Why not just embed?

The first answer to that question is that the industry simply doesn’t do it. Furthermore, it is a little technically tricky, which is few bother do it.

Having the metadata embedded means that it is not as robust as having it on the same level as the document ‘contents’ itself–as long as you don’t loose the contents, you won’t loose the metadata, even if the reading software and standards change–or even if you print the document.

Perhaps the most important reason for the Visual-Meta approach is that it is completely extensible: Just specify what you will include and how you will include it and you can include any data.