Glossary
Ranking methodology terminology — what we measure when we rank privacy tools and why.
Canonical version maintained at anonguide.com.
A
- Anonymous Account policy #
-
A provider account that does not require an email address or other identifier at signup. Examples: Mullvad's 16-digit numbers, IVPN's account IDs. Crucial when the threat model includes payment-link attribution.
I
- Independent Audit methodology #
-
A review of a provider's code, infrastructure, or operational practices by a third party with no commercial relationship to the provider. We weight recent audits (last 18 months) more heavily than older ones.
J
- Jurisdiction policy #
-
The country whose laws govern a provider. We note jurisdiction but weight it less than operators and competitors often do — surveillance partnerships (14 Eyes, etc.) are real but the absence of logs to compel matters more than the country.
K
- Kill Switch network #
-
Client feature that blocks all traffic when the VPN tunnel drops. We require it as table stakes — anything without one is ranked lower regardless of privacy claims.
M
- Multi-Hop network #
-
Routing through two or more VPN servers in different jurisdictions. Splits trust across providers/locations. Worth using only if your threat model includes a hostile single jurisdiction.
N
- No-Logs Policy policy #
-
A VPN provider's claim not to retain logs that could tie a session to an account. We only credit no-logs claims that have been verified by an independent third-party audit examining production infrastructure.
O
- Open Source methodology #
-
Source code is publicly available under an OSI-approved license, allowing independent review and rebuild. We credit open-source clients more than open-source servers because clients are the attack surface most users touch.
R
- RAM-Only Servers infrastructure #
-
Infrastructure where servers run entirely from volatile memory; on reboot, no persistent state remains. Reduces the value of physical server seizure for logs forensics.
T
- Threat Model methodology #
-
Who you are protecting from, what they would try, and what you can do about it. Our rankings cover three model tiers: "casual", "journalist/activist", and "high-risk" — products that win one tier may not win another.
W
- Warrant Canary policy #
-
A regularly republished statement that no warrants, gag orders, or subpoenas have been received. The statement disappearing implies that one has. Useful only for providers that update them on a strict schedule.