Pat Farrell's summary of the NIS&T conference (thanks for the report, by the way!) discussed a bunch of "criteria" that an "acceptable" GAK system should provide, including a couple that are supposed to limit the ability of law enforcement to use keys beyond the bounds of the search warrant. Another criterium that needs to be brought up (although I'm not sure of how one would phrase in a way that is either clean or polite) comes out of a debate between Philip Zimmermann and Dorothy Denning I ran across at one point: "How can a GAK system be arranged so that some future Congress cannot destroy the protections of the split-"escrow" system by issuing a resolution like 'All key components of suspected Comm^H^H^H^H terrorists shall be provided to the House Committee on Unamerican Activities'?" The links to McCarthyism are important here. GAK proponents can't claim this kind of thing won't happen -- it *has* happened, and could easily happen again. Anyone who claims otherwise is either terribly naive or is being outright misleading. In some sense, this criterium goes to the very heart of the whole GAK question -- even if you believe in the allegedly legitimate power of law enforcement to look through people's love letters, the "safety mechanisms" for the keys are nothing more than a set of flimsy policies that Congress could toss aside the next time something scary comes along. The protections that the GAK proponents are proposing (and proponing (-: ) are frightfully ephemeral. By the way, I'm a few days behind, so sorry if this is "old mail".