I'm very curious as to how this works at a basic level. Perhaps a 'positive-control' system in which he has to send a signal to some clandestine web service every 24 hours to prevent the keys from being released? Does he have an Arduino board strapped to his chest detecting his pulse? Or has he simply entrusted some mechanism to somebody else, who can determine whether the keys should be released depending on the nuances of the situation in which he is harmed/killed?
I guess you could just use the wikipedia API. Just try changing "is" to "was" in the first sentence and see what happens.
Complicated technical measures expose the data to risk of decryption prior to possible activation (by the NSA or someone else). Just printing the keys and giving to trusted people would be far less risky.
Or an arrangement with one or two friendly newspapers that a specific control phrase be inserted in to any editorial covering his death.
Then all Snowden would need is one or two cloaked servers somewhere that crawl each newspaper's site daily and trip the switch if they find an article containing the phrase.
Yeah, there are a lot of ways it could fail. You could probably build a pretty reasonable signal by looking at a lot of trends on Google News, Twitter, Wikipedia, etc. If you were building this in far enough in advance you could also test it by looking for high probability events besides your own death or demise.
What I like about it is that you could build a fairly hermetic system that, once you set it up, is never directly touched by you and just accesses public, high-volume websites. If you can set it up without leaving any fingerprints, that seems like a big plus.
