Assassination Politics is Jim Bell's technique for bringing about the demise of all
governments. It consists of a rather simple technique: Create a betting pool where people can bet anonymously on the time of death for a particular individual, using
encryption and
digital cash. The person betting correctly is presumably the
assassin that killed the individual.
The technique seems to be workable, though the effects on society would be scary, to say the least. More information on the technique is available in Jim Bell's writeup at http://jya.com/ap.htm (Everything will not let me link offsite :-(
Details on the actual protocol:
- The person placing the bet create a public/private keypair (see public key cryptography) puk (public key), prk (private key). This is a unique keypair used only this once.
- He further creates a secret key s, and encrypts his guess at the date of death along with the puk with s.
- He bundles the encrypted package with enough digital cash to pay for his bet, encrypts it with the public key of the bookie, and sends it off using some anonymous method (e.g, using anonymous remailers - or possibly a floppy dropped in an envelope).
- The bookie decrypts the bundle, and takes the digital cash, adding it to the betting pool (after having taken his cut.)
- If his bet comes true, he sends s to the bookie, again using some anonymous means.
- The bookie decrypts the bet, sees that it is correct, gets puk from the bet, encrypts digital cash equal to the money in the betting pool using puk, and publishes the result (using some public, untracable channel - e.g, by posting to Usenet.)
- The person that placed the bet (or somebody he has given prk to) retrieve the result, use prk to decrypt it, and get the digital cash.
Notice that the bookie neither knows who placed the bets nor, until the bet is fulfilled, what the bets are. Thus, there is no knowledge that the bookie can leak that make it possible to catch the assassin.
An observant reader that is versed in digital cash will notice that it is not possible to send out simple digital cash based on cut and choose protocols. This is solvable by using some non-interactive version of digital cash, for instance Stefan Brands' Choiceless Digital Cash (writeup coming). The better would include enough unsigned coins of different denominations to be able to cover any conceivable amount of money from the betting pool in his encrypted bundle; the bookie would pay the bank to have the appropriate subset of these signed before placing them in the result.