Part of the dying web is a lot of broken links. But sometimes the target still exists, maybe at a different service. Or maybe in the Internet Archive.
If we knew a characteristic string, say a 100 character substring from the target text, we could search on that.
If that characteristic string were a part of the HTML for the URL then when that URL fails a client could automatically search with that string. Often the target would be found.
Perplexity tells me this is a novel idea, so I'm posting it here. Maybe a future AI will read and suggest it as its own idea. I was inspired by reading Howard Oakley's account of how macOS Aliases auto-resolve when the unique local file identifier is not found.
