An interesting class of links are those embedded in a document without being marked up as a hyperlink. You can find them in places like news stories -- traditional media almost never hyperlinks urls that people say. They are also not uncommon in books, especially recent fiction. And also in audio and video.
These dark links are interesting partly because while they're not hyperlinks now, they could become hyperlinks in the future. For example, a book is converted to an ebook, and the ebook is read on a reader that automatically hyperlinks. It could happen with transcripted speech as well.
They're also interesting because some people will manually follow them in the present. And because the intent of the person who put the dark link in is often different than the usual intent of using a hyperlink. Some authors use dark links as a "rabbithole" leading to an Alternate reality game. Others make up a link that fits the context, without, apparently, worrying much about what's there in real life.
Grepping through piles of ebooks to find dark links, and domain squatting on them is one of those business models I have filed away in case I ever go bad. But it's probably being tried by others anyway. Although I have seen some dark links lately whose domain was not yet registered.
An early draft of RFC 2606 included reserved top level domains for ficticious domain names, including ".nil", ".xy", and ".tld", but that did not survive into the final RFC, which only provides "example.*". (".xy" is still probably a safe choice for fictional domains, since as the draft notes, that code will never be assigned to a country.)