Google does not use humans to read your email, only computers. These computers search for keywords that trigger the advertisements, however, if they hapen to find a catastrophic event or tragedy Google errs on the side of good taste and removes the ads altogether. […] After extensive testing I’ve discovered you need 1 catastrophic event or tragedy for every 167 words in the rest of the email. I usually toss in a couple extra for good measure. I’ve been told by an early adopter that the very elegant and self explanitory “These words are designed to kill advertisements” works wonderfully.
What fascinates me about this is that this is fundamentally the same idea as a popular form of NSA-baiting Usenet posters hit upon 15 years ago: sprinkle random incendiary keywords (e.g. terrorism, murder, child porn etc.) in your signature so the NSA (National Security Agency - or, if you prefer, No Such Agency) would pick up on your message and their computers would get swamped with false positives.
(Never mind that the NSA, if they were ever so slightly bothered by this sort of stuff, would just adjust their algorithms to mostly or even entirely ignore signatures.)
Then, people assumed that a great threat to their way of life was that the NSA - a major secretive government body with far too much computing power and secret technology - would tap into their usenet communications. Now people know that Google (or any semi-competent large company) already knows absolutely shedloads about them, and just wish for Google to leave them the hell alone.