The term "Happy Eyeballs" is basically a euphemism for "good user experience". The RFC was trying to balance network load and efficiency versus fast enough timeouts that users wouldn't readily notice that they were timing out and failing over to another address family. Thus the delay was short enough (200-300ms) that their eyeballs remained "happy" as it didn't negatively impact their experience.
I have just finished listening to the podcast about this topic and decided to subscribe to your RU-vid channel. Thank you for sharing interesting things!
Great explanation of Window's TCP/IP. Douglas Comer would be proud. P.S.: you have some very strange bots in your comments. Edit: How many people know who Douglas Comer is?
Oh that is a good question, I really don't know the answer. But If I would to guess the underlying kernel is a Windows, and linux comes on top, so we should see the retries .
I don't think people should use IPv6 without NPTv6 , having internal machines publicly addressable outside was a mistake. with NPTv6 the firewall can block access to protected resources, for ex, you might only allow SIP or voip to pass to internal machines, not everything. And yes I know about rfc4941, it just put too much burden on all devices in the network instead of a device made for it, a router/firewall.