Interesting People mailing list archives

IP: Techno-Hero or Public Enemy?


From: Dave Farber <farber () cis upenn edu>
Date: Wed, 06 Nov 1996 07:23:03 -0500

   Fortune, November 11, 1996, pp. 172-82.




   Techno-Hero or Public Enemy?


      James Bidzos of RSA Data Security wants to go global
      with a potent shield against computer break-ins. Uncle
      Sam's most secretive spy agency wants to stop him. At
      stake is the right to privacy and the health of the
      U.S. software industry.


   By David Stipp 




   If you work with a computer, chances are that embedded in
   your machine is a nifty little piece of software that FBI
   Director Louis Freeh calls a terrorist tool. Made by RSA
   Data Security, the software is woven into some 90 million
   copies of various applications, from Netscape browsers to
   Lotus Notes to products from Microsoft. Few, if any,
   programs are as ubiquitous. Yet RSA's software is so
   powerful that, as recently as October, it was classified
   as ammunition under U.S. laws that restrict arms exports
   -- along with cruise missiles and Stealth bombers.


   RSA is spreading everywhere for the same reason that the
   federal government doesn't want it to: It can thwart your
   enemies, whoever they are. Its power springs from
   encryption, which scrambles data sent over phone lines or
   stored in computers so that only those holding secret
   numerical "keys" can get access. Law enforcers fear it
   will be used to block them from eavesdropping on
   evildoers. But it also can stop bad guys from hacking
   into your company's computers, or stealing your credit
   card number if you shop on the Web. The tabloids would
   never have known Princess Diana was "my darling Squidge"
   to her secret amigo if she had used a cellular phone with
   encryption.


   As RSA's astronomical installed base attests, its
   software has become a de facto standard for safeguarding
   computer bits. If you have a Netscape Web browser, note
   the key icon in the lower left-hand corner of your screen
   -- it means RSA is inside and will leap to protect your
   credit card number when you electronically order things.
   As the Internet burgeons, so do applications for RSA. It
   protects the confidentiality of medical records exchanged
   by doctors via the Net and safeguards signals piped from
   video cameras at nuclear sites to an international agency
   monitoring atomic weapons.


   Indeed, RSA has quietly become to software what Dolby is
   to audio: the trusted brand most every vendor wants
   inside. The expected explosion of electronic commerce
   promises to multiply its licensees even faster. By 2000,
   International Data Corp. estimates, $78 billion of
   purchases will be made annually via the World Wide Web.
   RSA aims to keep all those cyberbucks honest by stamping
   electronic documents with "digital signatures" enabling
   online merchants to check that customers are legitimate
   (see box).


   Positioning RSA to be the key provider of trust in
   cyberspace, CEO Jim Bidzos recently did a strange thing:
   He decided not to make an initial public offering. Wall
   Street had been eagerly anticipating an IPO; last spring
   *ComputerLetter*, a New York City newsletter aimed at
   high-tech investors, called RSA the "poster-child of the
   networked future." Investment bankers during the recent
   Internet IPO craze estimated that an RSA offering would
   raise a hefty $300 million, says Bidzos. Not bad for a
   firm with 50 employees and 1995 revenues of $11.6
   million.


   But ignoring the bankers hot advances, in February he
   negotiated RSA's purchase for stock by Security Dynamics,
   a Bedford, Massachusetts, maker of computer-security
   devices. Completed in July, the $259 million merger is
   aimed at marrying RSA's software with Security's hardware
   to make the picks and shovels of the cybercommerce gold
   rush. For example, says Bidzos, who remains head of RSA
   in Silicon Valley, fraud-resistant smart cards that carry
   everything from digital cash to medical records.


   Bidzos maintains that the companies' synergy will give
   RSA more value than an IPO would have. Swept up in the
   speculation surrounding Internet stocks, Security
   Dynamics' share price has gyrated between $109 and $51.25
   since the merger was announced. Today Wall Street seems
   to agree with Bidzos that dominance in the security niche
   makes the company a choice Internet play: The Security
   Dynamics shares swapped for RSA were recently worth some
   $330 million. Analysts expect earnings to be around $23
   million on sales of $113 million next year, up from a
   projected $12 million on $74 million in sales in 1996.


   For Bidzos, 41, a former programmer with a standup
   comic's irreverence and wit, steering RSA against
   prevailing winds is almost second nature. For years he
   had little choice. After being named CEO of the fledgling
   company in 1986, he sought venture funding in vain. He
   recalls, "I would sit in meetings with venture
   capitalists, and somebody would say, 'Let me get this
   straight. There's no market for your technology, but
   you're going to try to create one by promotion. And the
   world's largest intelligence agency wishes you didn't
   exist, but you're just going to beat them at that game.
   Okay, it's been nice talking to you.' "


   Bidzos could hardly deny that the U.S. National Security
   Agency considers his company a loose cannon. The NSA,
   which declined comment for this story, was created
   secretly in 1952 to conduct electronic eavesdropping on
   foreign powers. It has sought to keep a lid on
   hard-to-break ciphers, the antiballistic missiles of the
   Information Age. The NSA's attacks often work -- a few
   years ago, for instance, it quietly persuaded AT&T to
   change its product strategy. But now the lid is being
   pried open, largely by software firms using RSA's
   programs. And when the NSA has tried to wrap its arms
   around the defiant little company, Bidzos has tweaked its
   nose and danced away.


   In his latest coup, Bidzos announced in June that RSA had
   agreed to put its imprimatur on encryption chips by
   Japan's Nippon Telephone & Telegraph Corp. The pact
   sidesteps U.S. export controls, which block RSA from
   developing such chips in the U.S. for the global market.


   The NSA is understandably irked. State-of-the-art ciphers
   are almost bulletproof -- deciphering a message scrambled
   with a "strong" version of RSA's software would take even
   the fastest computers millions of years. In the wrong
   hands, such ciphers could tie the agency in knots. It has
   tried hard to prevent that.


   Four years ago NSA officials learned that AT&T planned to
   launch a "secure" telephone system in the U.S. that
   scrambled conversations with a tough cipher called DES.
   Uncle Sam's cryptowizards were aghast -- the
   user-friendly device seemed ideal for spy-proofing the
   leaky communications channels they depend on. By early
   1992 the NSA felt obliged "to call in the cavalry," says
   Stewart Baker, the agency's former general counsel. "The
   FBI and Justice [Department] received increasingly urgent
   briefings" from the spy agency's brain trust about the
   dangers of spreading encryption. Meanwhile, an NSA team
   accelerated work on an alternative scrambler-on-a-chip,
   dubbed Clipper. It was designed so federal law enforcers
   with court orders would have access to codes for
   deciphering scrambled transmissions, preserving their
   ability to conduct wiretaps.


   AT&T soon heard that dropping DES in favor of Clipper
   would please some very important persons. Neither AT&T
   nor the federal agencies will comment on what happened
   next. But at 4 P.M. on October 13,1992, according to
   previously classified federal memoranda, then-FBI
   director William Sessions called AT&T CEO Robert Allen to
   discuss the government's "problem" with the DES phone.
   Sessions, who had been asked to make the call by Attorney
   General William Barr, reported back to Barr that "Mr.
   Allen indicated that he could support an accommodation."
   The memos were obtained through the Freedom of
   Information Act by the Electronic Privacy Information
   Center, a Washington, D.C., group opposed to federal
   restrictions on encryption.


   Whether Allen's arm was twisted is unclear -- Sessions
   noted to Barr that his conversation with Allen "was
   guarded because we were not talking on" a secure phone.
   But Allen may have had ample reason to take a hint --
   AT&T was starting work on a federal telecommunications
   contract that would be worth up to $15 billion if all
   went well. Besides, Clipper appealed to AT&T because
   products incorporating it would be legal to export.


   In April 1993 the White House unveiled the Clipper chip,
   touching off a storm of protest from computer companies
   invoking Orwell's 1984. Bidzos gleefully led the charge
   by calling for a boycott of products with "Big Brother
   inside." But AT&T was conspicuously absent from the
   critics' camp. In a press release coinciding with the
   Clipper launch, Ma Bell gushed that she was "pleased to
   be the first company to publicly commit to adoption of
   the Clipper chip" in secure phones. To AT&T's chagrin, it
   turned out to be the only big company to adopt Clipper.


   Besides galvanizing industry opposition to government
   restrictions on encryption, the Clipper debacle helped
   Bidzos raise RSA's profile -- in effect, the NSA handed
   him millions of dollars of free advertising, since
   quoting him became almost *de rigueur* for reporters
   covering the Clipper debate. Fumes a former government
   official: "Fighting the NSA turned out to be a great
   business move."


   Bidzos professes what seems sincere respect for the spy
   agency -- several of its former staffers even work for
   him. But he insists that it needs a gadfly: "I've gone
   there a few times and listened to their dark-side folks.
   I always come out feeling that I'm a traitor. I deserve
   to die. I'm causing the early demise of society and
   should just scuttle the company. But an hour after I
   leave the agency, my head clears, and I think, 'Okay,
   back to reality.' I wake up nights sweating after these
   meetings -- I have a nightmare that I'm cornered by NSA
   agents who are about to serve me with a security
   clearance."


   Recently the Clinton Administration has tried to distance
   itself from Clipper. On October 1 it proposed temporarily
   removing encryption technology from the federal munitions
   list. That may make exporting security products easier
   for U.S. industry. But encryption exports will still be
   reviewed by a federal panel that software companies fear
   will be dominated behind the scenes by law enforcement
   and intelligence agencies. "The new proposal is a
   positive move, but we still have a long way to go," says
   a spokesman for the Software Publishers Association.


   Moreover, the shift is contingent on industry commitments
   to market products with "key recovery" -- hardware and
   software built so law enforcers can obtain numerical keys
   for deciphering scrambled data. Software companies aren't
   opposed to storing backup copies of keys. But they argue
   that, depending on who is entrusted with the backups,
   billions of dollars of sales may be lost to foreign
   rivals offering secure products with no peepholes. Says
   Bidzos, "We know what it takes for one of the CIA's most
   powerful guys, Aldrich Ames, to give up what he knows --
   about $250,000 a year. What will a foreign government
   have top pay a $20,000-a-year federal bureaucrat to sell
   the keys for industrial secrets worth hundred of millions
   of dollars?"


   While the Clipper debate has raged, an equally important
   battle between Bidzos and the NSA has unfolded behind the
   scenes. At issue is a crucial question that must be
   settled before electronic commerce can take off: Will RSA
   or the federal agency provide a basic technology called
   digital signatures?


   The question looms large, since companies hosting
   Internet-based transactions will need standardized
   software to handle basic functions such as security --
   the alternative is a welter of incompatible programs few
   customers would use. Digital signatures will be
   especially important because they make it possible for
   money to change hands on the Net in a digital form that
   is practically impossible to forge.


   Such applications of scrambling don't worry the NSA. But
   it fears the demand they create will speed the global
   spread of RSA's versatile cryptoware. The software might
   end up being used for tasks that do spook the agency,
   such as encrypting digital phone calls. Thus, the NSA
   launched a preemptive strike in 1991 to stem the
   groundswell building for RSA as a standard for digital
   signatures.


   Using the National Institute of Standards and Technology
   (NIST) as its marketing arm, the publicity-shy agency
   developed and launched an alternative it preemptively
   named Digital Signature Standard, or DSS. Unlike RSA's
   offerings, DSS is designed exclusively to help users
   prove their identities -- it can't be used to protect
   other data, says Lynn McNulty, a McLean, Virginia,
   computer-security consultant who formerly worked at NIST.
   Translation: DSS can t be used to block up the NSA's
   far-flung ears.


   Thus, unlike RSA software, DSS is deemed safe for export.
   The government charges no royalties for its use.
   Moreover, federal agencies are required to use DSS for
   secure transactions, creating a sizable market. The NSA,
   it seems, figured Uncle Sam's purchasing power would trap
   the gadfly in its ointment.


   But Bidzos, a crack blackjack player who once won $10,000
   on a Las Vegas streak, was holding strong cards. One was
   his software's distinguished reputation. Over the years,
   RSA's patented algorithm has been relentlessly analyzed
   by a host of encryption experts -- researchers in the
   field love to one-up rivals by breaking their ciphers. So
   far, no one has found any significant chinks in RSA's
   armor, making it more tried and trusted than DSS.


   RSA's versatility was another Bidzos ace. Now that much
   of the world is wired -- and going wireless to boot -- 
   opportunities for data theft are cropping up everywhere.
   If you believe data piped over your company's intranets
   are safe, think of Princess Squidge: Unless your company
   has only one office, the information is probably flowing
   through the public telephone network, which includes a
   multitude of vulnerable nodes.


   Still not worried? Consider an unclassified section of a
   secret U.S. counterintelligence report to Congress last
   year. It quotes two former directors of France's
   intelligence service as stating that spying on U.S.
   concerns has long been a top priority at their agency.
   Recently the CIA released a report alleging that the
   French and Israeli governments aggressively spy on U.S.
   companies. With friends like these, who needs news of ex-
   KGB spies for hire to get freaked?


   The advent of RSA's technology to help counter such
   threats was "tremendously exciting," recalls Noel
   Matchett, a former NSA staffer and now president of
   Information Security, a company in Silver Spring,
   Maryland. Among other things, he notes, RSA makes it
   practical to continually change the keys for scrambling
   data sent over networks, robbing spies of the chance to
   stay tuned for long if they happen to break a code. DSS
   can't do that.


   Life at RSA wasn't always this glamorous. Its technology
   was invented in 1977 by three scientists at Massachusetts
   Institute of Technology: Ronald Rivest, Adi Shamir, and
   Leonard Adleman. By 1982 they had secured a patent
   through MIT and had started the company in Adleman's
   one-bedroom apartment. (RSA is named after their
   surnames' first letters.) But in those days most
   computers were far too slow for the software --
   scrambling data with RSA's algorithm demands a lot of
   number crunching. Says Rivest: "We were way ahead of the
   market."


   The company nearly went bankrupt before Bidzos was hired
   to turn it around, and the co-founders went back to their
   academic careers (Rivest has remained chairman). Bidzos
   recalls his first year on the job: "It was endless grunt
   work, flying cross-country and talking up RSA. For a
   while I didn't pay myself, to keep costs down."
   Eventually the networking revolution let him bring home
   some bacon. One of his first deals was with Fischer
   International Systems Corp., a Naples, Florida, software
   concern. After licensing RSA's patent, says founder
   Addison Fischer, "I thought, 'I never want to negotiate
   with this man again. He's just too skillful a salesman.'"


   Others who have dealt with Bidzos agree, describing him
   as a charming but ruthless riverboat gambler with a trump
   card in his hand and a six-shooter at his elbow. The card
   is RSA's patent. The gun is loaded with infringement
   lawsuits, mostly unfired.


   In 1987, Lotus signed up to use RSA for Notes, followed
   over the next few years by Motorola, Apple, Novell, and
   others. Suddenly RSA was on the map -- especially the one
   at the NSA in Fort Meade, Maryland. Says Bidzos: "For
   years the NSA seemed to ignore us, thinking we'd self-
   destruct. But by '91 it became clear to them that,
   'Whoops, this guy isn't going away, and he's beginning to
   integrate this stuff in user-friendly products. Uh-oh.' "


   Bidzos alleges that in early 1991 "the NSA tried to kill
   a major deal I was about to close. They called a
   well-known executive at a well-known software company and
   said, 'Why not wait for our DSS?' When my contact at the
   company told me, I was furious. I called the NSA and told
   them to call [the software company] back right now and
   tell them it was a terrible mistake. 'If you don't, you
   can explain why not to my Congressman. And if that
   doesn't work, you can explain it to the New York Times.'
   The guy said, 'I'll take care of it.' And he did."


   Bidzos declines to name the would-be customer. But in
   mid-1991, he landed a contract with a well-known software
   company: Microsoft.


   Despite such behavior, the NSA is usually far from inept.
   Interviews with a number of former staffers suggest that
   its top ranks are populated with smart people who
   understand the industry's beef -- even if it doesn't sway
   them. Indeed, the NSA has a big encryption operation of
   its own: Its No. 2 mission is to protect sensitive
   federal communications. Says Ed Hart, formerly the
   agency's deputy director in charge of information-systems
   security and now an executive at Science Applications
   International Corp.: "The NSA is rich with talent and is
   anything but bumbling. But it's insulated and sometimes
   doesn't appreciate what's going on in the outside world."


   Bidzos and his allies exasperate the agency, for they are
   trumping it in the court of public opinion, where it is
   hamstrung by its famous secrecy. Consider this: To get
   access to many U.S. nuclear secrets, all you need is "Q
   clearance," a step below top secret, according to a
   report by the Association for Computing Machinery. But to
   work at the NSA and get the skinny on SIGINT, or "signals
   intelligence," you must rate a top-secret clearance and
   then pass muster in a psychological strip search while
   lashed to a lie detector.


   "SIGINT is more valuable than dope," says a former Army
   intelligence officer, "because it goes directly to the
   personal power and prestige of the President." During the
   Cold War, for instance, the NSA reportedly briefed the
   White House regularly on what Soviet leaders were saying
   on their car phones.


   The NSA's accomplishments aren't confined to the geo-
   political. In recent years the economic benefits of
   SIGINT to U.S. industry have totaled tens of billions of
   dollars, states a landmark encryption report issued in
   June by the National Research Council, a congressionally
   chartered advisory group. The NSA declassified that
   astounding assertion at the council's request. The
   estimate is based on top-secret data, but a recent,
   widely reported episode suggests it's credible. Two years
   ago SIGINT helped U.S. officials blow the whistle on
   alleged French bribes to get a $6 billion contract from
   Saudi Arabia for weapons and commercial aircraft. After
   President Clinton informed Saudi King Fahd, an Airbus-led
   consortium that seemingly had the aircraft award in the
   bag was rejected in favor of Boeing and McDonnell
   Douglas.


   All of which helps explain the furrowed brows at a 1991
   meeting of federal officials involved in developing DSS.
   They had just learned that the technology, which they saw
   as a crucial part of the strategy to keep SIGINT flowing,
   was seemingly covered in a U.S. patent that had just been
   awarded to a German math whiz named Claus Schnorr.
   Hurriedly, they rejiggered DSS in hopes of preempting a
   Schnorr patent attack. It didn't work.


   The German soon sent a letter to NIST raising the
   possibility of a patent lawsuit. A few months later,
   security consultant McNulty, then still a NIST official,
   traveled with a colleague to Frankfurt to feel out
   Schnorr's intentions. Says Schnorr: "They explained how
   they thought they had circumvented my patent. But I had a
   different opinion." Indeed, he wanted $2 million for a
   patent license. NIST talked with him periodically about
   that, and eventually told him to get lost -- it now
   asserts his patent is a nonissue.


   But McNulty was worried: "As long as we didn't license
   Schnorr's patent, there would always be uncertainty about
   how a court might rule on an infringement suit." The
   prospect of a patent war could make industry less willing
   to adopt DSS. Moreover, says McNulty, "I argued that if
   we didn't take the Schnorr wild card off the table, it
   was going to be sitting out there for Jim Bidzos to pick
   up."


   He was right. In March 1993, Bidzos and a business
   associate met Schnorr in Marseilles, France, where the
   German was visiting a friend. Bidzos made his pitch over
   a lunch of monkfish and a bottle of Sancerre at the Trois
   Fortes hotel overlooking the city's picturesque harbor.
   After listening for four hours, the laconic mathematician
   suddenly said Ja: He handed over rights to his patent in
   return for a share of the royalties Bidzos hoped to
   squeeze from it. Schnorr says he was swayed by the fact
   that Bidzos had built up RSA from a single patent:
   "That's not trivial to do."


   Since then, Bidzos has rattled the Schnorr saber to the
   desired effect. "We've written letters to companies
   selling DSS," he says, "and have licensed Schnorr to a
   number of companies," including IBM. "People are afraid
   of getting involved in litigation."


   Still, not all has gone smoothly for Bidzos. The presence
   of the FBI in the encryption debate may make it harder
   for him and his allies to loosen their market's legal
   handcuffs. In alarming speeches, FBI director Freeh has
   repeatedly linked unrestricted encryption to terrorists,
   such as the Oklahoma City bombers. The FBI has more
   reason than the NSA to be hostile, experts say. It lacks
   the NSA's appreciation of encryption's ability to enhance
   national security, for one thing. It is also more likely
   to be stymied by encryption than the NSA, which is widely
   thought to be able to break some of the weaker codes now
   in use. (The FBI once regularly sought the spy agency's
   help, according to *The Puzzle Palace*, a 1982 book about
   the NSA. It quotes a former NSA staffer as saying that
   the FBI would ask the agency to unravel codes used by
   bookies: "We used to give them to our cryptanalysts to
   take cracks at in their spare time. They would break them
   over lunch and send them back to the FBI.")


   Bidzos also faces hard-liners closer to home. Two years
   ago RSA had a falling out with Cylink, a Sunnyvale,
   California, concern that holds encryption patents from
   Stanford University. The companies dissolved a
   partnership and began a messy patent battle. Both sides
   claimed victory after an arbitration-panel ruling last
   year; recently a U.S. district court denied Cylink's
   motion for a preliminary injunction against sales of RSA
   products, suggesting that RSA's patent is still golden.


   But when it expires in four years, Bidzos's main ace will
   be gone and he'll have to rely on another pair --
   marketing and the momentum of a big installed base.
   Potentially significant rivals already loom, such as
   Certicom, a Canadian company that claims its encryption
   algorithm is superior to RSA's for certain uses, such as
   in pagers and cell phones.


   Bidzos isn't fretting, though. On a recent typical day,
   RSA's bluejeans-clad boss drove one of his two BMW
   motorcycles to the company's modest digs at a Redwood
   City, California, office park, then got to work fielding
   media calls, reading E-mail, and monitoring Internet
   companies' stock prices. Interrupted by his secretary to
   sign a sheaf of contracts, he boasts that "the Internet
   probably spawns a dozen companies an hour, and almost all
   of them become RSA customers."


   Some form even closer ties: In lieu of royalties, RSA has
   taken stakes in Netscape (1%) and CyberCash (3%). Says
   Stephen Crocker, a senior vice president at CyberCash, a
   maker of payment systems for Internet shopping that use
   RSA: "It's a magnificent deal from Bidzos's point of
   view. If a company [paying with stock] goes out of
   business, he wouldn't get any royalties anyway. But if it
   succeeds, he wins much bigger than he would with
   royalties."


   As for DSS, sales for federal use are keeping it alive.
   But it appears to be a nonstarter in the crucial
   electronic-commerce market. There RSA is roaring ahead.
   This year, for example, MasterCard, Visa, and other big
   players announced specifications for Internet
   transactions based on RSA encryption. Indeed, RSA's
   success has made it something of a magnet for former
   federal encryption experts seeking private-sector jobs,
   including authorities on DSS. Says Bidzos: "I've gotten
   so many resumes from Washington, you wouldn't believe it.
   I've outlasted them all, and now they want to work for
   me. I love it."


   _________________________________________________________
   [Box]


   When g*%#B/L#h Isn't a Curse


   Military experts estimate that by cracking the Nazis'
   Enigma cipher, Allied codebreakers saved the world at
   least a year of war and millions of lives. Thanks to the
   feat, the Allies knew German U-boat positions, strategic
   plans, and, before D-day, where Hitler thought the attack
   would come.


   Enigma's fatal flaw was a problem that had always
   bedeviled efforts to keep government secrets by
   scrambling messages: The "keys" for unraveling them had
   to be widely distributed among intended recipients,
   usually making it only a matter of time before they fell
   into enemy hands. Enigma's keys were embodied in
   electrical devices, one of which was secretly obtained by
   a British team that analyzed it and broke the cipher in
   1940.


   But in 1976, Stanford University scientists proposed a
   way around this problem. Called public-key encryption,
   their idea was translated into a versatile datascrambling
   method a year later by a team at the Massachusetts
   Institute of Technology that founded RSA Data Security.
   Here's the gist: A key for coding messages is a number
   that, when fed into a computerized scrambler, sets the
   way it translates characters of the original text into
   the apparent gibberish of its encrypted form. (FORTUNE,
   for example, might become g*%#B/L.)


   RSA's method involves two keys that reverse each other's
   effects when fed into software that both scrambles and
   unscrambles messages. One, the "private key," is known
   only to an individual. Its mate is called a public key
   and is listed next to the owner's name in a kind of phone
   book for anyone to see. The keys are a lot longer than
   phone numbers, though -- the most potent can include more
   than 300 digits. Each person has a unique pair of such
   keys.


   If Alice wants to send an E-mail message to Bob for his
   eyes only, she looks him up in the directory and uses the
   dual-purpose software to scramble the message with his
   public key. Bob unscrambles it at his end using his
   private key -- no other key will work. Unless Bob does
   something dumb, like storing his private key where his
   foes can get at it, they won't be able to decipher
   intercepted messages. There's only one way to try: Derive
   Bob's private key. But due to RSA's clever math
   underpinnings, that could take millions of years using
   even the fastest computers.


   RSA software is also used increasingly for signing
   electronic documents, a process that's just the reverse
   of scrambling a message. Roughly, here's how it goes: Bob
   scrambles a computerized document with his private key
   and sends it to Alice along with an unscrambled version
   of the same document. Using his public key, she decodes
   the scrambled version. If the result matches the
   unscrambled version, she can be sure of two things: The
   message came from Bob, and it wasn't altered en route.


   Why? Only data scrambled with Bob's private key can be
   unscrambled with his public key. It's as though Bob has
   stamped his fingerprints on every character in a way that
   can't be forged.


   Encryption isn't all there is to computer security. If
   Bob loses his private key after scrambling all his data
   to keep it from prying eyes, he's sunk. He'll never
   decode his data again. To guard against that, several
   companies offer systems that store copies of private keys
   in the digital equivalent of bank vaults -- a leader is
   Trusted Information Systems in Glenwood, Maryland.


   But scrambling software is useless if implemented in ways
   that let outsiders steal keys. Such problems, not chinks
   in encryption's basic armor, account for almost all cases
   of hackers breaking secure systems -- such as a flap last
   year over holes in Netscape browsers. To help encode
   messages, the browsers used a so-called random-number
   generator routine; but grad students at Berkeley showed
   that the numbers it generated were actually predictable,
   and broke the code. RSA employs crack crypto-hackers to
   ferret out such holes. Says Netscape security chief Jeff
   Treuhaft: "After the problem was found, we definitely
   used RSA to review our fixes and make sure they were
   safe."


   _________________________________________________________


   [Photo] Turning Gibberish to Gold. Bidzos and RSA plan to
   cash in on the electronic-commerce bonanza with
   scrambling software that protects secrets.


   [Photo] Scholarly Merchant of Code Chairman Ron Rivest,
   an MIT professor, saw RSA almost fail before its software
   took off as a replacement for scrambler circuitry like
   this.


   [End]


Current thread: