[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Crypto Controls in the USA
Data Encryption Software
and
Technical Data Controls
in the
United States of America
Copyright (C) 1994 Michael Paul Johnson
PO Box 1151
Longmont CO 80502-1151
USA
[email protected]
303-447-7302
Distribute unmodified copies freely.
7 January 1994
1. Introduction
2. Problems with the Status Quo
2.1. Regulations Ignore Technology
2.2. Overly Broad Definition of "Encryption Software"
2.3. Overly Broad Definition of Export
2.4. Censorship and the First Amendment
3. National Security Issues
3.1. Signals Intelligence
3.1.1. Enemy Signals
3.1.2. Bugs & National Technical Means
3.2. Counter-Intelligence Activities
3.3. Our Military and Diplomatic Communications
3.4. Banking Transactions
3.5. Domestic Personal and Corporate Communications
3.6. Authentication in the Private Sector
3.7. Upholding the Constitution
3.8. Law Enforcement
3.9. Technology Base Migration and Loss
4. Technology Issues
4.1. Availability of Computers
4.2. Telephone Lines and Modems
4.3. The Internet
4.4. Information Services and Bulletin Boards
4.5. Books and Magazines
4.6. Availability of Encryption Software
4.7. DES is Dying
4.8. Unbreakable Encryption
5. Economic Issues
5.1. International Trade
5.2. Cryptographic Competition
5.3. Domestic Chilling Effect
6. Regulatory Issues
6.1. Clarity of Regulations and their Intent
6.2. The First Amendment
6.3. Enforcement
6.4. Consistency with Technology
7. Recommendations
7.1. Reevaluate National Security Impact
7.2. Deregulate Publicly Available Information
7.3. Deregulate Research and Publication
7.4. Replace DES with Better Public Standard
7.5. Control NSAs Cryptographic Technology
7.6. Alternate Intelligence Methods
7.7. Alternate Law Enforcement Methods
7.8. Clarify & Repair Export Regulations
1. Introduction
The current status of the regulation of encryption software in
the United States of America is, at best, confusing and harmful
to business. At worst, the current status is harmful to
National Security and violates U. S. Constitution. I invite you
to study this issue with me. I present what I perceive to be
the problems and the issues that must be considered, then
suggest some solutions. Even if you don't fully agree with all
of my statements, I hope that they prove helpful to your own
understanding of this situation.
2. Problems with the Status Quo
I perceive several problems with the current International
Traffic in Arms Regulations (ITAR) far beyond typos like the
reference to $120.10(d), which doesn't exist, in $120.10(1).
These problems are severe enough that I hope that they will be
rectified soon before they do even more damage. All of the
problems with the ITAR mentioned here have to with encryption
software, as defined in the ITAR.
2.1 Regulations Ignore Technology
The ITAR ignores the fact that software, like other technical
data, can exist in a multitude of forms, many of which know no
national boundaries. The ITAR ignores the fact that much of
what is prohibited to be exported exists in unlimited quantities
outside the USA. The ITAR hurts U. S. Business but doesn't
significantly reduce the availability of strong encryption
technology outside the USA. The ITAR ignores the widespread use
of purely electronic means to distribute software, such as the
Internet, Computer Bulletin Board Systems (BBS), and commercial
information services (such as CompuServe). The ITAR ignores the
fact that shareware publishing, which is a form of
Constitutionally protected publication, propagates software all
over the world with no formal distribution mechanism.
2.2 Overly Broad Definition of "Encryption Software"
"Encryption software" is defined in $121.8(f) and $121.1,
Category XIII(b) to include not only computer programs designed
to protect the privacy of information, but all of the technical
data about those programs. This naturally includes a great deal
of material in any large library or book store. William B.
Robinson, Director of the United States Department of State,
Bureau of Politico-Military Affairs, Office of Defense Trade
Controls, states in his letter to me of 30 November 1993, that
"the exemptions listed in $125.4 for technical data do not apply
to cryptographic software and source code." I conclude,
therefore, that the ITAR implies that the majority of the
libraries and larger bookstores in the United States stock
"munitions" on their shelves for anyone to read.
2.3 Overly Broad Definition of Export
$120.17 of the ITAR makes it clear that allowing a foreign
person to read a book containing encryption software constitutes
export. Therefore it seems possible that some perverse person
might state that all of the libraries and bookstores that
contain any book on cryptography must register as an exporter of
munitions. This situation gets even more interesting in its
electronic analogies. However, restricting domestic
distribution of technology that is perfectly legal and useful
within the USA just because a foreigner might see it is not only
unreasonable, it could probably not stand a Constitutional
challenge.
2.4 Censorship and the First Amendment
The ITAR does make some acknowledgment of the fact that not all
of the publications that it calls "encryption software" need be
subject to export restrictions, but doesn't even come close to
defining the difference. All it does is set forth a censorship
procedure called a "Commodity Jurisdiction Procedure" (see
$120.4).
From what I know of the First Amendment and Constitutional case
law (I'm not a lawyer, but I took a class on the subject), the
only way the Federal Government can legally take away U. S.
Citizen's rights to freedom of speech or freedom of the press is
when there is a clear danger that is caused by that expression,
or a significant infringement of the rights of another person.
The classic examples of this are yelling "FIRE" in a crowded
theater, or committing libel or slander. In the case of
technical data concerning encryption software that is already in
the "public domain" (as defined in the ITAR for technical data),
the damage (or benefit), if any, is pretty much already done and
further publication probably makes little difference. I believe
that any definition of what is a munition that makes the
nation's bookstores and libraries appear to be exporters of
munitions is not just ridiculous, it is unconstitutional.
When I tried to get clarification from the Department of State
on what the rules that they applied when performing their
censorship role (Commodity Jurisdiction Proceeding) were, all I
got back was two letters, one that clarified a point made muddy
by a typo in the ITAR and gave no help beyond the ITAR itself,
and one form letter that said that the Department of State would
not deal with hypothetical questions (even though most of my
questions weren't purely hypothetical).
This serious Constitutional question casts doubt on the
enforceability of any of the regulations in the ITAR against any
encryption software. It could be difficult to prove that the
publication of a particular piece of technical data or computer
program caused specific, measurable damage resulting from
intentional export without a license (even if you could figure
out who exported it). Yet, there cannot be any restriction to
U. S. Citizen's freedom of speech and freedom of the press
unless it can be proved that damage resulted from that speech.
3. National Security Issues
"National Security" means a lot of things. It means maintaining
the integrity and safety of our Constitution, our people, our
land, and our environment. It means the ability to defend
ourselves against anyone or any thing that would seek to harm
us. Our freedom, constitutional democracy, and fairness to all
citizens are our greatest protection against internal threats.
This gives us the strength and will to have a strong diplomatic,
economic, and military force to protect us against external
threats.
3.1 Signals Intelligence
In the context of encryption software, the most obvious
connection to National Security (if you ask the NSA) is the
impact on intelligence operations. In the process of spying on
enemies, it is a lot safer to listen to what they are doing
remotely than to send a person in to spy. The two main ways of
doing this are (1) to listen to and/or alter signals that they
generate for their own purposes, and (2) to listen to signals
emanating from devices that we have placed for the purposes of
listening.
3.1.1 Enemy Signals
Enemy signals may include telemetry, radio transmissions on
various frequencies for various purposes, telephone
conversations, computer data links of various sorts, etc. These
all may provide some kind of clue as to what evil deeds they may
try to perpetrate on us next, or may indicate significant
vulnerabilities for us to exploit in war time. The enemy knows
that we know this, and will probably try to protect at least
some of their signals using encryption, deception, jamming, or
data hiding (steganography).
It is possible that an enemy might use some of our own
encryption technology against us. The enemy may either directly
use a commercial product to hide the meaning of communications
from us, or use some published technology originated in the USA
and other free countries to build their own systems. They may
also add their own secret innovations to what they learn from
us. Of course, there is also the consideration that an enemy
would prefer to use cryptographic technology of their own
design. This would give them the advantage of not letting us
know which algorithm they are using. It would also deprive us
of the huge head start we have on cryptanalysis of things like
the ancient Data Encryption Standard (DES). This may not be
enough to stop me from protecting a proprietary cookie recipe
with the DES algorithm (or the triple DES variant if the cookies
tasted good and weren't fattening), but it would be a
significant consideration for a nation planning to bomb Pearl
Harbor. DES is probably a bad example, since everyone on planet
Earth who really cares already has a copy of a program that does
DES encryption, or can get one in a few minutes.
Using a commercial product like a spread sheet or data base
program that does encryption only as an extra feature against us
is something of a problem for an enemy, since such products are
not normally well suited to the applications needed in military
and diplomatic situations. Imagine giving a field commander a
laptop computer with a U. S. commercial spread sheet program on
it to decrypt orders from his commander. I may underestimate
the silliness of our enemies, but I don't think that this is
likely. A much more tamper-resistant device with better key
management would be much more appropriate for a military or
diplomatic application. Use of our publicly available
encryption design technical data in building more appropriate
military communications security devices is a more likely threat
in the case of a clever adversary. The only consolation in this
case is that we also have access to this same data as an aid to
cryptanalysis.
In the extreme case, strong cryptographic technology could
become so readily available and easy to use that most of the
interesting signals generated by enemies for their own purposes
are encrypted in such a way that we cannot decrypt or subvert
the communications without stealing their keys. In that case,
all nations might have to behave like gentlemen (and not open
the other's mail or read their electronic communications). Then
again, that is probably too idealistic to expect. It is more
likely that mankind will only figure out other ways of spying on
each other.
3.1.2 Bugs & National Technical Means
Even if the enemy takes great care to protect the secrecy and
integrity of their own communications channels, we can still spy
on them. Listening devices can be made so small and have such
inconspicuous output that they can be almost impossible to
detect or jam when planted properly. It takes very little power
to send a signal to a nearby relay to a satellite, and many
varieties of listening devices can be used. Even if an enemy
becomes wise to one kind, another kind may be in use. Suffice
it to say that all the encryption technology in the world could
not cut off this source of intelligence, since all valuable
intelligence exists in the clear at some point. If it didn't,
it would be of no value to the originator and intended
recipient.
Public use of strong cryptographic technology may limit the
points where listening devices must be planted to be of value,
but can never totally cut off this sort of intelligence.
Increases in knowledge cryptography and steganography may help
this sort of spying more than hinder it.
3.2 Counter-Intelligence Activities
Increased public use of strong cryptography makes it easier for
a spy to obtain a good cryptosystem. It also makes it easier to
send encrypted messages without arousing suspicion. That is
good for our spies, but bad for detecting spies in our own
country. Then again, it would be a pretty inept spy (ours or
theirs) who could not now obtain a good cryptosystem and send
messages home without arousing suspicion, under conditions much
worse than the USA right now. Of course, increased public use
of strong cryptography also makes it harder for a spy to find
valuable data to send back home. I think that the net effect
will be that spies in the USA (and some other developed nations)
will be harder to catch, but less effective.
3.3 Our Military and Diplomatic Communications
The greatest contribution of cryptography to our National
Security is in protecting our own military and diplomatic
communications from eavesdropping or alteration. Communications
of this nature must be private, must be authentic (not an
alteration or forgery), and must not have been altered in
transit. Increased public use of strong cryptography can only
help us to keep our most sensitive communications private. This
is because there will be more encrypted traffic to attempt
attacks on, making traffic analysis harder. It also may be that
discoveries made in the private sector help in the design and
evaluation of military and diplomatic cryptosystems.
3.4 Banking Transactions
We do so much banking electronically that failure to use strong
cryptography to protect these transactions would be criminally
negligent. It would be like not locking the vault and bank
doors and not posting a guard. The importance of the integrity
of our banking system to our economic well-being is obvious.
The cryptographic protection must also be economical, just as
the bank buildings, vaults, and other security systems must be,
or the banks will not remain competitive. We must balance the
cost of protection with the value of what is being protected.
Strong cryptography usually doesn't cost much more to implement
than weak cryptography, and may save a whole lot of money if it
can prevent some fraud.
3.5 Domestic Personal and Corporate Communications
Although there are strict and fairly consistent guidelines for
the protection of U. S. Government classified information, the
private sector is much more vulnerable. Some companies are very
security conscious, but some are not. Those which are not are
easy targets for foreign and domestic spies, either working for
governments or competing corporations (or both). Encouraging
good security practices in the private sector, including use of
strong cryptography, use of good crosscut shredders, etc., makes
the USA more secure against this threat.
Protection of personal communications with encryption is good
for privacy, just as locks on doors and curtains on windows are.
It becomes very important in some cases, such as when a
battered person is hiding from a stalker that is still at large,
or when coordinating activities that might attract criminals
like shipping diamonds. Encryption technology can help reduce
crime, just like dead bolt locks. Just as I prefer to manage my
own dead bolt keys, I'd rather not be forced to escrow a master
key to my data with Big Brother. This isn't because I do
anything evil with my dead bolts or cryptographic software, but
because I love freedom. This preference is nearly universal
among users of cryptography, and the countries and companies
that cater to this desire will have a big economic advantage.
3.6 Authentication in the Private Sector
Encryption technology is the only way to provide a signature on
a digital document. Nothing is totally fool proof, but digital
signatures, when done properly, are much harder to forge or
refute than pen and ink signatures on paper. Electronic
documents can be transmitted faster and with higher fidelity
than faxes, and the ability to sign them will be a great aid to
quickly and conveniently doing business with remote customers
and suppliers. As contract case law and technology evolve, this
will become more and more important to our economy.
3.7 Upholding the Constitution
Citizens of the United States of America have a right to privacy
guaranteed by the Constitution's Bill of Rights. This quaintly
stated right to be secure in our papers and effects is highly
cherished. The advance of technology has eroded privacy.
Corporations like Tandy openly track their customer's names,
addresses, buying habits, then shower them with junk mail.
Credit bureaus keep massive amounts of (often incorrect) data on
people all over the country -- information that is supplied to
lenders and in the form of prescreened mailing lists for
solicitors. Government organizations keep records of real
estate transactions, census data, and other such records that
are used by solicitors to pester owners of houses in selected
neighborhoods. Hospitals keep your patient records on computer
systems that can be accessed by many people. Cellular and
cordless telephones are trivial to monitor without physically
tapping any wires, and legislated privacy in these areas is
unenforceable.
Strong encryption can bring back part of the privacy that has
been lost to technology. No law can keep spies and criminals
from listening to phone calls made over radio links (including
microwave and satellite links for normal phone calls), but
encryption can make those calls unintelligible to criminals and
other unauthorized listeners.
3.8 Law Enforcement
The proper use of encryption technology by law enforcement
officers helps deny knowledge of monitoring operations to
criminals and fugitives. It helps them to keep records private
and protect under cover agents. It helps prevent tampering and
deception from being used against them in their own
communications. Unfortunately, this is a two-edged sword.
Strong encryption technology can also be used by criminals to
thwart the efforts of law enforcement officers to gather useful
information from court authorized wire taps.
Strong cryptography also provides a "safe" way for a criminal to
keep records of nefarious deeds that cannot be read by the
police and used as convincing evidence leading to a conviction.
Of course, fewer such records might be kept in the absence of
strong cryptography, and some records kept in this manner might
not be all that useful in obtaining a conviction. This is not
very assuring to law abiding citizens and law enforcement
officers, who want dangerous criminals to be caught well before
they meet the Ultimate Judge in Heaven. Fortunately, most of
the investigative tools available to law enforcement officials
are not affected by strong cryptography. It is also likely that
anyone stupid enough to engage in criminal activity is likely to
screw up in some way that leaks information about their actions.
Murder, terrorism, rape, and other violent crimes are not all
that hard to commit (for those devoid of conscience or with the
twisted conscience of a kamikaze), but these crimes are very
difficult to get away with.
3.9 Technology Base Migration and Loss
When a technology is discouraged by over-regulation, taxation,
or other means, that technology becomes less profitable in the
country where it is discouraged. Less profitable technologies
are not invested in as heavily. Therefore, the technology in
that country will tend to fall behind. Right now, it appears
more profitable to develop an encryption product for sale in
many other countries than in the USA because export of this
technology from the USA is discouraged but import is not. An
entrepreneur in New Zealand has an unfair advantage against one
in the USA. The New Zealander is not required to cripple key
lengths or deal with unreasonable and unreadable regulations
like our ITAR. This means that encryption technology in the USA
will tend to atrophy while it prospers in other countries. This
is bad for National Security.
4. Technology Issues
Any policy concerning encryption software that is to make sense
must take into account the realities of the current state of the
art in the applicable technologies. Failure to do so could at
best lead to confusion, and at worst do much more harm than
good.
4.1 Availability of Computers
It doesn't take a lot of computing power to perform strong
encryption (locking data up). It often takes a great deal of
computing power to do serious cryptanalysis (unlocking data
without the key). Strong encryption can be done with almost any
microprocessor on today's market. The original IBM PC (now
greatly outclassed by the current desktop computers) has more
than enough computing power to lock up significant amounts of
data so tight that all the spy organizations in the world
combined could not unlock it for thousands of years or more.
This class of computer is available in essentially any developed
or semi-developed country in the world.
4.2 Telephone Lines and Modems
There are still places in the world that don't have easy access
to telephone lines, but they are growing fewer all the time.
The places that do have telephones, computers, and modems are
those places where encryption technology is the most useful. Be
they friend or foe, these places all have one thing in common.
They are only a telephone call or two away from strong
cryptographic software if they know where to call, and it isn't
that hard to find out. Since many telephone connections are by
satellite, and since international telephone traffic is not
routinely monitored and censored by most free nations, any
technical data (including encryption software) can be
transmitted across almost any national border unhindered and
undetected.
4.3 The Internet
The Internet has grown to such a large, international collection
of high speed data paths between computers, that it has become,
among other things, one of the most effective examples of
international freedom of expression in existence. Physical
distances and political boundaries become irrelevant. I can
peruse data posted for public access on university and corporate
computer systems on five continents and many islands, no matter
if I'm in the USA or in Russia. This is a powerful research
tool. News groups provide discussion forums for subjects
technical and nontechnical, decent and obscene, conservative and
liberal, learned and ignorant, from Animal husbandry to Zymurgy,
and more. The Internet provides easy access to lots of strong
cryptographic technology and software that can be reached from
any nation with a connection to the Internet. A great deal of
this data originated from outside the USA.
The most complete and up to date collections of encryption
software on the Internet are published for anonymous ftp from
sites outside the USA. (Anonymous ftp sites are computer
systems that allow anyone to log in with the name "anonymous"
using the file transfer protocol program called "ftp" to
transfer files to their own system). There are several ftp
sites in the USA that carry some encryption software, and they
have varying degrees of barriers to export. Some sites make no
attempt at all to limit access to encryption software. Some
sites are very effective at not allowing export, but are totally
ineffective at distributing software domestically because of the
hassles they impose on users (who can just as easily get the
same stuff from Italy).
The strongest barrier to export that I've seen used at a U. S.
domestic ftp site for encryption software that doesn't totally
defeat most of the advantages of this form of software
distribution is the one used at rsa.com for the distribution of
their RSAREF package and RIPEM. The idea is to force you to
read a text file containing an anti-export warning before you
can find the data you are after. The text file that contains
the warning also contains the name of a hidden directory that
changes periodically. The encryption software is in the hidden
directory. Naturally, this doesn't prevent an unwelcome
intruder from stealing the data anyway, but the moral barrier
presented probably reduces the number of "exports" from that
site initiated by people in other countries. I support RSA Data
Security, Incorporated's right to publish this data, even though
I have observed copies of this data on several foreign computer
systems.
I tried hard to think of a better solution (and even called the
Department of State and the NSA for ideas), but there is
basically no way to widely and freely publish any data in the
USA without making it possible for a foreigner to steal that
data out of the country. Even if the data is confined to
physical packages and sold or placed in libraries only in the
USA, there is nothing to prevent someone (either a U. S. or
foreign citizen) from buying or borrowing a copy, then
transmitting a copy of that copy out of the country. Even if
positive proof of citizenship is required before release of the
data, all it takes is one citizen to release a copy of the data
outside the USA. You might argue that there would be a strong
moral barrier against this, but remember that all it takes is
one. What does it matter to someone if they send a copy of
encryption software to a friend or relative in another country
so that they can send private electronic mail back and forth?
All it takes is one copy out of the country, and that copy can
be copied any number of times. If rabbits multiplied so easily,
we would all quickly drown in them.
The bottom line is that the best solution to balancing freedom
of the press and the ITAR for encryption software ftp sites is
just an annoyance for the intended users and a way to make it
impossible to prove that the operators of the site intended to
break any valid law. This may or may not have any bearing on
the proliferation of encryption technology outside of the USA.
I am not a lawyer, but I know RSA Data Security, Incorporated,
has lots of them, and I don't believe they would do anything
stupid.
4.4 Information Services and Bulletin Boards
CompuServe, America Online, Genie, Bix, Delphi, and other
similar services offer massive amounts of data, including
encryption software and technical data, to callers. They often
act as common carriers between correspondents who carry this
data themselves, and really don't know the contents of what they
are carrying. Other times, they are well aware of what they
have. For example, CompuServe publishes a magazine promoting
some of the shareware that they carry, and featured some
encryption software in an article in their November 1993 issue.
These information services also serve customers outside of the
USA. Indeed, it would be very difficult not to do so, even if
they didn't want to bring some foreign money into their hands.
Computer bulletin board systems vary in size from hobby systems
running on a single PC in a home to large commercial systems.
Some are run as a hobby, some as a means of providing technical
support to customers, and some as profit-making information
services. A very large number of these systems have encryption
software on them with no export controls expressed, implied, or
implemented. Indeed, many of the operators of these systems
would laugh in your face if you claimed they were trafficking in
arms. These systems are normally accessible from anywhere with
a telephone, computer, and modem.
4.5 Books and Magazines
Encryption software and technical data about it can be found in
a large number of books and magazines in libraries, book stores,
and by subscription in and out of the USA. Some of these have
companion disks that can be ordered separately or that are bound
in the back of the book. Some have associated postings on an
information service. Some have printed computer program source
code listings in them. In those rare cases where the book and
disk sets are not distributed by the publisher outside the USA,
it is almost certain that the books and disks will appear
outside the USA, because most book stores don't restrict their
sales to U. S. Citizens. Indeed, to do so sounds rather fascist
and unamerican: "Let me see your citizenship papers before you
buy a book!" This country is both more pleasant and a lot more
secure without such nonsense.
4.6 Availability of Encryption Software
There is already a large number of free or very inexpensive
software packages available internationally from various
information services, computer bulletin boards, Internet ftp
sites, and commercial packages available off the shelf. These
include:
o Many DES implementations originating from many
countries.
o Several packages that implement the Swiss IDEA cipher.
o Several packages that directly implement triple-DES.
o Assorted implementations of published algorithms, some
of which probably exceed DES in strength.
o Assorted programs (such as utility packages, spread
sheets, database programs, and word processors) that
include some form of encryption that is incidental to
their main function. The security of the encryption
varies from so poor that it should be called false
advertising (like that used in Microsoft Word), to
probably good against all but professional cryptanalysts
(like PKZIP), to fairly decent implementations of DES or
better.
o Numerous proprietary algorithms, many of which probably
claim greater security than they merit, but some of which
may be very good.
o A few encryption packages that effectively use a
combination of the RSA public key encryption algorithm and
a block cipher (DES, triple DES, or IDEA) to encrypt
electronic mail.
o Several cryptographer's tool kits that implement large
integer arithmetic over finite fields, fast DES, IDEA,
and RSA implementations, and other data that facilitates
including these functions in other programs.
There are also a few cryptanalytical programs floating
around internationally to assist in cracking insecure
cryptosystems like the password protected files of Microsoft
Word and WordPerfect. In most cases, this software encryption
and cryptanalytical software cannot ever be eradicated (even if
you think it should be), because there are so many copies held
by people who think that this software is a Good Thing. Any one
copy can be copied again as much as desired. Hiding software is
much easier than hiding elephants.
The bottom line is that the cat is out of the bag, so to speak,
and no amount of regulation can ever put the cat and all its
millions of kittens back in again.
4.7 DES is Dying
DES was doomed to a limited lifetime from the beginning by
limiting its key length to 56 bits. This was probably done
intentionally, since there was much opposition to this decision
at the time. It is also possible that this key length may have
been an indication from the NSA that because of differential
cryptanalysis, the strength of the algorithm didn't justify a
larger key. Now a paper has been published that shows how DES
can be cracked for an amount of money that is within the budgets
of many nations and corporations (Efficient DES Key Search, by
Michael J. Wiener, 20 August 1993). Schematic diagrams of
showing how to build a device to accomplish this task are
included in the paper, which has been distributed
internationally electronically. I would be very surprised if
one or more of the world's major intelligence gathering
organizations had not already built DES cracking machines of
greater sophistication than Michael Wiener's. The only reason
that I say that DES is not totally dead is that it is still
useful in some cases, for the same reason that physical locks
that can be picked with a pocket knife or credit card in a
matter of seconds are still sold and used. DES encryption does
help keep unauthorized, honest, ladies and gentlemen out of your
proprietary and personal data. When used in its triple DES
variant, it might even keep dishonest people with big budgets
and lots of motivation out of your private data.
4.8 Unbreakable Encryption
One very well known algorithm (called the One Time Pad), when
properly used (i. e. with truly random keys used only once), can
never be broken by anyone, no matter what their computing power.
The One Time Pad has been known to the general public for many
years, but it has not caused the end of the free world. I've
never heard of a case of it being used for any criminal activity
except for spying (and there, I suppose, the use by "us" and
"them" somehow balances out). The One Time Pad is still used to
protect our most sensitive diplomatic communications. An
implementation of the One Time Pad in software is trivial, as
the following complete, non-hypothetical, Pascal program
demonstrates:
program one_pad;
uses dos;
var infile, keyfile, outfile: file of byte;
plain, key, cipher: byte;
begin
if paramcount < 3 then
begin
writeln('Usage: one_pad infile keyfile outfile')
end
else
begin
assign(infile, paramstr(1));
reset(infile);
assign(keyfile, paramstr(2));
reset(keyfile);
assign(outfile, paramstr(3));
rewrite(outfile);
while (not eof(infile)) and (not eof(keyfile)) do
begin
read(infile, plain);
read(keyfile, key);
{The following single line does the encryption/decryption.}
cipher := plain xor key;
write(outfile, cipher);
end;
close(outfile);
close(infile);
close(keyfile);
end
end.
The whole One Time Pad program is short enough to be written
from memory (for an experienced programmer, anyway). (For
instructions on using the above program, see your local library
or check out the sci.crypt Frequently Asked Questions document
on the Internet.) It could be argued that the trivial program
above isn't a complete encryption system, since it doesn't do
any key management.
Ladies and gentlemen, does this document contain a weapon of war
or other munition, or is it just free exercise of the author's
freedom of the press? Would the ITAR prohibit the export of
this document or not? I claim that the U. S. Constitution
specifically allows me to publish this document, no matter what
the ITAR says.
5. Economic Issues
While it seems clear that it is impossible to exercise our right
to freely publish encryption technical data and software in the
USA and at the same time prevent its export, it is very easy to
economically damage the USA with encryption export controls.
5.1 International Trade
It seems that the only encryption software that can be legally
exported for profit from the USA is either (1) crippled to
provide weak security (i. e. only a 40 bit key with RC-2 or
RC-4), (2) limited in function to certain purposes that do not
cover all market needs, or (3) limited in distribution to a
limited market. Therefore, encryption software export is not a
very lucrative field to enter. How can you compete with foreign
competitors who need not cripple their products?
5.2 Cryptographic Competition
There are sources of cryptographic software outside the USA
where the encryption software is not crippled, and is available
at a competitive price. Given a choice, the full-featured,
secure software is more likely to win. This means that other
countries will grow in this area and the USA will suffer
economically.
5.3 Domestic Chilling Effect
Export controls on encryption software discourage distribution
of strong encryption software in the USA and encourage the
weakening of domestic software to the same inadequate standards
forced upon exported software. It seems better to buy (real or
perceived) strong security from an external source than from a
domestic, persecuted supplier. Even though it would be
unconstitutional for the ITAR to disallow domestic distribution
of encryption software, few people want to be harassed by the
federal government or become a test case where the
unconstitutionality of the ITAR is conclusively proven in
court.
6. Regulatory Issues
The International Traffic in Arms Regulations are designed to
make the world a safer place by limiting the export of weapons
and military equipment. It also regulates classified or
otherwise non-public technical data about those weapons. Most
of the items regulated have a whole lot more to do with the
objective of limiting arms proliferation than encryption
software and technical data. The subject of this document,
however, is limited to a discussion of the regulation of
encryption technical data and software.
6.1 Clarity of Regulations and their Intent
For a regulation to be effective and enforceable, it must be
clear. No one should be compelled to guess what the state
requires or proscribes. Indeed, how could you be expected to
follow a law you don't understand? There should be a clear way
of telling what is and is not allowed without having to submit
an item for censorship. The intent of the regulation should
also be clear, so that a citizen could reasonably understand
what the regulation is for.
6.2 The First Amendment
The ITAR cannot override the Constitution of the United States
of America, in spite of its current claims that indicate that it
does. To the degree that it does violate the Constitution, it
is null and void. Any limitation on the freedom of speech and
freedom of the press of U. S. Citizens must be clearly linked
with a severe danger or denial of rights to another person that
can be proven in court. Worse things than encryption software
have been upheld in court as Constitutionally protected
expression.
When balancing defense and intelligence considerations with the
U. S. Constitution, it is important to remember that (1) the
whole point of defense and intelligence operations is to protect
and defend the Constitution and the people of the United States
of America, (2) the Constitution is the Supreme law of the land,
and (3) federal officials and military officers in the USA are
sworn to uphold the Constitution.
There is a theory among those involved in private sector
cryptography in the USA that there is an official or
semi-official policy of discouraging strong cryptography within
the borders of the USA, while giving the appearance of
supporting it. There is evidence to support this theory in
certain documents recently obtained under the Freedom of
Information Act by John Gillmore and released to the public.
This theory also explains a whole lot of otherwise difficult to
explain circumstances. Because such a policy, if openly stated,
would sound stupid at best and like treason against the
Constitution at worst, it is not openly stated as such. Export
control regulations and patent law appear to have been used as
tools to carry out this policy of discouraging strong
cryptography for the general public. In the event this scandal
is even partially true, then the policy must be reexamined.
This policy might not exist, but some alternate explanations for
some of the evidence is even more disturbing.
6.3 Enforcement
A regulation that cannot possibly be enforced is of questionable
value, at best. Ideally, it should be possible to detect all
violations and demonstrate beyond the shadow of a doubt to a
judge and jury that the violation was perpetrated by a specific
person or persons.
6.4 Consistency with Technology
Regulations cannot ignore technology, math and science.
Regulations cannot redefine pi to be exactly 3, repeal the law
of gravity, or stop radio waves at national boundaries. In the
same way, regulations (like the ITAR) that treat public
information like tanks, guns, and nuclear weapons make no
sense.
7. Recommendations
So far, I have pointed out problems and considerations that
cannot be satisfied concurrently. On the other hand, it is
possible to do much better than current regulations do.
7.1 Reevaluate National Security Impact
A study of the total impact of public use of strong encryption
software should be made that includes all of the considerations
mentioned above, as well as classified data concerning just how
much impact (if any) such software (which is widely available
now and projected to increase in both quality and quantity) has
on current U. S. and foreign intelligence operations.
7.2 Deregulate Publicly Available Information
Export controls on publicly available information, including
encryption software and technical data, are not only
ineffective, unenforceable, unclear, and damaging to U. S.
business interests, they are likely to be ruled unconstitutional
in any serious challenge. Deregulating this information would
help the U. S. economy, increase the use of strong encryption
software in the places where it does the most good, and have
minimal negative effects. Since so much strong encryption
technical data and software is available now, it is unclear if
any additional negative effects would even be enough to measure.
The desired effects of better security and technology in the
USA and a healthier economy would, however, be substantial.
7.3 Deregulate Research and Publication
Research and publication of scholarly work in the international,
public forums benefit the USA. The fact that this also benefits
other nations does not diminish the value to the USA. This does
not prevent the NSA from conducting classified research within
its security boundaries that is not available to the
international community. It does prevent the NSA or any other
government agency from interfering with or discouraging any work
in the field of cryptography outside its own facilities. The
NSA should maintain technological superiority by its own merit,
not by crippling all domestic competition.
7.4 Replace DES with Better Public Standard
DES is old and its key length is too short. The public wants a
more secure encryption standard that is fully public and can be
used in software implementations. The Swiss IDEA algorithm is
one likely alternative, but it would be better if an algorithm
that is royalty-free (like DES) could be made an official
standard. Clipper/Capstone key escrow is not the answer to this
need, although it might be useful within the Federal
Government.
Several possible replacements for DES have been suggested. One
that is much stronger than DES (and slightly stronger than IDEA)
and can be used royalty-free is the MPJ2 Encryption Algorithm,
which has been donated to the Public Domain by the inventor.
Technical details on this algorithm have been published, and are
available to U. S. Citizens in the USA.
7.5 Control NSA's Cryptographic Technology
While it is unreasonable to think that the general public's
cryptographic technology could possibly be confined to any one
country, it is not so difficult to control the technology in a
single organization such as the NSA. The NSA should be, with
very few exceptions, a trap door for information on cryptography
and cryptanalysis. They should strive to stay ahead of the
general public in these fields, and should not confirm or deny
what they can and cannot do to the general public without a
conscious decision by competent authority to do so (for example,
to endorse a DES replacement). In like manner, the NSA should
not discourage or encourage any cryptographic technology outside
of their walls but still inside the USA. Of course, even an
endorsement by the NSA is suspect, since their charter includes
reading other people's encrypted traffic. It would be better,
in my opinion, to preserve the NSA as a national treasure of
cryptographic expertise by dealing with public encryption
standards totally within the Department of Commerce, National
Institute of Standards and Technology (NIST).
It is probable that someone in the USA (or another country) will
independently invent something that someone inside the NSA has
invented, and that person will be honored with fame and fortune
publicly for what has already been done privately within the
NSA. This should never be construed as an excuse to censure the
public invention. Indeed, to do so would leak information about
the NSA's technology level and capabilities to the outside
world.
7.6 Alternate Intelligence Methods
To mitigate the effect of the inevitable improvement in both the
quality and availability of strong encryption software and
hardware all over the world, it would be wise to invest in
alternate intelligence methods, such as harder to detect and
easier to place bugs. Subtle long range bug delivery
mechanisms, relay devices, etc., could pay back great dividends
in intelligence value for the money for use in those cases where
strong encryption makes cryptanalysis impossible.
7.7 Alternate Law Enforcement Methods
There are many ways to catch a crook, no matter how
cryptographically sophisticated. After all, it is much easier
to plant listening devices around a suspected drug trafficker,
serial murderer, or whatever, in our own country (with a proper
search warrant) than it is to try to figure out how to bug the
command center of an enemy dictator surrounded by a loyal army.
An encrypted phone conversation may actually lull the bugged
suspect into a sense of false security, talking openly about
crimes on a secure line. An encrypted telephone does a criminal
little good if the room or car the phone is in is bugged.
7.8 Clarify & Repair Export Regulations
My specific recommendations to clarify the export regulations
with respect to encryption software, keep the encryption
technology that we use for our own military and diplomatic
communications safe, allow all reasonable commercial uses of
encryption technology in the United States, to make the
regulations much more enforceable, and to bring these
regulations into compliance with the United States of America's
Constitution follow.
$ 120.10 (1) should be altered (by removing the exception for
software defined in a nonexistent section) to read:
(1) Information which is required for the design development,
production, manufacture, assembly, operation, repair, testing,
maintenance or modification of defense articles. This includes
information in the form of blueprints, drawings, photographs,
plans, instructions and documentation.
$ 121.1, Category XIII, subcategory (b), items (1), (2) and (3),
should be modified to read:
(b) Information Security Systems and equipment, cryptographic
devices, software, and components specifically designed or
modified therefor, including:
(1) Cryptographic (including key management) systems, equipment,
assemblies, modules, integrated circuits, components or software
with the capability of maintaining secrecy or confidentiality of
information or information systems originated by the U. S.
Government or persons working under contract to the U. S.
Government, except for those specific items intentionally
released by the U. S. Government to the general public or
independently developed by a person or persons outside of the U.
S. Government. In case of any doubt about the status of any of
these items, see $120.4.
(2) Cryptographic (including key management) systems, equipment,
assemblies, modules, integrated circuits, components or software
which have the capability of generating spreading or hopping
codes for spread spectrum systems or equipment and which were
originated by the U. S. Government or persons working under
contract to the U. S. Government, and not independently
developed outside of the U. S. Government.
(3) Cryptanalytic systems, equipment, assemblies, modules,
integrated circuits, components or software originated by the U.
S. Government or persons working under contract to the U. S.
Government, and not independently developed outside of the U. S.
Government.
The above changes have the effect of maintaining strict controls
on the cryptosystems that we use in our own military and
diplomatic service, but has no ill effects on the U. S.
Constitution or economy. It also has the effect of costing less
taxpayer money to support censorship (Commodity Jurisdiction)
proceedings.
$ 121.8 (f) should be modified to read (deleting the exception
for encryption software):
(f) Software includes but is not limited to the system
functional design, logic flow, algorithms, application programs,
operating systems and support software for design,
implementation, test, operation, diagnosis and repair. A person
who intends to export software only should apply for a technical
data license pursuant to part 125 of this subchapter.