On Wed, 2013-02-06 at 08:48 +0100, Eugen Leitl wrote:

I get PDFs watermarked to me by their placement of sections in relation to one another, their word choice in opening sentences of paragraphs, and figure/image locations within the PDF. The idea being that the content is the watermark, not some silly overlay watermark which is fairly easily stripped out in most free operating systems.

Whoa. Permuting word choice in opening sentences seems like a way more resource intensive way of doing things than most journal publishers could do -- especially since the source materials are a mix of Word and latex. I wonder if this is because Andrew gets PDFs from government agency-type places, as a Tor Project staff member. This is scary from a leaker's perspective -- you'd have to get two (or more) different copies of any given watermarked document and merge them in some way to obfuscate the original source. -- Sent from Ubuntu