On May 16, 1985, at a Denny’s restaurant on LBJ Freeway, the cops and the investigators gathered at 4 a.m. to review the final plans for what would turn out to be a most unusual raid. Anybody in the restaurant could have taken one look at the eight officers and guessed what was up. There was going to be another bust, and this one seemed big.
In fact, this was a new kind of operation that hadn’t been tried before in Dallas- the cops weren’t sure if they would even need their guns-and all sorts of things could go wrong. Led by a veteran special investigator from the Dallas County District Attorney’s office, John Palich, a man whose hard face is dominated by piercing dark eyes, the officers knew only a few scraps of information. They were going to raid a small, specialized research company in Far North Dallas called Voice Control Systems Inc., they weren’t going to let anybody out, they weren’t going to let anyone use the phone, and they were going to start searching for evidence of a crime.
But what?’ Fingerprints, blood stains? Some kind of weapon?
Palich, of course, had already asked himself the same questions. At age fifty-one, he had been through the gamut of criminal investigations, from murder stakeouts to bank fraud. He had spent the last ten years investigating complicated white-collar crime. But now, he had to admit he was somewhat fuzzy about what would happen.
Trade secrets, he told his men. We’re looking lor computer secrets. He still didn’t know what a trade secret was, or what one looked like. None of the officers even knew how to work a computer. All Palich knew for sure was that the monolithic, Dallas-based electronics giant, Texas Instruments Inc., believed that this little seven-year-old company was trying to rob it blind. TI officials had even uncovered a “mole” inside Voice Control Systems who was passing along information about what VCS had.
At 7 a.m.. the officers came through the doors of VCS, met the perplexed employees as they arrived for work, held them incommunicado until noon, then sent them home as a tedious twenty-si x-hour search of the offices was conducted. Thus began one of the most curious criminal investigations ever seen in Dallas. The officers had walked right into the middle of a high-technology war that is reaching a fevered pitch, one in which billions of dollars are at stake, as some of the best computer researchers and scientists in the country hunt for the precise computer program that will, in time, completely change the way the world communicates with machines.
Dozens of high-tech companies are furiously searching for the formula that can make a computer competently and quickly understand the human voice. Once the stuff of science fiction movies, when Hal of the movie 2001 sparked the world’s interest in conversing with computers, the ability to talk to machines has slowly become a part of everyday life. Computers already can be trained to speak certain words, and they can recognize a very limited number of spoken words-but so far no one has come up with the formula that can make a computer recognize a large vocabulary, regardless of who is speaking. This new technology is still a few years away, but the breakthrough would force a quantum jump in the computerizing of society.
“Whoever breaks the logjam to make machines understand spoken words,” says Michael L. Dertouzos, director of the Massachusetts Institute of Technology’s Laboratory for Computer Science, “will gain control over the information revolution.”
Those companies, needless to say, also know they will make a fortune, creating products that allow a person to speak to typewriters, toys, household appliances, door locks, and automobile controls, And that’s why Texas Instruments wanted to know exactly what VCS was up to.
Over the last three years, several employees had left TI to work for VCS, which was trying to develop advanced computer software that would allow computers to recognize spoken words. TI, known worldwide for its semiconductor, consumer electronics, and defense products, has also been an industry leader in the field where humans interact with computers through voice commands. Not only were some Texas Instruments executives getting peeved at the pesky little cross-town company that was beginning to receive some attention, they were also becoming suspicious.
In today’s hotly combative high-tech industry, companies often sue ex-employees who they think have taken a trade secret with them (see box, page 106), yet almost never are criminal charges filed. Never in Dallas had anyone been prosecuted under the most recent (January 1, 1974) third-degree “theft of trade secrets” felony statute. When Texas Instruments secretly learned, however, that VCS might be using some TI computer programs, it went to the district attorney’s office, which in turn conducted the pre-dawn raid to look for evidence on who might have taken TI’s privileged information. Two TI engineers also aided in the search of the VCS offices.
The investigation ultimately resulted in an emotion-charged trial in a Dallas County courtroom, where two talented engineers were charged with stealing a series of TI’s computer speech programs allegedly worth $20 million. Tom Schalk, thirty-five, was one of the rising stars at Texas Instruments, a handsome, athletic engineer whom his supervisors say was destined for a top managerial position with the company. Gary Leonard, forty-two, was a quiet, bearded researcher, a skillful man who would sit unmov-ing for hours behind a computer terminal, trying to figure out how a computer could better understand the human voice. His work had already helped lead to a couple of important breakthroughs in computer-speech research.
Both of them worked intimately with one of the most famous names in speech science research, Dr. George Doddington, a daz-zlingly brilliant man whose fights with TI management had made him a legend of sorts among TI staffers. A man so involved in his work that he once spent two full days in his laboratory before remembering that he had neglected to tell his frantic wife where he was, Doddington, now forty-five years old. was the scientist TI counted on to lead the company into this new computer-speech information age.
Schalk and Leonard were accused of betraying Doddington, sneaking off with his secret research so they could find a shortcut to the top, Or was it that easy? Did they actually take a “trade secret” from TI, or were they simply scapegoats? Their attorneys argued that this was no theft at all, that the always hard-nosed TI management was looking for a way to shut VCS down and make it an example for those employees who were thinking about starting a rival company.
The unusually cutthroat fight also pointed out the extraordinary lengths to which high-tech companies now go to keep trade secrets under wraps. As the American economy depends more and more on information-based businesses, new technology becomes the driving force behind a company’s profit margin. In this age, the value of a company is almost entirely determined by the information that it has. And the new business superstars aren’t necessarily the Lee lacoccas and Carl Ichans. but the scientists and engineers, the ones who create the technology. Unseen and seldom publicized, they are the key players in a multimillion-dollar race to develop better products. And when a top researcher is lured away from one high-tech company by a competitor, the results can be devastating-especially if that researcher comes over with key information.
THE PECULIAR STORY BEGINS IN 1970, WHEN George Doddington, a charming, sometimes eccentric, and often brash young man, was hired by Texas Instruments to work on something called “speech processing.1’ Just out of the University of Wisconsin with a Ph.D. in electrical engineering-a degree almost denied him because he sent a blistering memo to a professor criticizing the way he spelled-Doddington was already one of these utterly absorbed researchers who had a great vision about the future of computer-speech technology and simply couldn’t understand those who disagreed with him.
During his first week on the job he demanded that TI send him to an engineering conference in Houston. When his supervisors balked, he went into a rage and threatened to quit. Obsessed with one of his projects, he would forget to comb his hair or even think about what he was wearing. The stories about Doddington’s clothing are notorious. When one researcher’s wife came to the office one day to meet the great Doddington, a man came out wearing a striped shirt, bermuda shorts, black shoes, and black socks.
“You must be George Doddington.” the woman murmured, shaking her head.
“How did you know?” asked Doddington, surprised.
The gangly, boyish-looking Doddington, with a disarmingly goofy smile, was the envy of many in TI’s central research laboratories because of the way he constantly needled the supervisors of the rigidly run company. Once, when top management decided it didn’t befit the corporate image to have its men wearing sandals, a memo was sent down stating that males at TI could not wear open-toed shoes. Doddington fired back a caustic memo asking why the rule shouldn’t apply to women as well.
Although Texas Instruments-made up of 77.000 employees, 25.000 of them in Dallas-has a reputation for being closemouthed and aloof with the media, strict in its rules and very poor in employee relations (“(hey expect a lot from you and don’t ever feel a need to pal you on the back.” said one TI lawyer), the company executives were intrigued by Doddington. “They’ll forgive some things, especially if you’re brilliant like George.1* said one TI middle management official. “We never knew what he was going to do, but we also felt like he was one of the great assets to our company.”
In the early Seventies, however, the big high-tech companies like TI weren’t sure what a speech scientist would ever do for them. It was hard to imagine when a good consumer product would ever come out of computer-speech research. Computers were able, through a voice synthesizer, to say certain words back to someone who typed in the proper commands, but that had little practical application. And although a lot of acoustical engineers were predicting the day would come when a machine could understand human speech, the truth was that no one was getting any where in figuring out how to get there. In 1969, John Pierce, then director of Bell Laboratories, likened such projects to “schemes for turning water into gasoline.”
At TI. Doddington worked on a couple of governmental projects, such as a security system for the Air Force whereby a person entering a room would say his name out loud for a computer to verify. A year later, another young Ph.D. engineer, Gary Leonard, an Oklahoma native whose father was a well-respected engineering professor at Oklahoma State University, came to TI. Painfully meticulous, Leonard was perfect for the marathon hours of trial and error required to teach a computer to recognize words. Doddington fondly called him “the staff computer guru.”
“For a long time,” said Gene Helms, another speech researcher who later came to work for TI, “there were just these two guys, Leonard and Doddington, off in a corner, virtually forgotten by TI mainly because the company didn’t have a due what they could do,” At one point, Doddington felt he had become lost in the bureaucracy and was being ignored. For an entire year, he would come to his laboratory and do his research, but no supervisor ever asked him what he was working on.
Doddington was fiercely independent, Even when Mark Shepherd, the colorful chairman of TI, known for his table-pounding temper and near-fanatical pursuit of new projects, picked Doddington to figure out how to design an automatic dictation system that would print out whatever a person would say to it, the scientist balked. Not that he didn’t like the concept. For a quarter of a century the idea of a “talkwriter” has captured the imagination of researchers. Doddington has even called it the “Holy Grail” of computer speech research. A machine that can automatically transcribe human speech will forever alter the information age- and make its inventor very famous. Almost every electronics company is racing to be the first to learn how to make such a machine.
But Shepherd or no Shepherd, the technology for a “talkwriter,” Doddington said, wasn’t complete. Showing a stunning disregard for the top corporate structure, Doddington fought the project almost every step of the way. He and his staff began sarcastically calling the idea “the Mark Shepherd Memorial Project.” Researchers such as Gary Leonard, whose quiet, reserved nature contrasted sharply with Doddington’s aggressiveness, would delight every time another Doddington memo came zinging out. headed for management.
The battle of the egos didn’t help the speech team’s standing at the company. They remained at the lower echelon of TI research, seemingly with nowhere to go, and sure that their jobs were in jeopardy-until 1976. Then everything dramatically changed.
In that year, with a mere children’s toy called Speak & Spell, Texas Instruments revolutionized the whole state of computer-speech research, and George Doddington became a star. And the breakthrough set in motion the events that would eventually lead to the arrest of Doddington’s two close associates.
IN LATE 1976, AN IDEA WAS BROUGHT to Doddington about a talking toy that would help children learn to spell. The concept was simple enough: develop a machine that would vocally “ask” for words to be spelled. Then, after the child punched in a response, the machine would tell him if he was right. It seemed like a solid idea, though rather unspectacular, and maybe something that could even make the company a little money.
But at this point in computer-speech research, no one had figured out an inexpensive and practical way to “synthesize,” or create speech through a computer. It took too long, it was far too costly, and the machinery was too bulky to permit a viable consumer product. In order to synthesize speech, a computer must first translate a word into digital signals, which it then reinterprets into sounds. Most top speech researchers thought this technology would not be functional in products until the Eighties-but Doddington thought he saw his chance to prove them all wrong. Mainly through the work of some of the company’s top researchers, Richard Wiggins, Paul Breedlove, and Larry Brantingham, the TI speech staff tried to put the entire “synthesis” process, which included 256,000 bits of digitized data, on a pair of computer chips the size of a thumb tack.
The success of the whole undertaking seemed to be a long shot at best, especially given Doddington’s incessant fights with management. “He combatted everything,” said Gene Helms, another key researcher on the project, “from the color of the wire to the size of the loudspeaker. He simply would go wild if the project directors disagreed with him. I don’t know how many times he wrote sharp memos announcing the demise of the whole project because of idiots outside his control.”
“All I can say,” Doddington explained, “is that I like to operate in an atmosphere where logic is the primary judge of action, rather than politics or other extraneous matters.’”
Working to exhaustion in their cluttered offices, the half-dozen youthful speech researchers were consumed by this one-pound, hand-held mechanical spelling teacher. In a scant year and a half, an unusually short time for new product research, Doddington and company finished their work.
Speak & Spell turned out to be one of the most profitable inventions that TI management could have ever wished for. Most astonishing was the cost: a computer speech synthesis product for $69.95. A triumphant Doddington flew with some of his staff in 1978 to an international computer-speech conference in Hawaii. His clothes looked ridiculous (shorts, Hawaiian shirts, and black wing tip shoes), but he was the hit of the conference. Everyone was asking him what he had known that they didn’t. Back in Dallas, senior management officials named him a Tl Fellow, an honor given only to forty top researchers in the company. They moved the speech research team into a new set of plush offices and, even better, bought them a new $500,000 computer.
The speech computer chip that was developed (and later imitated by other companies) was quickly incorporated into everything from automobiles (to remind drivers to lock the door or turn on the lights) to elevators (to tell people which floor they were on). The chips were put in home computers and video games.
But that wasn’t enough. In 1979, Doddington began expanding his staff to a dozen people, almost all of them in their late twenties or early thirties, to begin a full-scale assault on The Problem: getting the computer to recognize spoken words. The new recruits were at the top of their field in electrical engineering and computer research, and they were shockingly bright, discussing their projects constantly, using terms and mathematical equations no outsider could fathom.
One of the first new hires was a completely different kind of researcher. Tom Schalk, a recent Ph.D. graduate in biomedical engineering from Baltimore’s Johns Hopkins University, was seeking to understand the way speech signals are handled in the brain, and Doddington found him fascinating. Together, he and Schalk published a widely read and controversial paper in one of the computer-speech industry’s most respected journals, and they were also credited with devising an important computer program that could be used in personal computers so that the computers could talk, and, to a limited extent, understand speech.
But Doddington found Schalk fascinating in another way. Outgoing, energetic, the kind of guy who took his researchers out for a beer or for a game of racquetball, Schalk was the born leader that Doddington never could be or wanted to be. If the other engineers and computer experts in the speech program tended to come across as preoccupied, absent-minded scientists, Schalk looked like a former Boy Scout. Married to a beautiful Texas woman, Schalk kept the staff volleyball rosters on his personal computer file at the office and would send computer messages across the country to a friend, asking when they could go skiing in Vail. But Schalk could also chat easily with a fellow engineer bogged down in research. Eventually, Doddington would leave Schalk in charge of the lab when he went out of town. Schalk also began writing the lab’s weekly reports. “There was no question he was on his way up,” Doddington recalled. Another top TI official remembered, “You could take one look at him and realize he was a different breed than the usual engineer who came through this place.”
And so, in the central research laboratories of Texas Instruments, Doddington’s staff settled in to begin the long, wearisome process of teaching a computer to understand the human voice. The researchers would collect samples of hundreds of different voices, then program those voices into their computers, trying to discover that certain formula, or computer algorithm, that would allow the computer to recognize each speech sound.
Although the work sometimes seemed to go nowhere, there was a sense of urgency among the staff, because all the major hightech electronics companies were now on the hunt for the same kind of formulas. Bell Laboratories was hard at work, IBM was also making some noise about its voice-recognition machines, the Japanese were investing millions in their own projects, and several small start-up companies were forming as well-all because of the potential fortune that would come to the first one who could crack the code.
ONE OF THOSE NEW COMPANIES WAS LAUNCHED IN DALLAS in 1979. Doddington once called Voice Control Systems “purely a garage-shop operation ” and in truth, one could not say that VCS had the most auspicious of debuts. The founder was a smooth-talking real estate developer. Bill Bates, who was about to start serving ten months in prison on a drug smuggling charge. He hired an engineer named Bob Kirkpatrick, a self-described “maverick” scientist who worked on an Apple computer in one of the rooms of his home. Kirkpatrick began designing a voice-controlled robot to exhibit at trade shows. With little money coming from Bates and his small band of investors, the future didn’t seem particularly rosy, and no one at TI paid them any attention.
But things started to change in April 1983, when Tom Schalk shocked the speech laboratory at TI: he was going to work for VCS. The little company had taken out a large newspaper advertisement looking for engineers, and Schalk. who was getting bored doing often mundane research, decided to move. Like many at TI, he thought there was no room for quick advancement. He was thirty-one and he was moving into a role where his vision would help determine the company’s future. Schalk was off to become, in his obscure, scientific field, a star. “There was something sexy, so to speak, about the whole thing,” Schalk later said.
Doddington was stunned. He told Schalk he’d give him six months to try out VCS, and if he wanted to come back to TI. there would be a job waiting for him. The rest of the staff was puzzled as well, but in another sense, they understood all too clearly the need to get out from George Doddington’s control. He was, to put it mildly, not the best boss to work for. “George Doddington was sort of like the scatterbrained mad scientist,” said one top TI lawyer. “People would have to go in there and demand that it was time for their raise, but Doddington would be so consumed by some research project that he’d ignore them. One time, everyone got raises in the speech lab except George. The reason was he was too busy to ever get around to filling out his own performance review sheet that would have naturally led to his raise.”
According to testimony, the week Schalk left, the TI patent lawyer who was supposed to tell Schalk that he couldn’t walk out with anything that might be a trade secret, hardly mentioned it. They talked about racquetball. Schalk did sign a form vowing never to disclose any proprietary information he learned at TI, but the truth of the matter, court testimony revealed, was that everyone in Doddington’s speech laboratory routinely ignored TI management’s strict rules about classified information-even Doddington himself. He said in court that he didn’t know there existed a TI policies manual that explicitly stated how a trade secret should be treated. Doddington liked to think that his lab operated not so much in a competitive business climate but more in an academic environment-’free and open,” he called it. He also said on more than one occasion that the only way breakthroughs would ever come in computer-speech technology would be through the sharing of research.
So few suspected anything when Schalk jumped to the fledgling VCS. Only one person said something to Schalk that seemed a bit unusual. Roger Bate, the chief computer scientist for TI. had long ago taken a liking to Schalk; he saw him as the kind of young leader TI needed. Just before Schalk left. Bate told Schalk, “I don’t want you to do what I think you’re going to do.”
Schalk-young, handsome, sincere-looked at Bate and said he wasn’t doing anything wrong.
A month later, another TI computer-speech researcher left for VCS. One month after that, one of Doddington’s top men, Gene Helms, who had invented one of TI’s most popular speech analysis products and helped devise the computer program for Speak & Spell, left to become vice president of VCS. By the end of 1983, two more Tl’ers, specialists in something called image processing, had come to VCS.
Though Doddington persistently argued that VCS was not a competitive threat, the new rival was making strides. In late 1984, an aggressive, hard-driving Dallas business executive named Neal Robinson led a team of investors who bought controlling interest in VCS. Robinson, well known in Dallas from his experience as a vice president at Braniff Airways and then president and CEO of U. S. Telephone, pumped several hundred thousand dollars into the company. Soon, the proud Schalk was showing off to other TI speech researchers a prototype of a cellular car phone that was activated only by voice commands. A person never had to push any buttons, but just say the number he wanted to dial. The innovation added to a driver’s safety.
Doddington’s people had nothing like this-one solid consumer product based on voice recognition technology. Schalk was right. There was something sexy about a small company. In all, eight others from TI followed Schalk to VCS. The last one to come, in February 1985, was Gary Leonard, who had been with TI’s speech program from the beginning. It was Leonard’s resignation, more than any of the others, that finally got to Doddington.
“Gary was (he first person I intruded upon,” said Doddington about his long-time colleague. “It was the first time I told someone that he was making a mistake.” Doddington told Leonard, who would lead the research for the VCS team, that it was a crazy idea. Recalled Leonard: “George said to me, ’You know, TI can crush VCS.’ I said, ’Yeah, I know that. I know it’s risky. But I think the company is doing the right thing.’’”
The two men were not as close as they once were. In the year before Leonard left, Doddington had criticized him for not publishing enough in scientific trade journals and had written on a job evaluation that Leonard had been “coasting.” Doddington said later that Leonard “was not my best employee. He was sort of a C student.”
But it was Leonard’s work in putting together computer data bases consisting of different voice sounds that had helped Doddington become famous. It was the kind of job, as Doddington admitted, “that depended on him not making mistakes.” Leonard, in fact, was the first to begin piecing together what Doddingion called “the most advanced recognition system TI has developed.” Doddington began to work closely with Leonard, and soon the two produced a system that would recognize ten vowel sounds.
When Leonard quit, Doddington assumed that he was leaving because “he wanted to be a hero” in the computer-speech field. But. he said, it never occurred to him just how Leonard was going to go about it, not even when he learned that on Leonard’s final day at TI, he was at the office at 1:30 in the morning, doing some kind of work on the computer. “I guess I’m an absent-minded professor,” Doddington said. “I never worried, never suspected.”
But that was before April of 1985. when one of the VCS engineers, an odd, mysterious man who formerly worked at Texas Instruments, got in touch with TI’s security staff. And that’s when it all blew wide open.
SAM KUZBARY WAS BORN IN SYRIA, WORKED ON A TOP SECRET military project as an electrical engineer, and then in 1983 showed up in the United States seeking political asylum. Kuzbary left Syria rather surreptitiously, and many of the lawyers who were involved in the TI-VCS case assumed the CIA helped get him out to see what he knew about the Syrian military. In 1984, TI hired him to work in the corporate engineering center, but four months later he was fired because management decided he was a security risk. Among other things, it was learned that Kuz-bary kept a gun in his car.
But one of the computer people at TI did recommend Kuzbary to VCS. saying he was a good researcher and might be able to help out in VCS’s speech program. Kuzbary worked at VCS for eleven months, and then on April 25, 1985, he called a security executive with TI and said he wanted to talk about some material he believed was stolen from TI. His motive was simple. He said he wanted his job back at TI, and he wanted to prove that the security problem was not him, but native American citizens.
A few days later, Kuzbary met with a group of TI security officers and showed them a listing of the files that were in Schalk’s and Leonard’s computer directories, He gave them some documents and a tape he had found in boxes and shelves in the two men’s offices. The computer printouts were shown to Dod-dington. There, in the VCS computers, were listings of some of Doddington’s research files, even copies of some of his memos to management. Included in them were the special computer programs he and Gary Leonard had worked on, and even some of the early programs that had made Speak & Spell such a success. Doddington could barely speak. “I was flabbergasted,” he said. TI had spent between $15 and $20 million on speech research over the past ten years, and Doddington told the company’s security investigate rsinanffosMjiiMvas sitting over at VCS.
The word quickly got to the top executives of TI. They had been burned once before on a trade secrets case, and they weren’t going to let it happen again. In late 1981, a group of TI employees, led by TI’s top researcher in the personal computer department, broke away to start their own company, Compaq Computers, which was going to produce a personal computer of its own. TI filed a civil suit, claiming the employees were using trade secrets and proprietary information they had developed while working at TI. But in that case. TI had no strong evidence.
Eventually there was an out-of-court settlement in TI’s favor, but the damage was done. Compaq eventually developed a highly profitable personal computer. TI’s personal computer was a marketing disaster, leading to a $145 million loss for the company in 1983. Its entire personal computer division was ultimately disbanded. “If we had pursued the case against Compaq and had not settled,” said one lawyer for TI who was involved in the lawsuit, “I am convinced we would have won, taken them over, and they would have become a division of TI.”
So this time. TI would take no prisoners. Though TI officials didn’t rehire Sam Kuzbary. nicknamed the “Syrian snitch” by prosecutors, they did pay him $2,500 a month until the criminal trial began to ensure that he would tell his version of the VCS story. And they prepared a civil suit against VCS asking for a percentage of whatever income VCS made from the technology it developed. Neal Robinson of VCS claimed TI was simply out to ruin the small company’s reputation and litigate VCS out of business.
One of the lawyers who worked with TI in the case admitted, “The top dogs of TI were pissed off as hell about people walking out the door with TI information. And to be honest, they decided to make VCS an example of what could happen to anyone who tried it again.”
Regardless of the potential repercussions, the DA’s office made its raid early on May 16-and they went through everything. They accumulated five boxes of documents. They confiscated computer tapes and copied everything in the VCS computer, infuriating VCS officials, who were convinced that their own computer programs would end up in TI’s hands.
At one point, the investigators came across Leonard seated behind a computer terminal. He had sneaked into an office and was typing a command into the computer. Of course, after they got him away from the keyboard, the investigators couldn’t figure out what he was up to. But for the first time, many of them realized that they might be onto something big.
When it was all over, they found 7,985 computer files that had been copied from TI’s main computer. Tom Schalk had copied, before he left, 2,504 computer files. Leonard not only copied more computer programs from the speech lab, but he made a copy of TI’s private plans that outlined the direction of Doddington’s speech research over the next five years and told which products it would be applied to. The soft-spoken Leonard, the man who never made waves during his long tenure at TI, had gone up to the office late one evening, gotten into the personal computer directory of his old colleague George Doddington, and copied just about everything.
An open-and-shut case? Hardly. There was another version to this story, one that would soon come out in court, that TI officials didn’t even know yet-one that would throw into jeopardy all their efforts to stop VCS. It would also lead to a humbled George Doddinglon being stripped of his administrative leadership role in his beloved speech laboratory.
THE CASE WAS TRIED BY THE NEW CHIEF of the district attorney’s specialized crime division, thirty-five-year-old Ted Steinke, and assistant DA. Jane Jackson. Steinke was one of DA Henry Wade’s most articulate, sharp-witted prosecutors, and at the outset, the case seemed rather simple for him. Schalk would be tried for stealing five TI computer programs that allowed a computer signal to be turned into actual speech sounds, and Leonard would be tried for stealing five programs he and Doddington developed that allowed a computer to recognize vowel sounds. They had even admitted in a pre-trial meeting with the prosecutors that they had made copies of files on TI’s speech computer.
But the trial, which ended in September. was treated with extraordinary importance by Steinke because it was the first time the “trade secrets” felony statute was invoked in Dallas. And, as Steinke said, “You don’t want to always be known as the prosecutor to lose the first case of its kind.”
Steinke, however, could never have guessed just what he was going to run up against. He started learning on the first day of testimony, when a TI security official, under persistent cross-examination from defense attorneys Gerald Banks and Mike Carnes, said that company policy required all trade secrets to be marked with the words “TI-Strictly Private.” Doddington, of course, never did anything like that. He thought it was pointless. In effect, his employees had no clear criteria as to what was secret.
How could a man be accused of stealing a trade secret if the company itself didn’t even treat it like a trade secret? Suddenly, the defense was off and running. Several of Dod-dington’s ex-speech researchers testified that they couldn’t ever recall Doddington talking about what might have been a trade secret. One former TI engineer named John McClelland said he was required, when working in other TI departments, to go to periodic security meetings that explained just which part of their work was a “trade secret.” But at the speech lab it was never mentioned. “No one even mentioned security in general.” said another former TI speech engineer, Bruce Secrest.
A lot of engineers at TI, in tact, couldn’t believe that Schalk and Leonard were on trial. They subtly protested the matter by renaming a TI softball team that Schalk once played for as the VCS Trade Secrets. When the team finished second in the league and got its picture put in the TI company newsletter, a member of the team added Schalk’s face to the back of the group shot to taunt TI management.
The other problem was that Doddington encouraged his researchers to publish so much that some of the speech lab’s “sensitive” information ended up as public information. In fact, the company’s patent office only once notified the speech researchers that a certain computer program was a trade secret. But that very trade secret was published in Ph.D. dissertations by two of those who worked in the speech lab, so how secret could it be? Given Doddington’s policy of freely giving out some of Tl’s computer-speech research to competitors, it was easy to assume that the company didn’t go to much effort to protect its trade secrets in the speech lab.
“TI management went through the roof when it heard what Doddington was doing.” said one TI lawyer. “They had no conception how loose things were run down there.” Doddington was removed earlier this year from his administrative duties, although he did remain as the chief speech scientist.
But Doddington argued vehemently during the trial that he had often discussed with Leonard and Schalk the things that were , never to be published or released to other| companies. “It was well understood by everyone that our research product-our algorithms-would not be made available outside TI,” Doddington said at the trial. And further testimony revealed that the specific computer programs the two men had been indicted for stealing had never been made available to anyone.
“We’re not talking about school children,” said Steinke. “They knew the importance of the things they were taking.”
But there was an even more curious point. ’ The defense attorneys emphasized that VCS was not using any of the computer programs that Schalk and Leonard had brought over with them. Larry Brantingham, one of the senior scientists at TI who had been named, like Doddington, as a TI Fellow, even admitted the technology VCS was developing for its computer-speech system was different than what TI was using. The computer-speech scientist hired by the DA’s office to study what was on VCS’s computer discovered that VCS was working with different computer programs than those TI had.
It didn’t make sense. Why had these two men brought over nearly 8,000 computer files if they weren’t going to use them?
Both took the stand to tell their stories. Schalk charmed the jury with his demonstration of the VCS cellular car phone, and then he simply said he copied the TI speech directory to sift through at some later date in case he ever needed some of the TI programs that he thought had been released to the public domain. “I was not under the impression at any time I worked at TI that I was working with a trade secret.” Schalk said.
Leonard said he wanted to keep some of those computer programs that he had worked on for “historical value.” He said, “’Being a man with self pride and a souvenir collector, I wanted to have a record of what I had done [at TI].” Leonard claimed he went through Doddington’s personal file looking for some of the infamous Doddington memos he had enjoyed reading over (he years. Finally, he said he had to copy another computer directory because it contained the mailing list for his camera club and his Sunday school class roster.
Both men said they were utterly surprised to find certain computer programs, the ones that TI said were trade secret, in their own directories when they left TI. They also said their own sense of ethics wouldn’t allow them to ever misuse the TI files. But that’s where Steinke got them. Although both of them, when working at Tl, always asked for Doddington’s permission before they released some of the TI research programs to other individuals, they never once asked Doddington if they could take any of those 7.985 files when they left for VCS.
“One thing they can’t change,” said Steinke in his closing arguments to the jury in state District Judge Ed Kinkeade’s court. “They snuck these programs out without telling anyone.” What Leonard brought over in 1985, added to what Schalk carried with him in 1983. gave VCS the entire speech processing system that TI had spent years developing,
Added assistant prosecutor Jackson; “Do we treat these men any better because they have advanced degrees, because they have “Drs.” in front of their names? The law protects a company’s right to have property… and a guilty verdict is going to send out that message.”
The ten-woman, two-man jury had spent much of the three-week trial looking on with dazed expressions. It was a most complicated case-at one point the attorneys spent an entire day trying to determine what actually constituted a computer program. Almost everyone, from the defendants’ worried families sitting at the front of the courtroom to the serious Tl executives sitting in back, seemed exhausted when the testimony finally ended. Schalk and Leonard looked confident: Leonard even went back and sat in the witness box during a recess so his wife could take a picture of him as a keepsake.
But after a three-hour and twenty-five-minute deliberation, the jury convicted the two men. Schalk’s wife rushed weeping out of the courtroom. Leonard’s entire family broke into tears. Three of the jurors were crying as the verdict was read. After all, these were not unlikable men. “They were just the two guys that a big corporation decided to make examples of,” said one of their disgusted fellow researchers.
THE NEXT WEEK, BOTH LEONARD AND Schalk were back at work. They would later be sentenced to probation, two years for Schalk and four years for Leonard, and a $5,000 fine for each, avoiding a jail sentence (the two are appealing the case). So for them, the race was still on. Relentlessly, they were back trying to teach a computer to understand a human being’s voice.
“They’re still worried about us,” said VCS vice president Helms. “We’re at work on a technology that they aren’t able to do, and it’s going to pay off. They can’t stop us forever.”
Under a court agreement, TI made another search of the VCS offices to make sure they had no more TI documents (although some TI officials speculated that VCS could have made an extra set of copies of the TI computer programs). But TI had certainly proved a lesson to its employees, It would not tolerate someone walking out with TI information. The company also began studying ways to increase its security. In many ways. TI management was deeply embarrassed by the whole episode, One juror even said after the trial that the jury wanted “to slap TI’s hands as well” for its negligence in the case.
A subdued George Doddington was back at work as well. He had been stung by the trial-the defense attorneys spent a lot of time trying to browbeat him-and some of his old fire was gone. “I’ve always run my lab with the utmost respect for my colleagues.” he said. “We worked together on an equal basis, with the highest ethical standards, and now when one can’t trust his employees-well, I know it’s common, but it’s made things very difficult.”
The potential fortune to be reaped by their research remains elusive, tantalizing, just out of reach. Already, some companies around the country have released a few computer speech recognition products (VCS’s cellular telephone will come out this fall), but the big payoff is still waiting-which means the temptation to cheat one’s way to the top remains just as strong. “I don’t know what we’ve learned.” said one lawyer who was deeply involved in the case, “except that we all showed how easy it is to steal a company’s secret. And believe me. as the stakes go up in this war, I think you’re going to see a lot more Tom Schalks and Gary Leonards out there.”
On May 16, 1985, at a Denny’s restaurant on LBJ Freeway, the cops and the investigators gathered at 4 a.m. to review the final plans for what would turn out to be a most unusual raid. Anybody in the restaurant could have taken one look at the eight officers and guessed what was up. There was going to be another bust, and this one seemed big.