Nitpicking this a bit, IP addresses don't necessarily identify unique destination devices at all. They're just a label to help devices make a decision about pac

Nitpicking this a bit, IP addresses don't necessarily identify unique destination devices at all. They're just a label to help devices make a decision about packets they receive. IP addresses are more about the journey than the destination. To expand, I think it's crucial to understand that in order for an IP address to have any use at all, the device processing the packet must have additional configuration to know what to with it. Generally, devices either process packets locally, forward to other devices (possibly after modification), or discard them. This decision happens even at the end user's device, where you have a "default route" to a gateway device that then knows better. The default route is literally just a route that says "the entire internet is over here". Mail package analogies are common, but unlike physical mail, with IP packets it's no trouble to send copies of packets to multiple receivers (multicast, sometimes used to broadcast video streams) or to any receiver that happens to be closest to the user (anycast, common with DNS), or evenly distributed to any configured destination (load-balancing). IP networking is fundamentally pretty simple, but the simplicity gives you quite an array of options when you start to think about different mechanisms a router could employ to make routing decisions, especially if they communicate with other routers to share information about the networks they're connected to.