Defaults.Exposed › Trends

Trends

Each edition of the census is measured again from scratch, so two figures from two editions are only comparable when the same kind of thing sits under the line. These pages carry the series that pass that test, with the denominator named in every sentence that quotes a number.

• The DMARC policy mix across three editions: none, quarantine, reject
  Across three 2026 editions about half of published DMARC records sat at p=none, quarantine edged up, reject edged down. A p=none record changes nothing.
• HSTS and CSP presence across three editions
  HSTS share of answering domains: 19.89%, 19.11%, 19.24% over July to September 2026; CSP: 7.64%, 7.93%, 8.02%. Counts rose for both, the HSTS share did not.
• TLS 1.3 across three census editions: what moved and what didn't
  TLS 1.3 was negotiated on 94.81%, 94.87% and 94.41% of completed handshakes across July, August and September 2026; TLS 1.2 on 5.19%, 5.13% and 5.59%.
• What we will and will not compare between census editions
  Two census figures are compared only when both denominators are an observed record set in the same edition; shares of all domains are not compared.
• Why .nl and .fr look safer this month, and what changed in the counting
  September grades .nl at 39.8% F and .fr at 48.3% F of domains carrying a grade; what moved since the last edition is the inventory counted.

How a figure is measured before it is compared: the 34 checks and the grade. The current values are on the census data page.