Selective compression does seem like a pretty optimal approach. > We found that in most cases a secret within a webpage can be described in terms of a classica

Selective compression does seem like a pretty optimal approach. > We found that in most cases a secret within a webpage can be described in terms of a classical regular expression. Not sure I'd want to rely on a regex that identifies the right chunk most of the time..... that sounds fragile, easy to accidentally break with a routine update of the page. Explicitly tagging the secret portion seems like a more robust approach.