Re: impossibility of computer security?

From: Anders Sandberg (asa@nada.kth.se)
Date: Tue Sep 17 2002 - 08:27:38 MDT


In some sense there is no solution to the "security problem". A security
system is (if you idealize it) a program that reads an arbitrary input
and then does (or rather allows) something based on it (like opens a
lock, launches a missile or compiles a nanodesign). Essentially it
produces a single bit, "allowed"/"disallowed". Gödel and Turing
immediately cause problems: this is quite similar to a theorem proving
system. Either you have a system that is complete, that is that it will
give the right allow/disallow response to every legal input - but the
set of allowed responses has to be finite. Or you have a more flexible
system that allows responses to arbitrarily complex inputs (like
suggested nanodesigns, cryptokeys or programs to run) and it will either
be inconsistent (i.e. bad security) or incomplete (i.e. some inputs
causes it to freeze).

-- 
-----------------------------------------------------------------------
Anders Sandberg                                      Towards Ascension!
asa@nada.kth.se                            http://www.nada.kth.se/~asa/
GCS/M/S/O d++ -p+ c++++ !l u+ e++ m++ s+/+ n--- h+/* f+ g+ w++ t+ r+ !y


This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:17:06 MST