Tim May writes:

By the way, a conclusion I draw from this is that an alternative way to signal list members should exist, even if it's just a very long list to send mail to from an unaffected system. Sort of an "Emergency Broadcast System."

Excellent idea! Here's a first blush at a specification for an EBS: 1) Mailing list data should be distributed to "N" sites where "N" is a magic number that minimizes the chance of losing all the copies. (Mathematicians, sharpen your pencils!) 2) The list data should be encrypted somehow to address these concerns: A) The keepers of the EBS list can't screw around with its contents B) The exposure of the list itself is limited C) Duplicate copies of the list can be verified for authenticity 3) Some sort of signalling protocol should exist for activating the EBS, testing, etc. This protocol might also involve authentication (key swaps, signatures, whatever) that would require a minimum of "X" participants to accumulate enough data to unseal the encrypted copies of the list. 4) Implementation of the EBS site function should not require involved systems programming tasks- a member should be able to recieve and store the encrypted EBS data and follow fairly simple procedures to activate and test the system without being a [your favorite language here] guru. I would guess that many cypherpunks are reluctant to go to their systems people and say, "Hey, can you give me unlimited access so I can develop an emergency broadcast system for the cypherpunks?" Enough for now- I'll refill my coffee cup, sit back, and watch you pick this apart. 'Cypherpunks write code.' Well, here's your chance to implement a system that has wide appeal, would be fun to play with, and incorporates distributed data encryption. ........................................................................ Philippe D. Nave, Jr. | The person who does not use message encryption pdn@dwroll.dw.att.com | will soon be at the mercy of those who DO... Denver, Colorado USA | PGP public key: by arrangement.