We have a greyhound, and they are closer to cattle than pets at the beginning of their lives. Each dog has a unique combination of numbers tattooed in their ears. Then they are named according to a weird convention where litter-mates share a "first name" while having their own unique "second name". For example, my greyhound was named Del Sol Madison, and she had litter mates with names like Del Sol Martin, Del Sol Maxine, etc. These names are used for identifying the dogs, but they are not trained to respond to them because they aren't pets. Once we got her, we called her Maddie and that's just been her name ever since.

My family adopted a greyhound when I was a kid, was the sweetest dog. His track name was High Tech Kaddy Shack, we called him Kaddy.

When I worked at NASA, I basically did both. We had fixed format official "cattle" names that included the organization code and the computer's inventory number. Those names were useless to humans, so I always also registered a human-friendly "pet" alias. It was a bit more trouble, but it generally worked well. The pet names also has some structure, with prefixes differentiating classes of systems (desktop, server, lab), but they were otherwise free-form and picked to be useful to the users.

I've done the same: cattle names for the A record, pet names as a CNAME.

Then you get to argue about the PTR record ...

You can have multiple PTR records.

You're right, in most cases, it's possible.

It's generally considered to be a bad idea though, and in the case mentioned above, local policy forbade it.

The RFC was written in 1990. Very few computers were cattle then.

I named my pet after my neighbor.

Do you call your neighbor “Human Nigel”¹?

1. https://twitter.com/DawgBelly/status/862786851072909312