What humanity is building — disease eliminated, rights extended, ecosystems restored, knowledge made common — drawing on multiple independent sources, with disagreement shown rather than hidden, and costs honestly listed alongside every gain.
Our World in Data charts the trends. The World Bank publishes the macro numbers. The WHO records the disease eliminations. GAVI counts the vaccines. The IUCN tracks species recoveries. UNESCO holds the literacy series. Each project is excellent inside its scope. None of them, by themselves, gives a citizen a complete view of what is being lifted — by whom, by how much, and how durably.
The Lift combines them. Same record format, same verification standards, same lens for every gain — including the ones that are unglamorous, slow-burning, or located far from the centers of media attention.
We start with the visualization layer — the shape of what an integrated record makes possible. The ingestion pipeline is being designed alongside; live data flows incrementally as each source connector lands.
What was raised, by whom, by how much, and at what cost.
No borders, no comparisons. Just the numbers for everyone alive, from the World Bank's world aggregate. Every figure here is real and sourced — fetched from the World Bank API, not estimated. Where a year has no data, it's left out rather than filled in.
This isn't a ranking, and it takes no position on anyone's politics. It's five real measures of how people actually live — grouped into outcomes and lived experience — each fetched from an established dataset, with the weighting left to you. Singapore, for instance, has long lives and high prosperity but among the longest working hours on earth; whether that nets out as "doing well" depends on what you value. We don't make that call — move the sliders and the map is yours.
A curated selection of real, cited historical gains — each entry carries its sources, its costs, and (where scholarship disagrees on whether something even counts as a gain, e.g. agriculture or the Columbian Exchange) the dispute. The one estimated quantity is the dot size: "reach" is a deliberately coarse order-of-magnitude figure used only to scale the dots — not a measurement. The events and citations are the grounded part; the sizing is a reading aid.
Real change in life expectancy at birth, 2005 → latest, by country (World Bank). The biggest gains are concentrated in sub-Saharan Africa — Uganda, Ethiopia, South Africa — as the toll of the AIDS crisis receded and child mortality fell. The few declines come from war and economic collapse. Click any country for its full profile.
False precision in the positive direction is as much a distortion as false precision in the negative direction. When sources publish different numbers for the same gain, picking one is editorial. So we don't. Each panel shows every source's figure, the methodology behind it, and an honest finding that doesn't pretend more confidence than the evidence supports.
Fifteen real world-aggregate series, grouped into six domains — every one fetched live from the World Bank, not estimated. Each is the slowest-moving signal in its domain, the kind of curve that rewards patient measurement. Click any panel for the full series and its source.
The same gains reorganized by the cultural design that produced them. Treaty cascades are different machines from vaccine campaigns; both differ from open standards. Reading the pattern matters: if we want more of the kind of gain on this page, we want more of the kind of arrangement that produced it. The question is structural — what compounds, and why.
A running feed of verified recent gains — each a real, dated, sourced milestone. This is the one part of the page that needs a live monitor rather than a downloadable dataset, so it's not built yet.
A pattern emerges across the major rights-expansion and protection treaties: a few of them are nearly universally ratified. Whatever else is true about compliance and reservations, this is the body of conduct the world's states have collectively committed to.
Single-source records are not suppressed — they are flagged for what they are. The mix of tiers is itself an indicator: when too high a fraction is single-source, coverage is shallow; when too high a fraction is disputed, sources are diverging in ways worth investigating. For The Lift, adjudicated records get particular weight because regulator approvals and IUCN status changes are durable, time-stamped facts.
No source covers every category. The Lift's value is connecting the coverage that exists, not replacing any of the underlying projects — all of which we link, cite, and credit.
A few load-bearing promises. A record of gains has to refuse to be marketing copy from the first day or it never earns trust.
If we count Haber-Bosch among the gains because it feeds half of humanity, we count its environmental costs in the same record. If we count the Green Revolution, we count its ecological and equity tradeoffs. A gain that needs to hide its costs is not a gain that holds up.
Every claim links to the underlying source, with caveats disclosed when known. We do not paraphrase past the source's confidence interval. We do not silently average estimates.
OWID, World Bank, UN-SDG, WHO GHO, GAVI, IUCN, IEA, UNESCO UIS, IHME — these are the projects whose work this record stands on. Our job is connection and presentation, not replacement.
When a record changes — a magnitude revised, an attribution corrected, a tier reclassified — the change is recorded with timestamp and reason in a public log. Optimistic numbers get retracted on the same record they were celebrated on.
Every ingestion script, every normalization rule, every magnitude derivation is in the open. The data the public sees is the data the public can audit.
The Lift is foundation- and reader-supported. Companies whose products are part of the record (vaccine manufacturers, renewable-energy firms, gene-therapy developers) do not fund it. The funding sources are published.
Four stages. Each is independent and replaceable. Records keep their full provenance chain end to end — at no point does a normalized record lose its connection to the underlying source document.
One connector per source. REST API where available (World Bank, UN SDG, WHO GHO, IUCN, ITU); CSV download on a schedule for annual datasets (UNESCO UIS, OWID, IHME); structured scrape plus LLM extraction for narrative or PDF sources (IPCC reports, GAVI impact pages). Each connector writes raw records to a per-source staging table.
Per-source mapping rules translate raw records into the normalized gain schema (see methodology). Country names → ISO codes. Actors → canonical entity ids via embedding-based candidate matching plus LLM verification on edge cases. Original record kept alongside normalized form.
Records that describe the same gain are linked, not merged. Candidate matching by metric + actor + date + summary embedding. LLM verification pass. Output is a canonical gain with the list of source records, magnitude range across sources, and a confidence score derived from source count, agreement, and corroboration window.
Verification tier computed: multi, single, disputed, adjudicated. Records flow into the public ledger with full provenance and any disagreement preserved. Every change writes a row in the corrections log, which is public and queryable.
Order matters. The first connectors are the structural backbone — once they're flowing, everything else is incremental.