From: John Young <jya@pipeline.com>
Date: Mon, 26 Feb 1996 12:27:11 +0800
To: cypherpunks@toad.com
Subject: TWP on Crypto Export Policy
Message-ID: <199602260353.WAA13754@pipe1.nyc.pipeline.com>
MIME-Version: 1.0
Content-Type: text/plain


   The Washington Post, February 25, 1996, pp. H1, H4.


   Scrambling for a Policy on Encryption Exports

      As Technology Advances, U.S. and Industry Seek
      Compromise That Balances Public, Private Fears

   By Elizabeth Corcoran


   Keeping information about technology out of other people's
   hands gets tougher all the time. And in the realm of the
   Internet, where information ignores boundaries and some
   cybersurfers flaunt rules, it may become impossible.

   A big test of that statement is emerging in cryptography,
   the business of scrambling information so that it looks
   like gibberish to anyone lacking the keys for unlocking the
   code.

   Once considered an arcane subspecialty of mathematicians
   and espionage, cryptography is rapidly becoming big
   business as more and more of the world's information is
   exchanged on electronic networks and as more and more
   people want to protect their data from prying eyes.

   But in these times of international terrorism, drug
   trafficking and sometimes peculiar financial transactions,
   law enforcement agencies want to be able to legally
   eavesdrop. As technology has grown dramatically more
   powerful, the ability to peek at encrypted information is
   slipping from the hands of government.

   That balance -- how much access can government demand vs.
   how much privacy others want -- has long been a theological
   debate between civil liberties advocates and worried law
   enforcement officials.

   Now, giddy with the growth of the Internet, technology
   companies have joined the debate. Both the software
   industry and civil liberties advocates believe powerful
   encryption will spur the growth of electronic
   communications and the Internet. And they don't want
   encryption restrictions to curb that growth.

   There are no limits on what kinds of encryption people can
   use within the United States. But the government has used
   export restrictions to try to shape what encryption
   technology is used internationally, and by extension, what
   is available in the United States. Those export laws
   prohibit U.S. companies from selling their best technology
   overseas.

   The restrictions, companies contend slow the development of
   the Internet and harm a potentially lucrative market for
   U.S. manufacturing. Making two flavors of an encryption
   product, U.S. companies contend, is expensive. Yet even
   more worrisome is that foreign competitors are likely to
   move in and offer better technology. That could spell the
   loss not just of sales of encryption technology, but of
   many other products that rely on strong digital protection
   as well.

   So companies are looking hard for ways to wriggle around
   the rules -- and beginning to find them.

   "Trying to suppress this technobgy is like Prohibition,"
   said Whitfield Diffie, a cryptographer at Sun Microsystems
   Inc. and an outspoken advocate of widespread use of
   encryption technology. Companies will use anything at hand
   -- technology, business strategies and even the promise of
   congressional action -- to begin to get their home-brews
   out.

   Building and breaking encryption is hard. All information
   stored in computers -- whether pictures or sounds or
   documents -- is represented by ones or zeros or bits, the
   genetic code of the digital world. Encryption techniques
   amount to applying clever mathematical formulas to a
   collection of bits to make it look like gibberish to the
   uninitiated.

   Unlocking encrypted data requires a "key," a mathematical
   formula that can make sense of the tricks used to scramble
   the data. One common way to measure the sophisticaticn of
   an encryption scheme is by the number of bits in the key.
   The more bits, the harder it is to decode the information.

   A 30-bit key, for instance, could take as many as a billion
   random calculations to crack the code. A 60-bit key could
   take a billion-times-a-billion calculations.

   In past decades, governments were largely the only
   organizations with the money and need to tackle such
   expensive problems. But as the power of computers has
   soared -- and the cost of running millions of calculations
   has fallen -- companies and individuals have begun to
   clamor for sophisticated encryption.

   "We believe that encryption is a critical technology" to
   support many areas of electronic commerce, said Craig
   Mundie, a senior vice president at Microsoft Corp.

   Under current rules, U.S. companies can export encryption
   technologies that use up to 40-bit keys. A few years ago,
   such a lock might have stopped all but the most determined
   digital interlopers.

   No more. Within the past year graduate students at the
   Ecole Polytechnique in Paris and others at the
   Massachusetts Institute of Technology have shown they can
   break the 40-bit encryption used by Netscape Communications
   Corp. A few weeks ago, Diffie and six other well-know
   cryptographers began circulating a report in which they
   argue that to "adequately" protect information for the next
   20 years, keys should be as long as 90 bits.

   Even encryption wizards at the National Security Agency
   would have trouble unlocking 90-bit encrypted information,
   experts say.

   So the government has tried to craft a compromise. Last
   summer, the government suggested that it would likely let
   companies use up to 64-bit encryption -- provided they set
   up a way for law enforcement agents, with a court order, to
   unlock encrypted information.

   Under this proposal, a "trusted third party," such as a
   bank or an encryption company that typically handles
   sensitive information, would safeguard the key. The plan
   has since bogged down over such details as precisely who
   might qualify as a trusted third party.

   Last fall, Trusted Information Systems (TIS) in Glenwood,
   Md., in what it calls a test case, applied for a license to
   export a sophisticated (and still unexportable) 56-bit
   encryption system called DES. Steve Walker, who heads TIS,
   has invested months in outlining the sort of spare-key
   program that he believes both the government and his
   customers can stomach. In late January, he got approval to
   ship his product to Britain.

   "It's not perfect; it's not where we want to be," Walker
   insisted. He purposely submitted a case, he said, that was
   virtually certain to meet the government's still evolving
   criteria. "But it's a first, giant baby step," he said.

   Trusting a Third Party?

   Others are uneasy with putting the means to unlock files in
   the hands of a "trusted third party."

   "Ask anyone who owns a business: Are they willing to give
   copies of a spare key that leads to everything sensitive in
   their company to a third party," said Jim Bidzos, chief
   executive of RSA Data Security Inc., a leading encryption
   firm.

   But government officials get nervous if the only keys to
   the scrambled material are held by its owners. Ed Roback,
   an encryption policy specialist at the National Institute
   of Standards and Technology, puts it this way: "I know of
   few front doors that can't be broken down. It's a little
   different with encryption," when it literally might take
   10,000 years to break the code without the key.

   Roback and law enforcement officials say they'd be
   delighted to see Americans make more use of encryption,
   particularly if spare keys were held by a third party.
   "This nation, more than any other, relies on computers ...
   [so] there's a lot of vulnerability and encryption can help
   that," Roback said. "So it's a good thing -- but it can
   present a problem for national law enforcement."

   But momentum in the United States could swing toward
   widespread use of sophisticated encryption -- without spare
   keys -- if such technology was widely available. That's
   just what a recent announcement from Microsoft could help
   spur.

   In January, Microsoft told developers it had created a
   module in its operating system software that will let
   applications such as word processing programs or
   spreadsheets "plug in" to encryption technology.

   An application developer who built a software program for
   filing expense accounts would not have to add encryption to
   his product. Instead, the developer would need only to
   write a small program that taps the encryption technology
   available through the operating system.

   The strength of the encryption program could vary.
   Microsoft plans to include a 40-bit code with the version
   of Windows used principaUy by companies (called Windows
   NT). That encryption technology would be easy to export.
   But Microsoft also is encouraging other encryption firms,
   including RSA and TSI, to build more sophisticated
   encryption modules that could be used in the United States.

   Commercial products that take advantage of the new function
   are not likely to appear until the end of the year. But
   Microsoft is hoping it will spur more widespread use of
   encryption. "The single most pressing problem for
   electronic commerce is to create a secure payment
   structure," Mundie said -- and Microsoft is hoping to
   accelerate that work.

   RSA's Bidzos is among those in industry who would love to
   see the government give up on trying to control encryption
   technology. He worries that other countries are gearing up
   to snatch a big role in selling encryption while his
   company and other U.S. businesses remain entangled in U.S.
   policies.

   So he's testing the rules. Early this month, RSA announced
   that it had created subsidiaries in the People's Republic
   of China and in Japan. In China, RSA partners include the
   Chinese government. Bidzos plans to do joint research on
   encryption software with scientists there.

   Although Bidzos says he is planning to export only the
   approved, 40-bit encryption technology to his Chinese
   colleagues, "one genuine concern is that they might try to
   strengthen it themselves," he said. "It would be hard to do
   -- but not impossible. I've never had a conversation with
   [the Chinese] about it," Bidzos added.

   In addition, the Chinese have some interesting ideas of
   their own about new areas of cryptography, Bidzos said.
   "They're pretty advanced." And if the group developed more
   powerful techniques than even RSA has in the United States?
   Bidzos shrugged. RSA would likely take any promising ideas
   and develop them into products in the United States. As for
   Chinese export restrictions, "I haven't thought about it,"
   he said.

   Going Up to the Hill

   Industry also is fanning Congress's interest in taking a
   bigger role in the encryption debate. "Without
   congressional interest, the administration has no reason to
   liberalize exports at all," said Becca Gould, director of
   policy at the Business Software Alliance. "This issue is in
   Congress's front yard because it affects the economy" as
   well as U.S. citizens' privacy rights.

   Sen. Patrick J. Leahy (D-Vt.) and Rep. Robert W. Goodlatte
   (R-Va.) agree. They plan to introduce bills in the Senate
   and House aimed at loosening the restrictions on
   encryption. "The federal government is taking an attitude
   that's based more in the 1970s than in present time," said
   Leahy in a telephone interview.

   "This is a matter that should be decided by legislation,"
   he added. "We're talking about billions of dollars in
   revenues and thousands of jobs if we're handicapped in our
   global market, especially if what we're told to do is to
   build an export encryption program that is so outdated that
   our 12-year-old computer experts would laugh at it."

   The bills would do away with export licenses for any
   encryption technology considered to be "generally
   available," or "in the public domain." Leahy said that
   although he, too, worries about national security and
   terrorism, trying to bottle up technology won't solve the
   problem.

   Law enforcement has "got to figure out how to keep ahead
   ... and surprise, surprise, there will be some times when
   we won't be able to eavesdrop," Leahy said. Even now,
   criminals can make calls at pay telephones or avoid
   detection in other ways. The government shouldn't cripple
   the computer industry every time a new technology springs
   up that challenges law enforcement, he said.

   "What I'm suggesting is that if [the administration] works
   with the Congress, we'll find a solution," Leahy said.

   "We say over and over that we recognize that this is a very
   difficult issue," Roback said. But, he added, "the
   government has thought about [encryption policies] for a
   long time as well as industry," he said. To reach some
   resolution, he added, "compromise is going to be necessary
   on all fronts."

   [Photo] Jim Bidzos, chief executive of RSA Data Security
   Inc., a leading encryption firm, hopes the government wlll
   let go of encryption technology controls.

   _________________________________________________________

   Code Breakers

   Recent advances in technology have allowed much faster and
   cheaper invasion of encrypted information. The deciphering
   time, however, varies widely with the computer power of the
   attacker.

   Here are estimates of cracking times by one group of
   experts.

   Type of         Budget for      Time to       Time to
   attacker        computing       recover       recover
                   engine          40-bit key    56-bit key
   _________________________________________________________

   Pedestrian
   hacker          $400            5 hours       38 years

   Small
   business        $10,000         12 minutes    556 days

   Corporate
   department      $300,000        24 seconds    3 hours

   Big company     $10 million     7 seconds     13 hours

   Intelligence
   agency          $300 million    0.0002 sec.   12 seconds
   _________________________________________________________

   Source: Report by an ad hoc group of cryptographers and
   computer scientists Matt Blaze, Whitfield Diffie, Ronald L.
   Rivest, Bruce Schneier, Tsutomu Shimomura, Eric Thompson,
   Michael Wiener.
   _________________________________________________________

   [End]