I mean, "whoopsie" bad pins are the main reason that most headers that indicate caching have max enforced cache lifetimes. According to a previous draft of the

I mean, "whoopsie" bad pins are the main reason that most headers that indicate caching have max enforced cache lifetimes. According to a previous draft of the Client Hints spec, Accept-CH's cache lifetime was supposed to only be 10 minutes, for exactly that reason! But apparently the only implementation so far (in Chromium) didn't have any internal TTL-based eviction logic for these entries at all...