from at least 15 Mbps; the speed reached moves to context. (2) Income per person moves to context (2015–20 ends in the COVID year; 40% of cities scored 0). (3) Lines raised where too many cities maxed a measure: night-time light 100 = +10% a year (was +7.2%), research output 100 = ×8 a decade (was ×4) and scored only from 100 works in 2015, compact growth 100 = 0.35 (was 0.5). Every scored change measure now has at most 20% of cities at 100, except the clean share of new power (22%, where 100 means all new power is clean). (4) Iraqi research growth carries a caution on the card. Effect against v1.0: AA falls from 10 cities to 3; Chinese cities in the top 100 fall from 89 to 68; 184 cities rise and 326 fall, 23 of them by two bands or more; Ife, Nsukka and Ondo lose their grade for want of 2020 broadband data. Median score change: Africa +3.3, Latin America +3.8, Middle East −1.3, Europe −1.7, Asia −4.3, North America −8.7. The v1.0 cards are kept on record.
• v1.0 (5 Nov 2026, evening): Electrifying gets a second measure, the clean share of new power: of everything built within 100 km in 2015–25, how much is solar, wind, hydro, nuclear or geothermal rather than coal, oil or gas (Global Energy Monitor; scored for 875 of 1,030 cities that built at least 100 MW). One measure for a whole force was too thin, and size alone rewarded cities building lots of everything. 229 cities change grade (185 up, 44 down); two move two bands (Udaipur CCC → BB, Zurich B → BBB). Up: cities whose new power is mostly clean, such as Addis Ababa and Bengaluru (BB → BBB) and Delhi (B → BB). Down: cities building mostly coal or gas. No city reaches AAA. New power in the Middle East and North Africa is a median 1% clean (mostly gas); in Europe, 84%. Independently checked. The v0.9 cards are kept on record.
• v0.9 (5 Nov 2026): four changes after the index grew to 348 cities, where Chinese cities held every top-20 place and many maxed several measures. (0) Coverage rule: a measure is scored only when it covers at least 80% of graded cities; air traffic growth (172 of 348) and metro and rapid bus opened (39 of 348) move to context: shown on the city card and as ranking tabs, not scored, until bulk sources pass the bar. (1) AAA and AA need all six forces scored. (2) Three lines that were too easy to max out were raised: cleaner air (−8% a year, or at the WHO guideline of 5 µg/m³), greener built-up land (+0.10 a decade), internet speed (500 Mbps). (3) Income per person and human development are marked on each city card as regional (provincial) figures. 107 cities change grade (12 up, 95 down), almost all by one band; no city reaches AAA (Putian and Hefei were AAA); Cairo B → CCC (35.7 → 34.4). 19 of the top 20 are still Chinese: Chinese cities lead on 10 of the 13 measures (cleaner air, human development, income, sun and wind, greenness, internet speed among them), so no single line explains it; it reflects how fast they changed in 2015–25. The first city outside China is Dubai, 12th. The v0.8 cards are kept on record.
• v0.8 (4 Nov 2026): “where it stands today” now uses fixed lines instead of ranking the 50 cities against each other (thirds). Grades and scores are unchanged; 13 labels move (Delhi, Kolkata, Nairobi, Dar es Salaam, Johannesburg and Addis Ababa from low base to the middle; Moscow, São Paulo, Buenos Aires, Istanbul, Bogotá, Shenzhen and Shanghai from the middle to high base). Cairo stays low base (34). The v0.7 cards are kept on record.
• v0.7 (4 Nov 2026): scores are now absolute. Every measure is scored 0–100 against a fixed benchmark instead of a percentile among the 50 cities; the two halves (momentum, people) are replaced by six force scores of two or three measures each; the grade is their average, capped at two bands above the third-weakest force. New measures: income per person (GHSL), internet speed reached (replacing speed growth), transit per million residents (replacing total km); population growth and human development join Demographic turn. Each force ranking on /cities is now that force’s score. Grades moved for most cities; Cairo stays B (35.7). The v0.6 grades are kept on record.
• Data update (4 Nov 2026, late): Delhi added to metro and rapid bus opened (188.1 km, from DMRC’s own annual reports: network end-2014 vs 2024-25, plus the Noida Aqua Line); 37 cities now. No grade changes.
• Data update (4 Nov 2026, method v0.6 unchanged): metro and rapid bus opened now covers 36 cities (adds Jakarta, Mumbai, Seoul, Lahore, Dhaka, Ho Chi Minh City, New York, Bogotá, Baghdad and Tokyo), independently checked. Three cities change grade by one band: Jakarta and Berlin BB → BBB, Bogotá BB → B.
• v0.6 (4 Nov 2026): cleaner air now runs from 2013–15 to 2021–23 (three-year means) using Washington University’s ACAG satellite PM2.5 V6.GL.02.04 within 25 km of the city centre, instead of GHSL 2015 → 2020; “where it stands today” uses the 2021–23 PM2.5 mean. Seven cities change grade by one band.
• Data update (5 Nov 2026, night, method v0.9 unchanged): transit access now uses UN-Habitat’s official SDG 11.2.1 figure wherever one is published (UN SDG Global Database), as the method always intended: 355 cities move from NASA SEDAC’s estimate to the official figure (typically about 9 points higher) and 48 cities get a value for the first time, so 946 of 1,030 cities now have one. 132 cities change grade (108 up, 24 down; Campo Grande, Juiz de Fora, La Serena and Manaus move two bands up), 897 → 945 cities now have all six forces, and Handan and Yichang reach AA. Independently checked.
• Cleanup and top-up (5 Nov 2026, evening): 1,030 of the world’s largest cities. 41 cities with thin data are held back: each rested on 5 forces and missed at least one more scored measure (Goma, Kandahar, Cabinda, Nara and Xianyang among them). 52 cities added (Hamilton to Resistencia, about 0.4 million); 8 held under the same rule (including Bida, whose fixed-internet speed is recorded as 0 because no tests exist). Fixes to live cities: people within 100 km for Brighton, Gaya, Heyuan, Leipzig, Meizhou, Portsmouth and Shiyan (earlier runs had failed), and 2025 night light for Belo Horizonte, Hanoi, Kabul, Kazan, Lusaka and Tashkent; their scores move accordingly.
• Data update (5 Nov 2026, night): metro opened 2015–2025 now covers 74 cities: 35 Chinese cities added from the China Association of Metros’ own reports (end-2025 network, minus end-2015, plus lines it dates to 2015), independently checked. 17 are floors (extensions of older lines in 2015 are not published), rapid bus is not counted for them, and Wuhan is a floor by about 30 km (the reports’ line dates conflict). Still context, not scored: it covers under 80% of cities.
• Rule (6 Nov 2026, Mo): the grade rests only on measures that one official worldwide dataset covers for nearly every city in a single pull. Air traffic and metro opened have no such source, so they stay on every card as extra facts, filled from official bulk files as they arrive, and are no longer pushed toward the 80% coverage bar. Bulk sources are checked by rebuilding a random sample of 30 cities from the publisher’s own file; hand-collected numbers are still checked one by one.
• Metro and rapid bus opened: 63 more cities (6 Nov 2026). Line by line from operators, transit authorities and governments (e.g. HMRL, CMRL, PIB, Sound Transit, Metroselskabet, SITEUR, MRT Corp), each line independently re-read; transit opened now covers 137 cities. Street-median light rail on its own track counts (as for Los Angeles); trams in mixed traffic, commuter and regional rail do not. Many values are floors (sections with no published length are not counted, e.g. Bengaluru 2015–17, Kuala Lumpur LRT extensions, Ottawa Lines 2 and 4). Left out: Houston (only planned lengths published), Monterrey (no primary opening record), Daegu (Line 3 length from press only); Taipei counts only the Circular line (the Airport MRT runs mostly in Taoyuan). Context only; no grade moved.
• 44 more cities with air data (6 Nov 2026). Official airport files from Canada (Statistics Canada), Mexico (OMA and ASUR annual reports), Peru (OSITRAN), Chile, Morocco (ONDA), Tanzania, the Philippines (CAAP), Indonesia (Denpasar, BPS) and South Korea, each independently re-read; air now covers 416 cities. Morocco’s 2025 figures are published rounded to 0.1 million. Held back: Kampala (file could not be re-opened), Huancayo (2025 cut by runway works), Toronto (its second airport is not published) and five Indonesian and Philippine cities whose source tables disagree with themselves. Air stays context only; no grade moved.
• San Francisco, Durban and 48 airports added. San Francisco and Durban return with corrected centres: the population-weighted centre of their dense built-up area, because the UN point sat offshore (San Francisco) or 29 km inland (Durban). Durban’s centre lands in Westville, about 9 km west of downtown; its 30 km circle still covers downtown and the main townships. San Francisco’s circle measures cover the city itself; its urban population, GDP and transit access describe the wider Bay Area. San Francisco’s clean share counts only power built 2015–25, not its older gas fleet. Air data now covers 48 more cities (372 with both years), from official files in Japan, Türkiye, Argentina, Mexico, the UK, Colombia, South Africa and the US. Rosario (2025 is part-year), Kocaeli (near-zero traffic) and Tijuana (the Cross Border Xpress terminal opened in December 2015) are left out. Air stays context only. No existing grade moved.
• 197 more cities added (5 Nov 2026, later): 1,019 of the world’s largest cities, from Davanagere (0.56 million) to Bhilwara (0.43 million), under method v0.9; no existing grade moved. Thirteen held back: two UN cities with under 10 km² of built-up land, nine whose centre point sits over 25 km from their institutions, and Kikwit and Gusau (data for only 4 forces). Night light is withheld for Ramadi and Butembo (their 2015 values are near zero, which makes growth meaningless). Cities are not the 1,019 largest: about 360 larger UN cities are left out as UN boundary artefacts, suburbs inside a bigger city, war zones or cities with poor data.
• 94 more cities added (5 Nov 2026, later): 822 cities, from Yuxi (0.63 million) to Qarshi (0.56 million), under method v0.9; no existing grade moved. Sixteen held back: fifteen UN cities with under 10 km² of built-up land for 0.4 million people or more, and Seeb (centre point 27 km from its institutions). Suining’s transit-access value is withheld: two SEDAC centres share the name.
• 89 more cities added (5 Nov 2026, later): 728 cities, from Suqian (0.73 million) to Bordeaux (0.63 million), under method v0.9; no existing grade moved. Eleven held back: six UN cities with under 10 km² of built-up land for 0.4 million people or more (Bazhong, Dazhou, Jalpaiguri, Kamina, Mymensingh, Okara) and five whose centre point sits over 25 km from their institutions (Louhajang, Gulin, Fangshan, Longgang, Rasoi).
• 97 more cities added (5 Nov 2026, later): 639 cities, from Tirana (0.83 million) to Al Ahsa (0.73 million), under method v0.9; no existing grade moved. Held back: Al Mushahidah and Zakho (UN cities of about 0.8 million on under 10 km² of built-up land) and Taxila (centre point 27 km from its institutions). Goma’s 2025 night-light figure (27 times its 2023 value) is withheld pending review.
• 95 more cities added (5 Nov 2026, later): 542 cities, from Edmonton and Tainan (about 1.0 million) to Campo Grande (0.83 million), under method v0.9; no existing grade moved. Held back: Huadu and Bakhtiyarpur (centre points far from their institutions), Kavar, Kharameh and Kotwali (UN cities of about 0.8–1.0 million on only 1–3 km² of built-up land). Research growth is left out where a city had no indexed works in 2015 (Firozabad, Linfen).
• 99 more cities added (5 Nov 2026, later): 447 cities, from Salem (1.3 million) to Linyi (1.0 million), under method v0.9; no existing grade moved. Tshikapa is held back (data for only 4 forces). Nara’s research is left out (its institutions are mostly Osaka’s). City web addresses now spell accented letters plainly (Ürümqi → urumqi, Đà Nẵng → da-nang).
• 100 more cities added (5 Nov 2026, later): 348 cities, from Hohhot (2.0 million) to Mysuru (1.3 million). Same method v0.8; no existing grade moved. Neighbouring UN cities stay separate (Mesa and Phoenix, Vasai-Virar and Mumbai, Durg and Raipur, Johor Bahru and Singapore, Tijuana and San Diego); where they would share a value (an airport, a transit-access area) it is left out rather than counted twice. A replacement airport counts in its own year (Rajkot, Qingdao).
• 99 more cities added (5 Nov 2026, later): 248 cities, the next largest from Osaka (13.0 million) to Nashik (2.0 million); cities are now added 100 at a time, largest first, towards 1,000. Same method v0.8; no existing grade moved except Accra’s score (47.3, still BB), now that its 2025 night-light figure has arrived. Durban is held back: its UN centre point lies about 35 km inland. Chinese cities lead the new batch (Hefei AAA; eight AA), driven by night light, research, sun and wind, cleaner air and internet speed.
• 99 cities added (5 Nov 2026): the index grows from 50 to 149 cities, every city above 1 million people from a list balanced by region (Africa, Asia, Latin America, the Middle East, Europe and the US). Same method v0.8: scores are absolute, so no existing city’s grade, score or label moved. Global datasets are pulled by the vault for every city; air traffic from national tables (Eurostat, US BTS, Brazil ANAC, India AAI, China CAAC) and operators; transit opened not yet researched for the added cities.
• Data update (5 Nov 2026, later): transit opened covers 39 cities: Istanbul (a floor, from Metro İstanbul’s annual reports) and Karachi (an estimate; government and transit-authority pages) added, independently checked. Air traffic growth covers 31 cities: Bogotá added (Colombia’s civil aviation authority, Aerocivil: its 2015 airport bulletin and its 2025 per-flight open data), independently checked.
• Data update (5 Nov 2026): air traffic growth now covers 30 cities, independently checked: Seoul (Korea Airports Corporation, incl. transfer passengers), São Paulo and Rio de Janeiro (computed from Brazil’s ANAC flight-leg data, the same file for both years), Dar es Salaam (Tanzania’s statistics office, thousands; 2025 provisional). Two cities change grade: Seoul BB → B; Rio de Janeiro B → CCC (its airports carried 9% fewer passengers than in 2015).
• Data update (4 Nov 2026, night): air traffic growth now covers 26 cities, independently checked. Added from each country’s own airport or aviation authority, both years from the same publisher: New York (Port Authority, revised 2015 figures, Stewart and Atlantic City left out), Sydney (BITRE, scheduled services), Delhi, Mumbai, Bengaluru and Kolkata (Airports Authority of India, monthly reports summed to calendar years), Manila (MIAA), Johannesburg (ACSA, Lanseria not included), Lima (Fraport) and Buenos Aires (ANAC; 2025 published in thousands). Wikidata was used only to find each official file. Three cities change grade: Bengaluru BB → BBB; Mumbai BB → B; Johannesburg B → CCC. Seoul is held: its source could not be re-opened by the checker.
• v0.6 data update (4 Nov 2026): 2015 airport figures read for Berlin (Tegel and Schönefeld, from the airport company), Chicago (city aviation department) and Los Angeles (Los Angeles World Airports, same five airports as 2025); air traffic growth now covers 16 cities, up from 13. Bangkok’s 2015 figure is read, but it waits for a calendar-2025 total. Method unchanged; grades recomputed.
• v0.5 (4 Nov 2026): one measure per force, the same in the ranking and the grade. Rewiring: air traffic growth 2015–2025 (was internet speed growth 2020–23 in the grade and 2025 passengers in the ranking). Electrifying: solar and wind added 2015–2025 per resident within 100 km (was per resident of the dense core in the grade, and operating capacity in the ranking). Leapfrogging: metro and rapid bus km opened 2015–2025 (was network km per million in the grade and network km in the ranking). Grades moved; Cairo stays B. v0.4 (first edition) is kept on record.
• Corrections: if you find an error, write to [email protected]. Confirmed errors are fixed and logged with the version notes.
• Data: per-measure values and scores for every city are on each city card; the full dataset can be licensed.