Twenty full-screen data worlds · 2026-08-02

Twenty worlds hidden inside the data

Enter full-screen data worlds built from verified Our World in Data datasets. Each scene is an argument, an environment, and an instrument.

Drag the scene to look around
C for cinema, Space to play
↑ ↓ worlds, ← → acts
A for the atlas of all twenty

01SpaceThe sky is no longer empty.
02FoodWe built a second ocean.
03CommunicationThe planet skipped the wire.
04Human progressForty-three lanterns used to go dark.
05Energy accessThe planet learned to glow — unevenly.
06CitiesA city can become a country-sized organism.
07PopulationThe oldest district outgrew the nursery.
08LandWe rebuilt the surface of a planet.
09EnergyEvery energy age is still burning at once.
10TechnologyThe more we built, the cheaper sunlight became.
11ForestsThe world still loses a forest the size of a nation.
12BiodiversityThe average signal from 34,000 populations is fading.
13AtmosphereHumanity stopped tearing the shield.
14DisastersThe hazards stayed. The death rate collapsed.
15Nuclear weaponsHumanity built a mountain of endings.
16PeacePeace is not a switch. It is a network state.
17HealthA disease was pushed to the edge of the map.
18OceanThe migration routes went quiet.
19OceanFive centuries of catch ended in a cliff.
20DiscoveryThe night sky filled with worlds we cannot see.

World 01 Space

The sky is no longer empty.

What happens when launch becomes an industrial rhythm?

In 2025, 4,510 registered objects were launched into space. The orbital age has become an orbital swarm.

4,510objects launched in 2025

Since 1957: +4,508 objects
Peak: 4,510 · 2025
Depth: 13 layers · 69 steps

Act III

Access creates a new commons problem.

Object count does not tell us size, purpose, or whether an object remains active. It does show how quickly the shared orbital environment is filling.

Time

2025 · Objects launched: 4,510 · United States: 3,708 · Russia: 89 · China: 374 · United Kingdom: 10 · Japan: 30 objects

Data & limits

United Nations Office for Outer Space Affairs (2026)

1957–2025 · objects

Counts registered launched objects; not mass, function, operational status, or collision risk.

Open the OWID data page

World 02 Food

We built a second ocean.

How did farms overtake the wild sea?

Aquaculture surpassed wild capture in this global production series around 2013.

141.4Mtonnes from aquaculture · 2024

Since 1960: +139.32M metric tonnes
Peak: 141.41M · 2024
Depth: 18 layers · 65 steps

Act III

Growth moved behind nets and ponds.

Tonnes do not tell us stock health, welfare, feed inputs, pollution, or nutrition. They do reveal a structural transformation of seafood production.

Time

2024 · Wild catch: 94.4M · Aquaculture: 141.41M · China · farmed: 81.95M · India · farmed: 12.09M · Indonesia · farmed: 15.75M · Vietnam · farmed: 5.65M metric tonnes

Data & limits

FAO via World Bank (2026)

1960–2024 · metric tonnes

Production tonnage does not measure ecological sustainability, welfare, or nutritional quality.

Open the OWID data page

World 03 Communication

The planet skipped the wire.

How did billions connect without waiting for a landline?

Mobile subscriptions reached 111.5 per 100 people while internet use reached 73.6%.

73.6%people using the internet · 2025

Since 2005: +58 per 100 people / percent
Peak: 73.6 · 2025
Depth: 20 layers · 62 steps

Act III

Signal is not equal access.

Use in the last three months does not reveal speed, affordability, reliability, skills, or meaningful participation.

Time

2025 · Landline subscriptions: 9.9 · Fixed broadband: 20.11 · Mobile subscriptions: 111.5 · Internet users: 73.6 · India · online: 70 · China · online: 91.6 per 100 people / percent

Data & limits

International Telecommunication Union via World Bank (2026)

1960–2025 · per 100 people / percent

Subscriptions are not unique people and can exceed 100; usage does not measure quality or affordability.

Open the OWID data page

World 04 Human progress

Forty-three lanterns used to go dark.

How did childhood survival become normal — and where is it still denied?

The global estimate fell from 42.8 deaths per 100 births in 1800 to 3.74 in 2024.

3.74deaths before age five per 100 births · 2024

Since 1800: -39.06 deaths per 100 live births
Peak: 44.2 · 1856
Depth: 23 layers · 225 steps

Act III

A global triumph can contain local tragedy.

This combined long-run series measures outcomes. It does not assign causal credit or reveal inequality within countries.

Time

2024 · World: 3.74 · Nigeria: 11.56 · Turkey: 0.96 · Sweden: 0.24 · India: 2.66 · China: 0.57 deaths per 100 live births

Data & limits

Gapminder (2015); UN IGME (2025) — processed by Our World in Data

1800–2024 (source begins 1751) · deaths per 100 live births

Combined long-run estimates show outcomes, not individual causes or within-country inequality.

Open the OWID data page

World 05 Energy access

The planet learned to glow — unevenly.

Where did the darkness retreat, and where is it concentrated now?

657.3M people still lacked basic electricity access in 2024. Most lived in Sub-Saharan Africa.

657.3Mpeople without electricity · 2024

Since 1998: -949.03M people
Peak: 1.61B · 1998
Depth: 15 layers · 27 steps

Act III

The remaining darkness is geographically concentrated.

A connection is only a threshold. This dataset does not reveal reliability, affordability, or how much electricity a household can actually use.

Time

2024 · World · with access: 7.48B · World · without access: 657.26M · Sub-Saharan Africa (WB): 579.05M · South Asia (WB): 2.94M · East Asia and Pacific (WB): 34.4M · Latin America and Caribbean (WB): 12.94M people

Data & limits

World Bank, Sustainable Energy for All (2026)

1998–2024 · people

Basic connection does not measure reliability, affordability, quality, or consumption.

Open the OWID data page

World 06 Cities

A city can become a country-sized organism.

How did the center of the megacity world move — and where might it go next?

Jakarta held about 39.1M people in 2020 under a harmonized urban-footprint definition. The next chapter is projected, not observed.

39.1MJakarta residents · 2020

Since 1950: +45.79M people
Peak: 53.17M · 2070
Depth: 24 layers · 31 steps

Act III

The skyline becomes a scenario.

After 2020, the flight enters model projections. African megacities may dominate growth, but the further horizon is inherently uncertain.

Time

2100 · Jakarta: 49.7M · Dhaka: 55.01M · New Delhi: 31.97M · Tokyo: 24.07M · Shanghai: 24.67M · Lagos: 17.92M people

Data & limits

European Commission, Joint Research Centre (2025)

1950–2020 estimates; projections to 2100 · people

City footprints use a harmonized density definition; values after 2020 are projections.

Open the OWID data page

World 07 Population

The oldest district outgrew the nursery.

What does a world look like when its age architecture changes?

Around 2019, the global 65+ population overtook the population under five in this series.

201965+ crossed above under-five

Since 1950: +682.06M people
Peak: 808.37M · 2023
Depth: 15 layers · 74 steps

Act III

Longevity changed the skyline.

Wide age bands are not equal cohorts. This scene shows population totals, not dependency or health in old age.

Time

2023 · 65+: 808.37M · 25–64: 4B · 15–24: 1.26B · 5–14: 1.37B · 0–4: 654.03M · Asia (UN) · 65+: 469.4M people

Data & limits

United Nations, World Population Prospects (2024)

1950–2023 · people

Age bands have different widths; totals do not measure dependency or healthy ageing.

Open the OWID data page

World 08 Land

We rebuilt the surface of a planet.

When did human land use stop being a trace and become a planetary force?

Twelve thousand years of modelled land history reveal a slow transformation that accelerates sharply after 1800.

12,000 yearsof reconstructed land change

Since 10000 BCE: +19.18M square kilometres
Peak: 19.24M · 1960
Depth: 16 layers · 75 steps

Act III

The modern era compressed the transformation.

These ancient values are model reconstructions, not satellite observations. The broad scale of change is clearer than any single estimate.

Time

2017 · Urban: 2.49M · Villages: 10.27M · Cropland: 19.18M · Rangeland: 35.54M · Seminatural: 39.48M · Wild woodland: 9.74M square kilometres

Data & limits

Gauthier et al. (2021) — processed by Our World in Data

10,000 BCE–2017 · square kilometres

Ancient values are modelled reconstructions with substantial uncertainty.

Open the OWID data page

World 09 Energy

Every energy age is still burning at once.

Did new energy replace the old — or stack on top of it?

The 225-year transition is not a clean sequence of replacements. New sources mostly joined a growing system.

186,382.8TWh of primary energy · 2024

Since 1800: +55,292.08 terawatt-hours
Peak: 55,292.08 · 2024
Depth: 10 layers · 77 steps

Act III

The newest towers are growing fastest.

Primary energy uses the substitution method. The visual compares energy equivalents, not direct electricity output.

Time

2024 · Traditional biomass: 11,111 · Coal: 45,850.54 · Oil: 55,292.08 · Gas: 41,278.27 · Nuclear: 6,871.84 · Hydro: 10,860.75 terawatt-hours

Data & limits

Energy Institute (2025); Vaclav Smil (2017)

1800–2024 · terawatt-hours

Uses substitution-method primary energy and combines historical source series.

Open the OWID data page

World 10 Technology

The more we built, the cheaper sunlight became.

Can scale bend the cost curve?

Module prices fell from $128.27 to $0.258 per watt as cumulative capacity reached 1.9M MW.

−99.8%real module price since 1975

Since 1975: -128.01 US dollars per watt / MW
Peak: 128.27 · 1975
Depth: 2 layers · 50 steps

Act III

A cheap module is not a free system.

The relationship does not prove one cause, and module prices exclude installation, storage, grids, finance, and land.

Time

2024 · Module price: 0.26 · Cumulative capacity: 1.85M US dollars per watt / MW

Data & limits

IRENA (2025); Nemet (2009); Farmer and Lafond (2016)

1975–2024 · US dollars per watt / MW

Correlation with cumulative capacity is not causal proof; module price excludes total system costs.

Open the OWID data page

World 11 Forests

The world still loses a forest the size of a nation.

Can gains in one place cancel losses in another?

Global net forest loss slowed to about 5M hectares per year — still deeply negative.

−5Mnet hectares per year · 2025

Since 1991: +5.69M hectares per year
Peak: -3.24M · 2016
Depth: 16 layers · 35 steps

Act III

A plantation cannot simply replace a primary forest.

Net change can hide simultaneous natural-forest loss and planted-forest gain.

Time

2025 · World: -5M · China: 1.61M · Brazil: -3.26M · Indonesia: 81,479.69 · Democratic Republic of Congo: -282,893.75 · India: 86,000 hectares per year

Data & limits

FAO Global Forest Resources Assessment (2025)

1991–2025 · hectares per year

Net change can mask natural-forest loss with planted-forest expansion.

Open the OWID data page

World 12 Biodiversity

The average signal from 34,000 populations is fading.

What does a 73% average decline mean — and what does it not mean?

The global Living Planet Index fell from 100 in 1970 to 27.1 in 2020.

−73%average indexed population change

Since 1970: -72.87 index · 1970 = 100
Peak: 100 · 1970
Depth: 9 layers · 51 steps

Act III

An index is not a headcount.

A 73% decline does not mean 73% of all animals disappeared. Coverage is uneven and the average is sensitive to the monitored sample.

Time

2020 · World: 27.13 · Africa: 23.97 · Asia and Pacific: 39.6 · Europe and Central Asia: 64.72 · Latin America and the Caribbean: 5.38 · North America: 60.96 index · 1970 = 100

Data & limits

WWF and Zoological Society of London (2024)

1970–2020 · index · 1970 = 100

Not globally representative; an average index decline is not the share of animals or species lost.

Open the OWID data page

World 13 Atmosphere

Humanity stopped tearing the shield.

What does a global environmental success look like before the atmosphere fully heals?

Consumption of the most damaging ozone-depleting substances collapsed after the Montreal Protocol, while atmospheric recovery follows decades later.

−99%+CFC consumption since 1986 peak

Since 1986: -1.09M ODP tonnes
Peak: 1.08M · 1986
Depth: 18 layers · 34 steps

Act III

Stopping damage is not instant repair.

Consumption is an emissions proxy. Negative values can reflect destruction or trade, and long-lived gases remain aloft for decades.

Time

2021 · CFCs: -847.9 · HCFCs: 16,232.84 · Halons: -268.6 · Methyl bromide: 242.1 · Carbon tetrachloride: 1,321.3 · Methyl chloroform: -21.7 ODP tonnes

Data & limits

UN Environment Programme (2023)

1986–2021 · ODP tonnes

Consumption is an emissions proxy; negative values can reflect destruction or net exports.

Open the OWID data page

World 14 Disasters

The hazards stayed. The death rate collapsed.

How did societies become harder to kill?

Disaster deaths still spike without warning, but the century-scale decline reflects forecasting, infrastructure, food systems, and response capacity.

16,373recorded disaster deaths · 2025

Since 1900: -1.25M deaths
Peak: 3.72M · 1931
Depth: 21 layers · 126 steps

Act III

Hazards are not destiny.

Earlier records are incomplete, especially for smaller events and heat. The downward death rate is real, but the database is not a complete history of every disaster.

Time

2025 · All disasters: 16,373 · Earthquake: 6,353 · Storm: 3,641 · Flood: 5,236 · Drought: 89 · Landslide: 958 deaths

Data & limits

EM-DAT, CRED (2026)

1900–2025 · deaths

Pre-2000 records are incomplete; extreme-temperature deaths are especially undercounted.

Open the OWID data page

World 15 Nuclear weapons

Humanity built a mountain of endings.

The peak passed. What remains beneath the surface?

The estimated global arsenal peaked at 70,374 warheads in 1986. It is smaller now, but still concentrated and opaque.

12,187estimated global warheads · 2026

Since 1945: +12,181 warheads
Peak: 70,374 · 1986
Depth: 11 layers · 82 steps

Act III

Decline is not disappearance.

Stockpiles are secret and estimated. Counting warheads does not measure yield, readiness, doctrine, or the human consequences of use.

Time

2026 · World: 12,187 · Russia: 4,400 · United States: 3,700 · China: 620 · France: 290 · United Kingdom: 225 warheads

Data & limits

Federation of American Scientists (2026)

1945–2026 · warheads

Secret stockpiles are estimated; totals may include retired warheads awaiting dismantlement.

Open the OWID data page

World 16 Peace

Peace is not a switch. It is a network state.

How did hostile dyads become security communities?

The number of state relationships grew with the state system itself, while rivalry, negative peace, warm peace, and security communities formed different layers.

363security-community dyads · 2020

Since 1900: +363 country-pair relationships
Peak: 389 · 2013
Depth: 17 layers · 121 steps

Act III

Institutions can change the topology.

Absolute dyad counts rise partly because the number of states rises. Shares and categories must be read together.

Time

2020 · Severe rivalry: 27 · Lesser rivalry: 28 · Negative peace: 1,572 · Warm peace: 143 · Security community: 363 · Africa · rivalry: 3 country-pair relationships

Data & limits

Diehl et al. (2025)

1900–2020 · country-pair relationships

The number of states and possible dyads changes over time; absolute counts need share context.

Open the OWID data page

World 17 Health

A disease was pushed to the edge of the map.

Why is the last mile of eradication the hardest?

Estimated paralytic polio cases fell from hundreds of thousands in the early 1980s to 3,752 in 2023.

3,752estimated paralytic cases · 2023

Since 1980: -364,658 estimated cases
Peak: 460,159 · 1981
Depth: 15 layers · 44 steps

Act III

Near zero is not zero.

The estimate includes more than wild poliovirus. Conflict, access, trust, surveillance, and vaccine-derived transmission shape the endgame.

Time

2023 · World: 3,752 · Africa: 3,563 · Asia: 189 · Europe: 0 · North America: 0 · South America: 0 estimated cases

Data & limits

World Health Organization (2019, 2024)

1980–2023 · estimated cases

Estimated paralytic cases include more than wild-virus cases and depend on surveillance quality.

Open the OWID data page

World 18 Ocean

The migration routes went quiet.

Can an industry make giants disappear — and can restraint bring them back?

Recorded whale catch peaked at 82,194 animals in 1964. In 2024 it was 1,085.

82,194whales killed at peak · 1964

Since 1900: -566 whales killed
Peak: 82,194 · 1964
Depth: 23 layers · 125 steps

Act III

Catch fell. Recovery is slower.

Catch is not population size. Some species have recovered strongly; others remain far below historical abundance.

Time

2024 · All whales: 1,085 · Blue: 0 · Fin: 32 · Sperm: 0 · Sei: 26 · Humpback: 2 whales killed

Data & limits

Rocha et al. (2014); International Whaling Commission (2026)

1900–2024 · whales killed

Catch is not abundance; illegal and unreported historical catches may be missing.

Open the OWID data page

World 19 Ocean

Five centuries of catch ended in a cliff.

What does collapse look like when extraction outpaces a living stock?

Eastern Canadian cod catch reached 810,014 tonnes in 1968. By 2019 it was 10,559.

−98.7%catch from 1968 peak to 2019

Since 1508: +10,403 tonnes caught
Peak: 810,014 · 1968
Depth: 3 layers · 512 steps

Act III

Closing the fishery did not rewind the ocean.

Catch is not biomass, but the collapse and prolonged low catches reveal how slowly ecological systems can recover.

Time

2019 · Eastern Canada cod catch: 10,559 · 25-year average: 4,554.24 · Cumulative catch since 1508: 49.69M tonnes caught

Data & limits

Schijns et al. (2021)

1508–2019 · tonnes caught

Catch is not fish biomass; early observations are historical reconstructions. The average and cumulative layers are derived from the same reported series.

Open the OWID data page

World 20 Discovery

The night sky filled with worlds we cannot see.

How does the instrument decide what kind of universe we discover?

By 2025, the catalogue held about 6,070 confirmed exoplanets. Most were found when they crossed their star.

6,070confirmed exoplanets · 2025

Since 1992: +4,478 cumulative planets
Peak: 4,478 · 2025
Depth: 4 layers · 34 steps

Act III

A catalogue reflects its instruments.

Discovery method shapes which planets are easiest to find. This is not a map of habitability or of every planet that exists.

Time

2025 · Transit: 4,478 · Radial velocity: 1,158 · Microlensing: 262 · Other methods: 172 cumulative planets

Data & limits

NASA Exoplanet Archive (2026)

1992–2025 · cumulative planets

Discovery methods have strong selection effects; counts do not represent habitability or spatial abundance.

Open the OWID data page

Twenty worlds hidden inside the data

The sky is no longer empty.

Access creates a new commons problem.

Global total1

Launch operators12

We built a second ocean.

Growth moved behind nets and ponds.

Global production2

Aquaculture nations8

Capture fleets8

The planet skipped the wire.

Signal is not equal access.

Global technologies4

Internet users by country8

Mobile subscriptions by country8

Forty-three lanterns used to go dark.

A global triumph can contain local tragedy.

Country trajectories10

Long historical records7

World regions6

The planet learned to glow — unevenly.

The remaining darkness is geographically concentrated.

World2

Regions in the dark5

Countries in the dark8

A city can become a country-sized organism.

The skyline becomes a scenario.

Urban footprints24

The oldest district outgrew the nursery.

Longevity changed the skyline.

Age bands5

Ageing by region5

Under five by region5

We rebuilt the surface of a planet.

The modern era compressed the transformation.

Land classes8

Cropland by region8

Every energy age is still burning at once.

The newest towers are growing fastest.

Energy sources10

The more we built, the cheaper sunlight became.

A cheap module is not a free system.

Learning curve2

The world still loses a forest the size of a nation.

A plantation cannot simply replace a primary forest.

National net change10

Continental net change6

The average signal from 34,000 populations is fading.

An index is not a headcount.

Biogeographic realms7

Confidence interval2

Humanity stopped tearing the shield.

Stopping damage is not instant repair.

Substances6

HCFCs by region6

CFCs by region6

The hazards stayed. The death rate collapsed.

Hazards are not destiny.

Hazards9

Deaths by continent6

Earthquake deaths by continent6

Humanity built a mountain of endings.

Decline is not disappearance.

Stockpiles11

Peace is not a switch. It is a network state.

Institutions can change the topology.

Relationship types5

Severe rivalry by region6

Security communities by region6

A disease was pushed to the edge of the map.

Near zero is not zero.

World regions7

Last-mile countries8

The migration routes went quiet.

Catch fell. Recovery is slower.

Species11

Whaling nations12

Five centuries of catch ended in a cliff.

Closing the fishery did not rewind the ocean.

Reconstructed catch2