Methodology, Sources, & licenses
Is any of this made up? No. Atlas Resolve collects no data of its own and guesses nothing. It is a confidence-scored synthesis of public, named datasets—each carried with its source, recency, and how sure we are of it. Where the data is thin, we say so and score it lower, rather than inventing a number to fill the hole. This page explains how a score is built; the Sources & licenses tab names every input.
How a score is built
We do not claim to compute where you should live—“best place to live” is not an objective quantity, and no honest tool can measure it. What Atlas Resolve offers is decision support under irreducible uncertainty: a way to weigh many criteria, on your own terms, while being explicit about what is and isn’t known. The construction follows the OECD/JRC Handbook on Constructing Composite Indicators, and we flag where we depart from it.
Every value is stored with its source, recency, geographic resolution, and a confidence level. A place is described as a profile across nine dimensions—eight livability axes plus long-run durability—never collapsed into a single hidden number. Each indicator is normalized to a 0–100 scale against a fixed reference, so a place stays comparable as it improves or declines, and overlapping sources for the same construct are layered as coverage-fill (the widest source as a floor, higher-resolution data on top) rather than double-counted.
Within a dimension, facets are combined so a genuinely bad facet cannot be fully cancelled by a good one—excellent courts do not make rampant crime “averagely safe.” (This is the same non-compensatory correction the UN’s Human Development Index adopted in 2010.) Across dimensions there is deliberately no single objective score: you set the weights, and you can set a floor below which a place is set aside rather than averaged in. Whether you can actually move somewhere is treated as its own question—feasibility, as a probability—kept separate from how good life there would be.
Confidence travels with every value and is always shown; where coverage is thin we lower it, and community knowledge is kept distinct from measured data. We do not validate against a single external index—that would only measure how redundant we are—but against purpose-built benchmarks and informed judgment for the places we know well, reporting the uncertainty rather than a bare score.
Why we don’t use crowdsourced prices
The most-quoted numbers in this space come from open crowdsourced databases—anyone can submit, and almost nothing is audited. The failure mode is documented: in 2017 a single contributor’s submissions made Lund, Sweden rank among the world’s most dangerous cities within a day, and the viral 2022 claim that Bradford was ‘Europe’s most dangerous city’ traced to the same kind of index before being debunked against official statistics.
We use named, licensed public datasets instead—slower to update, thinner in places, and honest about both. Every source is listed, with its license, on the sources page; every value carries its vintage and confidence.