Dr. Denning Speaks

Concerns/Objections Presented at NIST Review, June 2-4, 1993

           Compiled by Dorothy Denning, June 3, 1993

The following is a summary of some of the major concerns and
objections that I heard expressed at the NIST Review. My listing
of these claims does not imply that I endorse them or accept
them as "truths," I have not attributed the claims to specific
individuals since many were expressed in some form or another by
more than one person.

Review Process

The review time is too short. More time is needed to formulate a
good solution that meets everyone's needs. The issues are very
complex and there are many questions. It is not even clear
exactly what is the "problem" that is being solved. The
government appears to be "end- running" the process.

The government should work with industry toward a policy that is
acceptable to both.  Industry should be involved in the review.
The process should be open and not secret. The cold war is over
and such secrecy cannot be justified. There is more openess and
industry/government collaboration in other countries, which will
allow them to find workable solutions.

The secret process has led to fear, uncertainty, and doubt. It
does not inspire trust in the people or in industry. Knowledge
is a necessary basis for trust.

There is no reason to race to a solution given that there has
been no significant law enforcement problem so far.

Impact on Industry and Economy

The initiative is bad for business and for the economy though
hard to measure costs.

Foreign markets will reject since algorithm is classified and
not subject to public scrutiny.  Some domestic users will reject
for same reason.

The electronic infrastructure is global. We need international
standards and tools and technology that can be used everywhere.
This requires international standards. There are already
international standards for non-Clipper cryptography.
Clipper/Capstone is not compatible with the huge installed base
that has gained international acceptance. Fragmenting the net by
nationalities would be bad since the economy and market is

U.S. companies will be at a disadvantage if Clipper is mandatory
since they'd have to design around a more costly technology.
Since there would be no significant foreign demand, foreign
companies will have an advantage since they can market DES or
other high-grade security products.

It would be more cost effective to give the FBI an extra $80
million to handle investigations by means other than wnretapping
Criminals will find some way to subvert the law enforcement
field or encrypt with a non-escrow based system anyway, so the
key-escrow program is not worth it.

Businesses will want to escrow their keys with agents of their
choice so that they can get the keys if needed. The government
holding the keys does not solve their problem.

Impact of Export Laws on Industry

Export restrictions on encryption software seriously harms the
software industry and should be lifted. There are more foreign
than domestic encryption programs and products. These products
use the DES, RSA, and other strong methods. Foreign vendors are
responding to a market for software-only encryption solutions.
Foreign companies will buy (and are buying) foreign products if
they cannot buy U.S. ones, resulting in U.S.companies losing
substantial market share (they currently have 75%). Domestic and
foreign companies are sophisticated in their knowledge about
encryption and they want strong encryption.

Encryption is becoming an integral part of many software systems
(for authentication, access control, data privacy, signatures,
proprietary software protection, etc.) and not just a separate
function..Companies are migrating to distributed computing
environments where encryption must be an integral component.
Thus, if export policies are not liberalized, the entire
software industry is at risk. The software industry is a major
industry and growing, contributing $36 billion to economy.
Computer software and hardware industry is second largest
manufacturing industry. Thus, it is critical to protect it.

There is serious economic loss of "leveraged sales" involving
hardware, software, and services as well as initial product

Export restrictions should be liberalized for encryption systems
that are "in the public domain" or "generally available to the
public" (mass market). Other countries that have decontrolled
"public domain" include Britain, Japan, and France.

Export controls must be dropped in order to achieve world-wide
commerce over the telecommunications infrastructure.

Clipper should be readily exportable. But liberalizing export of
Clipper chip technology will not help the software industry.

The export licensing process is costly and time consuming for
companies. Multinationals have to choose between two undesirable
situations: using expon-approved weak cryptography or going
through the licensing process.

The Clipper Hardware Solution

The necessity to keep the algorithm classified will result in
more costly hardware.

A software solution is needed. It would be cheaper and more
versatile. It will be a long time before a hardware solution is
cheap enough that it could be built-into a PC as a standard

There is a problem trying to come up with a world-wide hardware
solution that is acceptable, given that it is classified. A
non-classified solution is needed to achieve acceptance.

There is less trust in a classified Skipjack than in DES.
solution. People trust DES and don't want to give it up.  as
long as it remains unbroken.

 Micali's "fair public key cryptography" may be better since it
allows software as well as hardware implementation, it is
completely voluntary, and it allows multiple incarnations of key
escrow agencies.

Triple DES would be a better DES should continue to be certified

Voluntary Program

A voluntary scheme won't meet the law enforcement objectives.
But a mandatory program would be unacceptable and would severely
damage U.S. industry, so it must be voluntary.


There are serious doubts about whether the escrow agents and
process can be trusted. If the entire scheme can be cornpromised
by bribing only 3 people, there's a real possibility of a
*digital Pearl Harbor' or a *Clipper Gate.' There have been
corrupt people in the government in the past and there will be
in the future. More escrow agents might help.

The key escrow process needs a review similar to that for the
encryption algorithm.

There should be some contingency plan in case all the keys are
compromised. Even so, it would be extremely costly and time
consuming to replace all chips in order to change keys.

Giving out chip keys introduces vulnerabilities. It would be
better if the escrow agencies gave out session keys.

This is not an "escrow" arrangement in the true sense in that
the agents have no fiduciary responsibility, and are not
private. Government can't "escrow" for the government.

Legislation is needed to establish escrow agencies, set up rules
of operation, appropriate funds, etc.


Secret searches are not allowed for physical papers. By
extension, they should not be allowed for electronic papers in
the 'virtual office' that exists on top of the network
infrastructure.  People telecommute, so "papers" travel over
networks. There is a fuzzy line between searches for papers and

The 4th Amendment and wiretap law are intended to restrict the
government, not coerce the public. The key escrow scheme
presumes everyone is guilty, which is contrary to US. law.

There are holes in the warrant requirements with FISA, namely
insufficient protection for oversees conversations by Americans
as well as foreigners.

Widespread escrow raises 5th Amendment issue of waiving right to
not incriminate yourself.

A mandatory program would violate 1st Amendment rights.

Larger Policy Issue

The premise behind the initiative is that telecommunications
equipment should be designed to facilitate surveillance. This is
a bad policy since it does not protect privacy and limit
government abuse. Wiretaps are not that effective and there is
no legal or policy basis for such a premise.

To catch the few who misbehave, the entire population is put at

All content COPYRIGHT SOFTWAR (C) 2000. Any reproduction or use of content herein must be approved by SOFTWAR.