Re: NANO: Hacking assembler security

From: hal@finney.org
Date: Thu Feb 10 2000 - 11:04:59 MST


Eliezer S. Yudkowsky, <sentience@pobox.com>, writes:
> What's more likely is that if Zyvex wants to sell more than one Anything
> Box, their Boxes will be programmed to build only those objects which
> have been signed with by Zyvex's public RSA key; which, of course, will
> not include Anything Boxes. Or submachine guns.

Following up on the crypto angle, there's also a technology called
"group" or "threshold" signatures, which requires a group of people
(or some subset) to work together to create a signature. Years ago I
suggested the ultimate democracy, Anything Boxes which would only create
objects which were signed by a threshold signature meaning that (say)
80% of the human race approved that design.

I think Drexler first proposed the idea of the "limited assembler" back
in Engines, something which would make consumer and industrial devices
but wouldn't make super-dangerous things. (Obviously even something as
simple as a baseball bat can be used to club someone, unless it's a pretty
damn smart bat.) Along with that he proposed a centimeter-sized isolation
lab where you could build anything you want, but it couldn't get out. This
is how new designs would be tested and approved.

> (As modified to take into account the "nanotech -> quantum computing ->
> no more RSA" chain. Open (?) question: Using quantum computing, is it
> possible to create a public-key encryption method immune to quantum decryption?)

First of all I am skeptical that it will ever be practical to build
quantum computers able to crack large RSA keys, because of environmental
noise.

However if they do get built, there are some public key schemes which are
not vulnerable to their attacks. These are relatively obscure, involving
coding theory or CAs, and haven't been examined very thoroughly. They
tend to have disadvantages of having very large keys and/or signatures.
But if quantum computers destroy the factoring and discrete-log schemes
that are widely used today, these other methods would hopefully be able
to take over.

Hal



This archive was generated by hypermail 2b29 : Thu Jul 27 2000 - 14:03:38 MDT