Methodology
Every number on this site is built only from measurable, open data. Here is exactly how the score works — simply first, then in detail.
We add up how much planet-warming gas each country and company puts out in a year. The biggest emitter gets a score near 100. We then nudge the score a little up or down depending on whether they’re getting worse or better over time. That’s the whole idea — bigger and rising means a higher Climate Damage Score.
The Climate Damage Score
The score runs from 0 to 100 and is built from two measurable things only: how much an entity emits, and which way that’s trending.
logScore = log₁₀(E) / log₁₀(E_max) × 100
trendModifier = clamp( annual % change ÷ 10, −5, +5 )
Why a log scale? Emissions span a huge range. The largest oil state emits hundreds of times more than a mid-sized utility. On a straight scale, almost everyone would score near zero. A log scale (like the Richter scale for earthquakes) keeps the ranking meaningful. The biggest emitter anchors the top at 100.
Trend. We fit a line through up to five years of emissions. Roughly +1 point per 10%/year of growth, capped at ±5. The adjustment stays small so size always dominates — a fast-improving giant is still bigger than a small one.
Data confidence (shown, never scored)
Confidence is about our measurement, not the emitter’s behaviour. We never lower a polluter’s score just because the data about it is fuzzier — that would be unfair and would reward staying hidden. Instead we show a separate confidence badge, combining how well Climate TRACE measures the underlying assets with how much of an entity’s emissions we can tie to a named owner.
Attribution — who gets the blame
We rank entities (companies, state producers), with individual assets as the proof beneath them. Each asset’s emissions are split across the owners Climate TRACE lists for it. The dataset records owners but not exact ownership percentages, so we split equally across the named owners. Where no owner is known, those emissions are set aside as an honest “unattributed” figure rather than blamed on the wrong party — they are never ranked as a villain.
Everyday equivalences — how we convert tonnes
We never lead with raw tonnes. Every entity, country and share-card page shows the same emissions as something an ordinary person can picture — cars, flights, beef burgers, phone charges. The reader can swap between them using the pill row above the headline. The conversion factors below are constants in src/lib/equivalence.ts; nothing is approximated server-side beyond what the source itself reports.
- 1 petrol car for a year ≈ 4.6 t CO₂ (US EPA average)
- 1 home’s annual energy ≈ 2.7 t CO₂
- 1 large (1 GW) coal plant ≈ 4 Mt CO₂/year
| Equivalent | kg CO₂e per unit | Approximate? | Source / note |
|---|---|---|---|
| cars driven for a year | 4,600 | No | Average US passenger car ≈ 4.6 t CO₂/yr (US EPA Greenhouse Gas Equivalencies). |
| round-trip London → New York flights | 1,000 | No | Round-trip economy London↔New York ≈ 1.0 t CO₂ per passenger (DEFRA / ICAO ranges). |
| phone charges | 0.00822 | No | Charging a smartphone ≈ 8.22 g CO₂e per full charge (US EPA / IEA derivation). |
| hot showers | 0.5 | Yes | 10-minute hot shower ≈ 0.5 kg CO₂e on a mixed grid (varies widely by fuel and water-heater type). |
| cups of tea or coffee | 0.05 | Yes | Boiling enough water for one cup of tea or coffee ≈ 50 g CO₂e on a mixed grid. |
| beef burgers | 3 | No | Single beef burger ≈ 3 kg CO₂e (Poore & Nemecek 2018, Science). |
| people's entire yearly carbon footprints | 4,700 | No | Global average per-person footprint ≈ 4.7 t CO₂e/yr (Our World in Data, 2022). |
| trees soaking it up for a year | 21 | Yes | A mature tree absorbs ≈ 21 kg CO₂/yr — varies hugely by species, climate and age (US Forest Service). |
The three rows marked Yes (showers, kettle, trees) depend on the local electricity grid, water-heater type or tree species and age, so we badge them as approximate in the UI. Cars, flights, burgers, jeans and people use widely-cited single-figure benchmarks.
What we deliberately leave out
For now the score uses only measurable emissions and their trend. We deliberately do not fold PR claims, lobbying records, litigation history or ESG ratings into the headline number, because there is no globally consistent, free, objective dataset for them yet. The model is built so these could be added later as separate, clearly-labelled indicators — never quietly baked into the headline score.
Reproduce it yourself
Everything here is open. The same input data and the formulas above always produce the same score — no hidden weights, no training on private data. The reference year is 2024 (the latest complete year of Climate TRACE data). Refreshes from Climate TRACE on the 1st of each month.
Sources
- Climate TRACE — asset-level emissions & ownership (CC BY 4.0)
- World Bank — population (for per-person figures)
- Carbon Majors — historical producer emissions (context; Non-commercial)
The full technical specification lives in docs/scoring-spec.md in the project repository.