This method of storing metadata is robust since as long as the content of the document is available, the metadata will also be available, even to the point of printing the document, then scanning it and performing OCR on it.
Furthermore, the actual Visual-Meta contains instructions for how to implement it in plain language, which will allow any developer to integrate Visual-Meta import or export, now, and in hundreds or thousands of years in the future. This is far and away the most robust way to store rich digital text metadata currently available.
Data which exists only in proprietary formats are more likely to become unreadable since there are less of them and thus less of a reason for future developers to support access to them. Data which is accessible through the web relies on upkeep of paying for domain names and server costs. Data which is contained in a widely shareable, open format, such as PDF, on the same level as the ‘contents’ and which connects using the citation method of specifying the bibliographic details of a source so that it can be located and used from any location (like a traditional printed journal) rather than only from a web addressed repository, makes for a robust, long term solution for publishing and sharing our knowledge.
This approach does not violate any PDF standard since it is indeed just textat the end of the document and thus PDF documents with Visual-Meta can be opened by any PDF viewer.