6.8.1. Restrictions on cryptography--difficult as they may be to
enforce--may also impose severe hardships on secure operating
system design, Norm Hardy has made this point several times.
- Agents and objects inside computer systems will likely need
security, credentials, robustness, and even digital money
for transactions.
6.8.2. Proofs of identity, passwords, and operating system use
- ZKIPS especially in networks, where the chances of seeing a
password being transmitted are much greater (an obvious
point that is not much discussed)
+ operating systems and databases will need more secure
procedures for access, for agents and the like to pay for
services, etc.
- unforgeable tokens
6.8.3. An often unmentioned reason why encyption is needed is for
the creation of private, or virtual, networks
- so that channels are independent of the "common carrier"
+ to make this clear: prospects are dangerously high for a
consolidation under government control of networks
- in parallel with roads
+ and like roads, may insist on equivalent of licenses
- is-a-person
- bans on encryption
- The Nightmare Scenario: "We own the networks, we won't
let anyone install new networks without our approval, and
we will make the laws about what gets carried, what
encryption can be used, and how taxes will be collected."
- Fortunately, I doubt this is enforceable...too many ways
to create virtual networks...satellites like Iridium,
fiber optics, ways to hide crypto or bury it in other
traffic
+ cyberspace walls...
+ more than just crypto: physical security is needed (and
for much the same reason no "digital coin" exists)
- processes running on controlled-accesss machines (as
with remailers)
- access by crypto
+ a web of mutually suspicious machines may be sufficient
- robust cyberspaces built with DC-Net ("dining
cryptographers") methods?
By Tim May, see README
HTML by Jonathan Rochkind