Datasets
Security base rates, measured in the open 32 controls measured passively across defined populations of domains, packages, certificates and networks — sliced by country , company size , ecosystem and issuing CA. Each figure is the share not meeting the control, with a 95% confidence interval. Aggregate-only; no company is ever named; free to reuse (CC BY 4.0).
100% What share of certificates issued by ZeroSSL GmbH have a validity window longer than the 47-day maximum being phased in? ZeroSSL GmbH (certificates) · n=735 Almost every one · above the global rate of 91% · little changed since 8 November Chosen automatically, rotating weekly: each week this is one of 6 controls that most of their population fails (n≥400), highest severity first · the selection rule is fixed, the figures update as measurements change All checks, by rate 28 controls, ranked by the share of the population not meeting each. Each row shows the control's worst credible segment — the slice of the population (n≥400) where the gap is widest.
Subject All Certificates 7 Email auth 5 Email transport 4 Supply chain 3 Transport 3 Identity 3 DNS integrity 2 Routing 2 Attack surface 2 Web hardening 1
Measured over All domains 25 packages 3 certificates 2 networks 2
% not meeting the control · higher is worse 0 50 100%
Certificate lifetime readiness What share of certificates issued by ZeroSSL GmbH have a validity window longer than the 47-day maximum being phased in? worst credible segment: ZeroSSL GmbH (certificates) · n=735 · sev medium 100% MTA-STS What share of domains worldwide have no enforcing MTA-STS policy (absent or testing mode)? worst credible segment: global domains · n=14,820 · sev medium 100% TLS reporting What share of domains worldwide have no TLS-RPT record? worst credible segment: global domains · n=14,804 · sev low 100% Nameserver resilience What share of UK companies have all nameservers on a single provider? worst credible segment: UK companies · n=2,618 · sev low 98% DANE for mail What share of UK small companies have no DANE/TLSA record binding their mail server's certificate? worst credible segment: UK small companies · n=531 · sev low 98% CAA record What share of UK small companies have no CAA record? worst credible segment: UK small companies · n=932 · sev low 98% DNSSEC What share of UK small companies have no validated DNSSEC (unsigned, or signed with no DS at the parent)? worst credible segment: UK small companies · n=932 · sev medium 95% RPKI route origin validation What share of networks in South Korea have no RPKI ROA, so their route origins cannot be validated? worst credible segment: South Korea (networks) · n=933 · sev high 91% DMARC policy What share of domains worldwide have a DMARC policy that is absent or monitor-only (p=none)? worst credible segment: global domains · n=14,820 · sev high 85% Security headers What share of domains worldwide have no HSTS or Content-Security-Policy header? worst credible segment: global domains · n=13,549 · sev medium 83% Package provenance What share of packages on npm have no build-provenance attestation tying the artifact to its source? worst credible segment: npm (packages) · n=677 · sev medium 79% DMARC reporting What share of domains worldwide have dmarc-reporting? worst credible segment: global domains · n=6,915 · sev medium 72% SPF policy What share of domains worldwide have no enforcing SPF record (absent, ~all soft-fail, or +all)? worst credible segment: global domains · n=14,820 · sev high 72% Package maintenance What share of packages on npm have had no release in over two years? worst credible segment: npm (packages) · n=683 · sev medium 57% Post-quantum key exchange What share of domains worldwide have tls-pq-readiness? worst credible segment: global domains · n=1,721 · sev medium 48% HTTPS redirect What share of France companies have no redirect from http to https? worst credible segment: France companies · n=716 · sev medium 33% Certificate lifetime What share of UK medium & large companies have certificates issued for longer than the 47-day maximum? worst credible segment: UK medium & large companies · n=728 · sev low 32% Package install scripts What share of packages on npm have an install script that runs on every install? worst credible segment: npm (packages) · n=677 · sev medium 32% Subdomain sprawl What share of UK medium & large companies have ten or more discoverable subdomains? worst credible segment: UK medium & large companies · n=1,072 · sev low 19% Legacy TLS What share of domains worldwide have legacy TLS 1.0/1.1 still enabled? worst credible segment: global domains · n=13,385 · sev medium 15% Certificate trust What share of UK medium & large companies have a certificate that is self-signed or invalid for its hostname? worst credible segment: UK medium & large companies · n=842 · sev high 10% Mail server reverse DNS What share of France companies have a mail server with no reverse-DNS (PTR) record? worst credible segment: France companies · n=814 · sev low 9% Certificate expiry What share of UK medium & large companies have a TLS certificate expiring within 30 days, or expired? worst credible segment: UK medium & large companies · n=725 · sev high 7% Mail server certificate What share of France companies have a mail server presenting a self-signed or expired certificate? worst credible segment: France companies · n=599 · sev medium 6% Subdomain takeover What share of Norway companies have a subdomain that is a takeover candidate? worst credible segment: Norway companies · n=983 · sev high 5% Federated identity provider health What share of UK companies have a federated identity provider endpoint that no longer answers? largest segment: UK companies · n=105 · sev high 2% Mail transport encryption What share of UK companies have a mail server that accepts mail without offering STARTTLS? worst credible segment: UK companies · n=1,112 · sev high 2% Certificate crypto What share of UK small companies have a certificate with a weak key or signature (SHA-1 or sub-2048-bit RSA)? worst credible segment: UK small companies · n=646 · sev medium 1% No datasets match that combination.
Context measures Composition of the same population — not failures, and deliberately not ranked above. They give the denominators the checks above sit inside.
For AI agents: MCP server Every dataset here is also served over MCP — an agent can look up a base rate, get the full measurement doc with citation metadata, and run a live check of a specific domain against it. Free, no auth; live checks are rate-limited.
claude mcp add --transport http quiet-failures https://mcp.quietfailures.com/mcp Six tools, other clients' config, limits and licensing: full MCP documentation → Prefer raw files? The catalog is at /datasets/manifest.json and every dataset page has .json / .csv.