Guardrails, not gates: rethinking policy in platform teams

Good framing from Koray Oksay: policy on a platform team fails as a mental model before it fails as tooling. His observation is that most Kyverno and Gatekeeper deployments end up roughly 90% validation, while the platforms people actually like lean on mutation and generation, fixing or scaffolding instead of rejecting. The diagnostic is cheap: look at your last 30 policy interactions and count how many blocked versus quietly handled.