Summary

  • César Díaz says the same device and location produced different access to a set of repair diagrams after he changed internet providers.
  • The change points to a connection-path difference, but the essay does not publish the DNS, response, address or policy evidence needed to distinguish GeoIP gating from filtering or another delivery-path cause.

A changed provider is a clue, not a diagnosis

The useful fact in Díaz’s account is not that a GeoIP block has been proved. It is that access reportedly changed when the provider changed while the device and physical location did not. In a 14 October post, Díaz—identified by LACNIC Blog as its Telecom Affairs lead—says diagrams on an automotive repair site were unavailable over one connection and returned over another. He says he found information suggesting that a copyright notice had preceded a location-based restriction. The site, notice, providers, country, addresses and implementation are not named. The page also says contributor opinions do not necessarily represent LACNIC.

The comparison narrows the symptom to a difference between two connection paths. It does not hold the network constant. A new provider can supply a different public egress address, resolver, route, peering path or content-delivery edge. The service may also receive a different location estimate for the new address. Any one of those changes could matter; the essay does not supply evidence that selects among them.

That distinction matters because “the diagrams did not load” is an outcome, not a mechanism. A site might use an IP-linked country estimate to decide whether to return a page or asset. An ISP could apply a network-level rule to an address. A DNS response, origin decision, cache or CDN path could also differ. The account does not establish that any of these alternatives occurred. It does not say whether the site’s main page loaded, whether only image requests failed, what the browser displayed, or whether the result was reproduced at the same time.

RFC 7754 offers a useful way to ask where a failed transaction could have been interrupted: at an endpoint, a rendezvous service such as DNS, a network, or the server. The RFC is an architectural analysis, not evidence about Díaz’s unnamed site. RFC 9505 separately describes network-level IP blocking and its limits, including the risk that an address shared by several services can reach beyond the intended target. Neither document identifies the cause in this account or decides whether any particular measure was lawful.

The source’s account is therefore a lead for diagnosis. A provider switch changes several variables together. Keeping a phone or laptop and a street address constant does not isolate the egress address, the resolver, the route, the content edge or the site’s own policy. A two-provider comparison can show that a result is path-dependent as reported; it cannot, by itself, prove that a geolocation database or an ISP block produced it.

What would separate the explanations

A reproducible record would begin with the exact page and missing asset, timestamp, provider, public egress address and the DNS answers seen on each connection. It would then preserve the HTTP or TLS result, any block page or error, the destination address and—if a site-side country rule is alleged—the location dataset and policy version used for that request. An ISP-level claim would need different evidence, such as the applicable rule or an operator response. These are proposed diagnostic fields, not records that Díaz says he obtained.

The distinction between a network-range location and a person’s location is equally important. RFC 8805 defines a format for operators to publish geographic information about address ranges; RFC 9632 describes ways to find and use geofeed data, including optional authentication. Such a feed can improve the provenance of a prefix-level claim. It cannot establish where an individual user is sitting, identify the location estimate a website actually consumed, or show that a given access rule caused a failure.

The evidence boundary is narrow. Díaz’s essay establishes that he made the report and describes his own inference. It does not independently verify a cease-and-desist notice, identify the site or provider, publish measurements, or show the later fate of the diagrams. The regional study discussed in other LACNIC coverage is a separate subject; this account does not supply a case denominator or a second data point for it.

Sources