The PDFs are a couple of GB every year - I could limit that further by putting more aggressive filtering in place. There is also a rule that copies the contents

The PDFs are a couple of GB every year - I could limit that further by putting more aggressive filtering in place. There is also a rule that copies the contents of image and PDF URLs verbatim without converting them. Over time I'm going to extend all of this to a more fully-fledged "sousveillance" dataset. I would love to add health and location data, but alas Apple is such a walled garden. But I did add a note-taking system, Zettelkasten-style.