Amazing!

I can imagine how it works on the server side, but what's really fascinating is that my client (macOS) can handle negative response times at all.

Does anybody know how that's possible? Do ICMP sockets just queue incoming "responses" until there is a pending matching request?

Or does the ping command take complete ICMP "ownership" over... something? more than a socket? while it runs, and this is an implementation quirk of that userspace tool's synchronization/threading/event handling mechanisms?