Some metadata may be more effectively carried at the system level than at the codec level. As the VSEI specification continues to expand - from 33 SEI messages in the 3rd edition to 48 in the 4th edition, with additional candidate SEI messages under consideration - it may be timely to reflect on whether each metadata type is being standardized at the most suitable layer of the media stack. For a number of metadata types that are primarily descriptive of content or intended presentation, file-format-level representations in ISOBMFF may offer practical advantages, including improved accessibility, greater robustness in remuxing and transcoding workflows, and reuse across multiple codecs. This contribution discusses these considerations, proposes a file-format-aware evaluation principle for future metadata work, and suggests closer coordination between video coding and systems groups when defining new metadata mechanisms.