HPKP, Expect-CT, or plain CT monitoring: why did pinning lose to transparency?

HPKP, Expect-CT, or plain CT monitoring: why did pinning lose to transparency?

How did the industry decide to defend against rogue certificates, and why was one approach abandoned? The history is a clean case study in operational risk beating cryptographic ambition.

HPKP (HTTP Public Key Pinning, RFC 7469) had servers send hashes of allowed public keys; browsers refused certificates outside the set. Powerful in theory, catastrophic in practice — a pinning mistake or lost key could brick a domain for the pin's lifetime, and attackers could exploit it for ransom-style hostile pinning. Browsers deprecated and removed it.

Expect-CT was a transitional header forcing certificates to comply with Certificate Transparency, reporting or blocking violations. It served its purpose and is now obsolete because CT enforcement became mandatory in browsers by default.

What replaced both is CT itself (RFC 6962/9162) plus monitoring: no brittle client-side state, just universal logging you observe.

— Never deploy HPKP; it is removed and dangerous.
— Ignore Expect-CT; CT enforcement is now built in.
— Rely on CT-log monitoring for the detection HPKP promised without the self-inflicted outage risk.

Further reading: RFC 7469 (historical); RFC 6962; browser HPKP-removal notes.

Bottom line: pinning's blast radius killed it; transparency delivered the same threat coverage without the ability to brick your own domain.