If I remember right, that is almost exactly what they thought. Or rather he. I think it was one guy. The one who wrote the RFC. And no-one called him on it because at the time, that did not seem unreasonable.

4.3 billion devices that all need their own unique address? It’s not like everyone on Earth will need one.

What then followed was allocations of giant swaths of IPv4 addresses to large organisations, compounded by the fact that similarly large swaths were already reserved for special uses, leaving the whole thing with a problem basically from the outset.

I believe that one guy has said that he wishes he’d made it 64 bit and even thought about it at the time. But the “save every byte” mindset of the pre-Internet era was still very much alive and well, and I think that’s why he went for the smaller option.

“DNS never breaks. Nobody will ever have to type in an IP address”

• ipv6 inventors probably

It’s all allocated, but not all those allocations are for routing on the internet. Eg private ranges, localhost space, multicast, experimental ranges. Unfortunately you can’t repurpose those ranges as there is already kit out there that is hard coded to treat them a particular way.