No black boxes

How we score

Every Sketch Score is produced by math, not vibes. Here's the whole recipe — including the parts we're still building and the things we deliberately refuse to guess.

Where the data comes from

  • Four governments. Travel advisories from the United States, the United Kingdom, Canada, and (where reachable) Australia — each an independent read on the same country. When they disagree, we show you all of them and flag it.
  • World Bank governance data. Political stability and rule-of-law estimates from the Worldwide Governance Indicators.
  • CDC travel health notices. Active disease and outbreak warnings, by level.
  • UN homicide statistics. UNODC intentional-homicide rates — the crime sub-score's backbone.
  • ACLED political-violence data. Recorded political-violence events and fatalities over the trailing year (via the UN's Humanitarian Data Exchange), blended into the political sub-score so it reflects what's actually happening, not just perception indices.
  • Numbeo city crime indices. Resident-submitted crime perception per city — this is what powers the city pages' crime numbers.
  • WHO road safety data. Estimated road traffic deaths per 100,000 population — the roads sub-score's backbone.
  • INFORM Risk Index. Natural-disaster hazard exposure (earthquake, flood, tsunami, cyclone, drought, epidemic) from the EU Joint Research Centre and partners — the nature sub-score's backbone.
  • Real traveler chatter. Reddit threads (quoted with links) and Wikivoyage safety sections (attributed, CC BY-SA) — words, never numbers.
  • UK FCDO terrorism briefings. Real per-country prose, shown as-is with attribution — not a score. No dataset gives a clean per-country terrorism-risk number without inventing a severity heuristic, so this stays words, never a fabricated number.

More sources — terrorism data, women's-safety indices — are still being wired in. Until a source is live, we leave its sub-score off the page rather than invent a number.

The math

The overall score is a weighted blend of the sub-scores we actually have data for, each normalized to 0–100 (higher = safer). If a source is missing for a country, its weight is redistributed across the rest — we never treat "no data" as "zero."

advisory consensus · weight 0.40

political stability (0.7 WGI + 0.3 ACLED violence rate) · weight 0.25

health notices (CDC) · weight 0.20

crime (UNODC homicide rate) · weight 0.20

rule of law (WGI) · weight 0.15

roads (WHO road-death rate) · weight 0.08

nature (INFORM natural-hazard exposure) · weight 0.08

The hard caps

Some situations override the arithmetic entirely. A slow-moving governance index shouldn't keep a country's score high when a war just broke out, so:

  • • Any government's "do not travel" advisory caps the score at 39 ("Nope.").
  • • Two or more governments at "avoid non-essential travel" caps it at 54.
  • • An active CDC Level 3 health notice caps it at 64.

The verdicts

85–100 Go.

70–84 Go, but pay attention.

55–69 Go with a plan.

40–54 Think twice.

0–39 Nope.

What we won't do

We don't hide the sources — every country page has a "show the receipts" section linking the exact feeds behind its score. We don't let language models set numbers; they only write the plain-English summaries, and only from cited facts. And we don't pretend a quiet data gap is a safety guarantee. Right now 195 countries are scored; the rest of the planet is coming.

Sketch Scores are informational opinions synthesized from public data, not professional security advice. Always check your own government's official travel advisory before you go.