This is generally a problem with titles or summaries.

I never thought of Twitter like that. And under these thoughts 140 char limit makes even more sense. The content IS the reference. And in nostr we have note hashes which address perfectly.

But nevertheless we are visual animals. So a link preview to a product page of a shoe probably should show this exact shoe because the link represents a very specific thing. An image increases the value of the addressing scheme in this case, don‘t you think?