Summary

  • In the 30 October APNIC Blog report, the Chile test put 9.5% of analysed web requests at server IPs geolocated to Chile through ExpressVPN, against 62.7% through Surfshark and 64.2% through NordVPN.
  • The authors’ original paper compared four VPN providers across 14 countries. It found large country-location differences in eight; six showed little difference. Chile was the loudest demonstration, not a universal correction factor.

A national Internet can look local or foreign depending on which tunnel a researcher rents. The Chile table is the giveaway. Through ExpressVPN, 57.1% of the tested requests pointed to IPs geolocated in Brazil; through NordVPN, the Brazil share was 2.4%. The websites in the experiment did not have to move. The observation path did.

VPN-Scope scheduled 25 visits to each of 1,000 sites per country and provider over three weeks in March and April 2025. The authors included government, locally popular and globally ranked sites, loaded full pages and counted requests for page components as well as the initial domain. The Chile percentages describe requests in that tested set. They are the numbers a single-vantage measurement could turn into a claim about national infrastructure.

The mechanism runs through the VPN exit, DNS answers and content delivery routing. Two exits advertised as being in the same country can ask different resolvers, get different addresses, or reach different replicas. The paper reports a Jensen–Shannon distance of 0.51 between ExpressVPN and NordVPN for Chilean destination-country distributions, but 0.03 between NordVPN and Surfshark. Provider choice mattered more than the country label in that comparison.

For a public assertion that Chilean web requests stay inside Chile, the missing record is the provider, exit location, resolver and repeated cross-provider result. Publish those beside the percentage. Without them, a neat national score may just be a portrait of the measuring tunnel.