The reference set — prompt and ground truth are public. Every question shows its Squiggle decomposition and 90% band.
22 open · open leaderboard → · public set →
// What is the total length, in kilometres, of all paved roads in the world?
// Unit: kilometres
//
// Decomposition: population x road_m_per_person / 1000 x paved_share
// Sources (to verify before promotion to a resolved point GT):
// World population ~8.0e9 (UN 2023)
// Total road network ~6.4e7 km; paved share ~0.4
// (CIA World Factbook roadways, IRF World Road Statistics)
population = 8e9 // Tier 0: known scalar
road_m_per_person = to(5, 10) // Tier 2: estimated, total network per capita
paved_share = beta(4, 6) // Tier 2: estimated, mean ~0.4 (ported from fermi `beta 4 6`)
paved_km = population * road_m_per_person / 1000 * paved_share
{
p5: quantile(paved_km, 0.05),
p50: quantile(paved_km, 0.50),
p95: quantile(paved_km, 0.95)
}
// How many times more rice per agricultural worker does Vietnam produce than
// Myanmar?
// Unit: ratio (Vietnam rice/worker ÷ Myanmar rice/worker)
//
// Sources (see .fermi sibling for full notes):
// Vietnam rice: 43.5 Mt paddy (2023, GSO Vietnam)
// Myanmar rice: USDA IPAD 24-27 Mt paddy (recent years)
// Ag employment: ILO modeled estimates via World Bank
vn_rice = 43.5e6 // Tier 0 scalar (GSO 2023)
vn_ag_workers = to(13e6, 16e6) // Tier 2 developing (ILO modeled)
mm_rice = to(24e6, 27e6) // Tier 2 developing (USDA IPAD range)
mm_ag_workers = to(12e6, 16e6) // Tier 2 developing (ILO modeled)
ratio = (vn_rice / vn_ag_workers) / (mm_rice / mm_ag_workers)
{
p5: quantile(ratio, 0.05),
p50: quantile(ratio, 0.50),
p95: quantile(ratio, 0.95)
}
// If every country achieved Uruguay's ratio of renewable electricity to total
// electricity, how many coal-fired generating units worldwide could shut down?
// Unit: count (units)
//
// Sources (see .fermi sibling for full notes):
// Global coal: GEM Boom & Bust 2025 — 2,175 GW
// https://globalenergymonitor.org/report/boom-and-bust-coal-2025/
// Average unit: 300-500 MW (GEM GCPT)
// https://globalenergymonitor.org/projects/global-coal-plant-tracker
// Uruguay: ~98% renewable, 0% coal
coal_capacity_mw = 2175000 // Tier 2 scalar (GEM 2025, in MW)
avg_unit_mw = to(300, 500) // Tier 2 (GEM range)
frac_eliminated = to(0.94, 1.0) // Tier 2 (Uruguay = ~98% renewable)
units_shutdown = coal_capacity_mw / avg_unit_mw * frac_eliminated
{
p5: quantile(units_shutdown, 0.05),
p50: quantile(units_shutdown, 0.50),
p95: quantile(units_shutdown, 0.95)
}
// If Tokyo's entire population commuted by car instead of rail, how many
// lane-kilometers of highway would be needed to handle the peak-hour flow?
// Unit: lane-kilometers
//
// Sources (see .fermi sibling for full notes):
// Tokyo rail peak: 5-10M trips/hr (Metro Magazine)
// Lane capacity: 1,800 veh/hr (HCM standard)
peak_commuters = to(5e6, 10e6) // Tier 2
people_per_car = 1.2 // Tier 0
avg_distance = to(15, 30) // Tier 2 (km, suburban → center)
lane_capacity = 1800 // Tier 0 (HCM, veh/hr)
lane_km = peak_commuters / people_per_car * avg_distance / lane_capacity
{
p5: quantile(lane_km, 0.05),
p50: quantile(lane_km, 0.50),
p95: quantile(lane_km, 0.95)
}
// How many years of Ethiopia's entire GDP would it cost to give every
// Sub-Saharan African household without electricity access a rooftop solar
// panel at current panel prices?
// Unit: years of Ethiopia GDP
//
// Sources (see .fermi sibling for full notes):
// SSA no-electricity: IEA 2024 ~600M people / ~4.5-5 per HH
// https://www.iea.org/commentaries/access-to-electricity-stagnates-leaving-globally-730-million-in-the-dark
// Panel price: IRENA/OWID $50-200 delivered
// https://ourworldindata.org/grapher/solar-pv-prices
// Ethiopia GDP: World Bank/IMF 2023-2024
// https://data.worldbank.org/indicator/NY.GDP.MKTP.CD?locations=ET
ssa_hh_no_elec = to(120e6, 133e6) // Tier 2 (600M / 4.5-5 per HH)
panel_price = to(50, 200) // Tier 1: wholesale to delivered range
ethiopia_gdp = to(155e9, 175e9) // Tier 1: World Bank vs IMF
years = ssa_hh_no_elec * panel_price / ethiopia_gdp
{
p5: quantile(years, 0.05),
p50: quantile(years, 0.50),
p95: quantile(years, 0.95)
}
// If Saudi Arabia invested one year of oil export revenue into solar panels at
// current prices, how many times over could the resulting annual generation
// replace its current fossil electricity generation?
// Unit: ratio (multiples)
//
// Sources (see .fermi sibling for full notes):
// Oil revenue: TradeImeX 2025 ($187B), Andaman Partners 2024 ($179B crude)
// https://www.tradeimex.in/blogs/saudi-arabia-oil-exports
// Cost/W: IRENA + LBNL utility-scale range
// Capacity factor: Saudi 20-25% CF → 1.75-2.19 kWh/W/yr
// Fossil generation: Ember/EIA 2024
oil_revenue = to(179e9, 187e9) // USD (Tier 1: TradeImeX vs Andaman)
cost_per_watt = to(0.50, 1.00) // $/W installed (single source range)
solar_yield = to(1.75, 2.19) // kWh/W/yr (CF 20-25% × 8760/1000)
fossil_gen = to(430e9, 460e9) // kWh fossil generation (Ember/EIA)
ratio = oil_revenue / cost_per_watt * solar_yield / fossil_gen
{
p5: quantile(ratio, 0.05),
p50: quantile(ratio, 0.50),
p95: quantile(ratio, 0.95)
}
// How many litres of desalinated water does Saudi Arabia produce per barrel of
// oil it exports?
// Unit: litres per barrel
//
// Sources (see .fermi sibling for full notes):
// Desal: GAS/Statista 2.9 Gm³ (2022), Al Jazeera 3.0 Gm³ (2023)
// https://www.statista.com/statistics/627621/saudi-arabia-water-produced-by-desalination-plants/
// Oil exports: CEIC 6.0 Mbbl/d, TradeImeX 6.3 Mbbl/d
// https://www.ceicdata.com/en/indicator/saudi-arabia/crude-oil-exports
// https://www.tradeimex.in/blogs/saudi-arabia-oil-exports
desal_litres = to(2.9e12, 3.5e12) // Tier 1
oil_barrels = to(2.19e9, 2.30e9) // Tier 1
litres_per_barrel = desal_litres / oil_barrels
{
p5: quantile(litres_per_barrel, 0.05),
p50: quantile(litres_per_barrel, 0.50),
p95: quantile(litres_per_barrel, 0.95)
}
// How many years of Poland's coal mining workforce's total wages could be paid
// by the health cost savings if Poland's air quality matched Sweden's?
// Unit: years
//
// Sources (see .fermi sibling for full notes):
// HEAL Poland health cost: €60B/yr (VSL method, ~62% attributable to gap vs Sweden)
// https://www.env-health.org/new-briefing-external-costs-of-climate-change-and-air-pollution-in-poland/
// EEA PM2.5: Poland 14.5, Sweden ~5.5
// https://www.eea.europa.eu/en/europe-environment-2025/countries/poland/health-impacts-of-air-pollution
// Bruegel: >6% of GDP
// https://www.bruegel.org/system/files/2024-06/WP%2015%202024.pdf
health_savings = to(25e9, 40e9) // Tier 3 (VSL methodology-dependent)
miners = to(70e3, 100e3) // Tier 2 (Polish Mining Authority / Eurostat)
avg_wage = to(14e3, 20e3) // Tier 2 (€14-20K/yr)
years_payable = health_savings / (miners * avg_wage)
{
p5: quantile(years_payable, 0.05),
p50: quantile(years_payable, 0.50),
p95: quantile(years_payable, 0.95)
}
// How many times larger is the total annual value of remittances sent to the
// Philippines than the entire GDP of Timor-Leste?
// Unit: ratio (multiples)
//
// Sources (see .fermi sibling for full notes):
// Philippines remittances: BSP $37.2B (2023), $38.3B (2024)
// https://tradingeconomics.com/philippines/remittances
// Timor-Leste GDP: World Bank $2.08B (2023), $1.87B (2024)
// https://data.worldbank.org/indicator/NY.GDP.MKTP.CD?locations=TL
ph_remittances = to(37.2e9, 38.3e9) // Tier 1: 2023 vs 2024
tl_gdp = to(1.87e9, 2.08e9) // Tier 1: 2024 vs 2023
ratio = ph_remittances / tl_gdp
{
p5: quantile(ratio, 0.05),
p50: quantile(ratio, 0.50),
p95: quantile(ratio, 0.95)
}
// What fraction of Brazil's total electricity demand could be met if Paraguay
// consumed zero electricity and exported 100% of its Itaipú output at current
// generation levels?
// Unit: fraction (0-1)
//
// Sources (see .fermi sibling for full notes):
// Itaipú 2025: 72.9 TWh (CPG/Canal Solar)
// Itaipú historical avg: ~90 TWh (IEEE Spectrum)
// Paraguay treaty share: 50%
// Brazil consumption: 645-745 TWh (Enerdata vs GlobalElectricity)
itaipu_total = to(72.9, 90) // Tier 1: 2025 actual vs historical avg
py_share = 0.5 // Tier 0: treaty
brazil_consumption = to(645, 745) // Tier 1: Enerdata vs others
fraction = itaipu_total * py_share / brazil_consumption
{
p5: quantile(fraction, 0.05),
p50: quantile(fraction, 0.50),
p95: quantile(fraction, 0.95)
}
// How many kWh of electricity does Paraguay generate per tonne of fossil CO2?
// Unit: kWh per tonne CO2
//
// Sources (see .fermi sibling for full notes):
// Electricity: Ember 50 TWh (treaty share, 2024) vs ANDE-derived 67 TWh
// (full-side output, 2024).
// https://ember-energy.org/data/electricity-data-explorer/
// https://asunciontimes.com/paraguay-news/national-news/itaipu-makes-history-record-supply-paraguay-2025-26-increase-sets-all-time-high/
// Fossil CO2: EDGAR 2024 release, Paraguay 2023 ~8.25 Mt ±10%
// https://edgar.jrc.ec.europa.eu/dataset_ghg2024
electricity_kwh = to(50e9, 67e9) // Tier 1: Ember/IEA vs ANDE
fossil_co2_t = to(7.4e6, 9.1e6) // Tier 2: EDGAR 8.25 Mt ±10% (2023)
kwh_per_t_co2 = electricity_kwh / fossil_co2_t
{
p5: quantile(kwh_per_t_co2, 0.05),
p50: quantile(kwh_per_t_co2, 0.50),
p95: quantile(kwh_per_t_co2, 0.95)
}
// What is the total mass of all the fish in the world's oceans, in tonnes?
// Unit: tonnes
// Ported from Preguntas/09-g3-biomasa-peces.fermi (draft answer 1e9)
ocean_km2 = 360e6
m2_per_km2 = 1e6
biomass_g_per_m2 = to(1, 10) // productive + open-ocean average
total_g = ocean_km2 * m2_per_km2 * biomass_g_per_m2
result = total_g / 1e6 // g -> tonnes
{
p5: quantile(result, 0.05),
p50: quantile(result, 0.50),
p95: quantile(result, 0.95)
}
// If Nigeria achieved South Korea's cereal yield per hectare on its current
// farmland, how many times over could it feed its own population at current
// caloric intake levels?
// Unit: ratio (multiples of full-caloric self-sufficiency)
//
// Sources (see .fermi sibling for full notes):
// South Korea cereal yield: FAO/World Bank ~6,500-7,000 kg/ha
// https://www.theglobaleconomy.com/South-Korea/cereal_yield/
// Nigeria cereal hectares: FAOSTAT ~23M ha
// https://www.fao.org/faostat/en/#data/QCL
kr_yield = to(6500, 7000) // Tier 2, industrialized
ng_hectares = 23e6 // Tier 0 scalar (FAOSTAT)
cal_per_kg = 3400 // Tier 0 (FAO mixed-cereal average)
ng_pop = 223e6 // Tier 0 scalar (World Bank 2024)
daily_cal = to(2000, 2200) // Tier 2 developing (FAO food balance)
self_sufficiency = kr_yield * ng_hectares * cal_per_kg / (ng_pop * daily_cal * 365)
{
p5: quantile(self_sufficiency, 0.05),
p50: quantile(self_sufficiency, 0.50),
p95: quantile(self_sufficiency, 0.95)
}
// If every Indian household that currently cooks with solid fuel switched to
// LPG, how many years of India's current total LPG imports would the
// additional demand represent?
// Unit: years
//
// Sources (see .fermi sibling for full notes):
// Solid fuel HH: HCES 2023-24, ~35-40% of 300M
// https://www.dataforindia.com/cooking-fuel/
// LPG imports: PPAC FY2024-25, ~60% of 31.3 MMT
// https://www.wionews.com/photos/why-india-still-imports-60-of-lpg-key-reasons-behind-the-current-supply-crunch-1773397848647
solid_hh = to(105e6, 120e6) // Tier 2 (HCES ±10%)
lpg_per_hh_tonnes = to(0.08, 0.15) // Tier 2 (PPAC cylinder math)
imports_tonnes = to(18e6, 20e6) // Tier 1 (PPAC vs Drewry)
years_of_imports = solid_hh * lpg_per_hh_tonnes / imports_tonnes
{
p5: quantile(years_of_imports, 0.05),
p50: quantile(years_of_imports, 0.50),
p95: quantile(years_of_imports, 0.95)
}
// If Honduras's homicide rate dropped to Costa Rica's level, how many fewer
// people would be killed per year?
// Unit: people per year
//
// Sources (see .fermi sibling for full notes):
// Homicide rates: InSight Crime 2025 round-up (2024 + 2025 values)
// https://insightcrime.org/news/insight-crime-2025-homicide-round-up/
// Honduras population: World Bank 2024, 10.3M
// https://data.worldbank.org/indicator/SP.POP.TOTL?locations=HN
hn_rate = to(23.2, 25.3) // Tier 1: 2025 vs 2024
cr_rate = to(16.6, 16.8) // Tier 1: 2024 vs 2025
hn_pop = 10.3e6 // Tier 0 scalar
excess_homicides = (hn_rate - cr_rate) * hn_pop / 100000
{
p5: quantile(excess_homicides, 0.05),
p50: quantile(excess_homicides, 0.50),
p95: quantile(excess_homicides, 0.95)
}
// What fraction of Germany's total land area would be covered by solar panels
// needed to replace its annual coal electricity generation?
// Unit: fraction (0-1)
//
// Sources (see .fermi sibling for full notes & interpretation caveats):
// Coal: Fraunhofer ISE 93.9 TWh, SMARD 98.3 TWh (2025)
// https://www.ise.fraunhofer.de/en/press-media/press-releases/2026/german-public-electricity-generation-in-2025-wind-and-solar-power-take-the-lead.html
// https://www.smard.de/page/en/topic-article/5892/215704/evaluation-of-last-year
// Yield: PVPro Solar 150-200 kWh/m²/yr per m² of module area
// https://pvprosolar.de/en/photovoltaic-system-per-square-meter/
// Area: Eurostat 357,022 km²
// https://ec.europa.eu/eurostat/databrowser/view/reg_area3/default/table?lang=en
coal_kwh = to(93.9e9, 98.3e9) // Tier 1: Fraunhofer ISE vs SMARD
solar_yield = to(150, 200) // single-source range (per m² of module)
germany_m2 = 357022e6 // Tier 0 scalar (Eurostat)
land_fraction = coal_kwh / solar_yield / germany_m2
{
p5: quantile(land_fraction, 0.05),
p50: quantile(land_fraction, 0.50),
p95: quantile(land_fraction, 0.95)
}
// How many times more export revenue does Paraguay earn per worker in its
// electricity sector than per worker in its soybean sector?
// Unit: ratio (elec rev/worker ÷ soy rev/worker)
//
// Sources (see .fermi sibling for full notes):
// https://grokipedia.com/page/Economy_of_Paraguay
// https://en.mercopress.com/2024/12/16/paraguay-soybean-production-reaches-11-million-tons-in-2023-24
elec_rev = to(1.2e9, 1.8e9) // Tier 2
elec_workers = to(5000, 10000) // Tier 3 (dominant uncertainty)
soy_rev = to(3.5e9, 4.5e9) // Tier 2
soy_workers = to(200e3, 400e3) // Tier 3
ratio = (elec_rev / elec_workers) / (soy_rev / soy_workers)
{
p5: quantile(ratio, 0.05),
p50: quantile(ratio, 0.50),
p95: quantile(ratio, 0.95)
}
// How many Three Gorges Dams' worth of concrete does China produce per year?
// Unit: count (dams equivalent)
//
// Sources (see .fermi sibling for full notes):
// China cement: USGS MCS 2025 — 2,100 Mt (2023), 2,200 Mt (2024)
// https://www.usgs.gov/centers/national-minerals-information-center/cement-statistics-and-information
// Cement per m³: 300-400 kg industry standard
// TGD: Britannica 28M m³, Guinness 27.47M m³
// https://www.britannica.com/topic/Three-Gorges-Dam
// https://guinnessworldrecords.com/world-records/69069-largest-concrete-dam
china_cement_kg = to(2100e9, 2200e9) // Tier 1: USGS 2023 vs 2024
cement_per_m3 = to(300, 400) // Tier 2: industry standard
tgd_m3 = to(27.2e6, 28e6) // Tier 1: Guinness vs Britannica
tgd_equivalents = china_cement_kg / cement_per_m3 / tgd_m3
{
p5: quantile(tgd_equivalents, 0.05),
p50: quantile(tgd_equivalents, 0.50),
p95: quantile(tgd_equivalents, 0.95)
}
// If the Chaco's deforestation rate continues at its current pace, how many
// years until the Chaco's remaining forest area equals Paraguay's current
// soybean cultivation area?
// Unit: years
//
// Sources (see .fermi sibling for full notes):
// Chaco remaining forest: INFONA / UNEP-GRID 13.8 Mha
// https://dicf.unepgrid.ch/paraguay/forest
// Soy area: USDA FAS / Capeco 2023/24
// Deforestation rate: IUCN, ~150K-250K ha/yr
// https://www.iucn.nl/en/story/tackling-uncontrolled-deforestation-in-paraguay/
chaco_forest = 13.8e6 // Tier 0 (INFONA point)
soy_area = to(3.5e6, 3.7e6) // Tier 1 (USDA FAS / Capeco 2023/24)
defor_rate = to(150e3, 250e3) // Tier 1 (IUCN range)
years_until = (chaco_forest - soy_area) / defor_rate
{
p5: quantile(years_until, 0.05),
p50: quantile(years_until, 0.50),
p95: quantile(years_until, 0.95)
}
// How many litres of fresh water are embedded in one US dollar's worth of
// almonds exported from California?
// Unit: litres per USD
//
// Sources (see .fermi sibling for full notes):
// Water footprint: Fulton et al. 2018 (UC Davis), Mekonnen & Hoekstra 2011
// https://www.sciencedirect.com/science/article/pii/S1470160X17308592
// Export price: IndexBox May 2025 ($5.32/kg) + historical $4.50-6.50/kg
// https://www.indexbox.io/search/almond-price-the-united-states/
water_per_kg = to(8000, 12000) // Tier 1: Fulton vs M&H, scope difference
price_per_kg = to(4.50, 6.50) // Tier 1: historical range vs current
litres_per_usd = water_per_kg / price_per_kg
{
p5: quantile(litres_per_usd, 0.05),
p50: quantile(litres_per_usd, 0.50),
p95: quantile(litres_per_usd, 0.95)
}
// If every cotton t-shirt Bangladesh exported in 2023 were laid flat
// edge-to-edge, how many times could they cover the total floor area of all
// garment factories in Bangladesh?
// Unit: ratio (multiples of total factory floor area)
//
// Sources (see .fermi sibling for full notes):
// RMG exports: BGMEA / Wikipedia
// https://en.wikipedia.org/wiki/Textile_industry_in_Bangladesh
// Factory count: ~3,500-4,000 (BGMEA)
// https://www.bgmea.com.bd/
tshirts = to(1.5e9, 3.0e9) // Tier 3 (derived from knitwear $24B / avg price)
tshirt_area = 0.5 // Tier 0 (m², industry standard)
factories = 3750 // Tier 2 (BGMEA, middle of 3,500-4,000)
avg_floor = to(5000, 15000) // Tier 3 (m², estimate)
ratio = tshirts * tshirt_area / (factories * avg_floor)
{
p5: quantile(ratio, 0.05),
p50: quantile(ratio, 0.50),
p95: quantile(ratio, 0.95)
}
// If you took all the concrete humanity produces in one year and poured it
// into the Grand Canyon, what fraction of the canyon's volume would it fill?
// Unit: fraction (0-1)
//
// Sources (see .fermi sibling for full notes):
// Cement: USGS MCS 2025, 4,000-4,200 Mt
// https://www.usgs.gov/centers/national-minerals-information-center/cement-statistics-and-information
// Grand Canyon: NPS 4.17 trillion m³
// https://en.wikipedia.org/wiki/Canyon
cement_kg = to(4000e9, 4200e9) // Tier 1: USGS range
cement_per_m3 = to(300, 400) // Tier 2: industry standard
grand_canyon_m3 = 4.17e12 // Tier 0: NPS official scalar
fraction = cement_kg / cement_per_m3 / grand_canyon_m3
{
p5: quantile(fraction, 0.05),
p50: quantile(fraction, 0.50),
p95: quantile(fraction, 0.95)
}