Eric (Hollander; my last post was to Eric Hughes)- Your "small laptop with (goodies)" was EXACTLY what we were trying to go for in 1983/84. This was the "Cryptex CS-3" project. Remember that there was no such thing as a laptop at the time. What I was proposing was a self-contained portable encryption terminal. It would have measured about a foot wide by 10" long by about 2-3" thick, had an LCD for about six to eight lines of text at a time, two 3-1/2" FDDs, a pair of sockets for "Codepacks" (hardware key storage devices which would have been tamper-resistant and password protected), a good quality keyboard, a modem with modular jack and optional acoustic couplers (for payphones: low-tech anonymity)... the Tempest feature would have been achieved by putting 100dB of white noise on the metal housing, which would have been imperfect but decent enough. There was no plan for a hard drive; the operating software would have been a simple line editor and the crypto routines, all burned into ROM and part of the main processor board. No plans for thermite linings at the time either, though we did have a password routine with a duress option, which would have erased anything on the FDDs or the Codepacks. My first intended market for this thing was the political dissident community, where communication has always suffered to a small but noticeable degree by the "we can't talk on the phone" factor. It never got going because the market was too small. Now it would seem the time has definitely come... though whatever ultimately arises out of the cypherpunk scene will be many many times more sophisticated and versatile. -gg
I thought of another thing to add to my wish list for a dedicated cryptosystem: analog input and output, for use as a phone line scambler. Such a system could be manufactured for not too much money, I think. It would be like a specialized version of the Apple Newton. Apple would never make something like this, though; they are becoming good buddies with our favorite agency. Also you would probably need a permit to own thermite. e
Also you would probably need a permit to own thermite. I don't think there's a problem with owning it or making it -- only (perhaps) selling it and transporting it; thermite is not strictly an explosive. You may wish to consider alternate ways of destroying the data, especially if you wish to ever transport the device on a commercial airline; if you've only got one RAM device that has critical data in it, then simply arrange for the battery backup circuit to have a "high current mode", perhaps feeding more of the
cryptosystem: analog input and output, for use as a phone line scambler. Such a system could be manufactured for not too much money, I think. It Let's see -- an ISDN-quality (quality? I use the term loosely) codec should be under $50 single quantity, the data rate isn't very high so you don't need much of a CPU (6811 might even be enough, and
pins -- a "light emitting RAM" should be just as blank as one burned through by thermite. they're easy to interface to things -- lots of on-chip I/O). You'd need a modem-style encoder for the output (running digital from box-to-box -- "analog" scrambling (Time or Frequency domain) is way too easy to break) so maybe another $50 DSP chip... after all, you don't need to support 30 different baud rates, just one data rate with perhaps a low-line-quality backoff. The connectors and the box are probably the major recurring cost (the chip prices will go way down in quantity.) Am I missing anything? The technology level of the Newton seems to be a bit of overkill (unless you actually want that kind of user interface.) _Mark_ <eichin@cygnus.com> <eichin@athena.mit.edu> : note that this is an unsigned message.
Let's see -- an ISDN-quality (quality? I use the term loosely) codec should be under $50 single quantity, the data rate isn't very high so you don't need much of a CPU (6811 might even be enough, and they're easy to interface to things -- lots of on-chip I/O). You'd need a modem-style encoder for the output (running digital from box-to-box -- "analog" scrambling (Time or Frequency domain) is way too easy to break) so maybe another $50 DSP chip... after all, you don't need to support 30 different baud rates, just one data rate with perhaps a low-line-quality backoff. The codec is pretty cheap, but you want a nice low bit rate so you can send the encrypted data over the phone. Some talk of how to do this is happening on sci.crypt. I have a scheme which I plan to float around at the cypherpunks meeting on Saturday. However, the hardware ends up being on the expensive side ($150 or so).
participants (4)
-
Eric Hollander -
George A. Gleason -
Mark W. Eichin -
phr＠napa.Telebit.COM