NetNews Usenet Archive 1992 #27/NN_1992_27.iso/spool/sci/crypt/4827

Internet Message Format | 1992-11-15 | 4.7 KB

Path: sparky!uunet!europa.asd.contel.com!darwin.sura.net!zaphod.mps.ohio-state.edu!cs.utexas.edu!sun-barr!news2me.EBay.Sun.COM!exodus.Eng.Sun.COM!cloudbase!jfarrell
From: jfarrell@cloudbase (Jerry Farrell)
Newsgroups: sci.crypt
Subject: Re: Limits on the Use of Cryptography?
Date: 15 Nov 1992 20:04:59 GMT
Organization: Sun Microsystems Inc., Mountain View, CA
Lines: 74
Message-ID: <[e-mail]>
References: <[e-mail]>
NNTP-Posting-Host: cloudbase
X-Newsreader: Tin 1.1 PL4

First, let me express my awe and admiration for the patient persistence
Terry Ritter has shown in returning this discussion to its topic (or
at least to the topic *he* proposed), whenever it veered toward one
of the surrounding infernos.  Thank you, Terry.

Now, to return myself to the topic:  It seems to me there are three
usable arguments that have come up so far, of varying applicability.

1) Futility.  The means to strong encryption are readily available from
numerous sources within and without the country.  Pre-registration would
not prevent the use of secure channels by parties willing to break the
law; quite possibly, it would only be detectable by an abuse on the part
of snooping authorities.  Post-facto demands for a key could be met by
stonewalling or legitimate ignorance of a session key; a simple "I
don't rmember" sufficed for James R. Hoffa, and would likely reappear
as needed.  And the penalty for failure to provide a key on (warranted)
demand would very likely be less than the penalty for the hidden crime;
we could expect some people to accept jail for contempt (or whatever)
in lieu of revealing the keys LE would *really* like to see.  There
seems to be *some* force to an argument that making unenforceable laws
is a waste of effort and tax dollars (although there are plenty of
counterexamples).

2) Potential for abuse.  There is sufficient history of people in
positions of trust abusing their powers to give this some weight.  Examples
include tax records under the Nixon administration, the State department
records of the Clintons, mere et fils, and various reported leaks of
information from the NCIC to private investigators.  At lower levels
of government, the record seems worse.  At the extreme, we have the case
of Dutch population records falling into the hands of the Nazis.

All these examples argue primarily against pre-registration.  It seems to
me opposition to post-facto demands would be stronger if based on (1) or
(3).

3) Restraint of progress.
 a. Much of the most interesting research in the use of cryptography
    involves one-time or session keys.  The user is hardly expected to know,
    far less retain these keys.  A requirement to produce such keys on
    demand effectively prohibits protocols where that is impractical.
    This in turn severely restricts the use of electronic media in
    domains where privacy, authentication, and accountability are critical.
 b. The research mentioned in (a) has immediate application in the
    economic life of this country.  Making it impossible (or even
    too painfully impractical) to use cryptography will put our
    information industries at a competetive disadvantage, open all our
    industries to espionage, and put our financial structure at risk.

I expect these points could be better expressed, but they seem to
address the arguments I have seen proposed.  I'd be very interested to
see something new.

[Note added in proof:  another argument might be constructed, parallel
to the use of time-clocks in safes to protect bank officials against
threats designed to get them to reveal the combination.  In fact, many
of the arguments in this cable of threads take on an interesting light if
you view a bank or corporation as a conspiracy to commit commerce.]

It seems to me the arguments against pre-registration are quite strong,
perhaps even strong enough to sway a drug-crazed congress.  I am less
certain about post-facto demands for revelation.  If it could be arranged
not to obstruct the use of transient keys, I might even be persuaded that
the power to compel keys might be tolerable, *as long as it does not
imply the necessity of retaining keys simply to respond to such a demand.*
That is, as long as it is a legitimate response to a warrant for my keys
to say "I don't know them; I threw them all away as soon as the session
ended" or "I never knew them", I'm not sure I'd object to such a warrant
issuing (except as a taxpayer concerned about the waste of my dollars on
on futile pursuits).  My suspicion is that arguments of the form (3a)
would be most effective in securing this middle ground.  This is vaguely
troublesome: must we stoop to bashing NEC or Peugeot or the gnomes of
Zurich to get our government to Do the Right Thing?

jf