by Alec Foege
Life at PARC was never the same. Despite a recent decision to expand its research budget significantly, Xerox did not capitalize on the personal computer interface it developed. Instead, it tunneled millions into the development of silicon-based integrated circuits, despite the fact that such circuits were already readily available elsewhere.
Of course, PARC invented other technologies that would prove influential in the not so distant future, including the laser printer, Ethernet networking, the optical disc, and LCD technologies, among others. Unfortunately, Xerox was never properly able to capitalize on most of those innovations, either. The laser printer market was first developed by IBM in the mid-1970s, after Xerox delayed the sale of its own 9700 printer for five years while it endlessly debated the cost-analysis merits of doing so. Cisco Systems and 3Com later cornered the network hardware business, due to Xerox’s initial instinct to keep the technology secret, followed by its decision to license it to anybody for a one-time license fee of $1,000.
But the real consequence of the squandering of PARC’s tinkering capital was the effect it had on corporate research in the decades after. No large corporation today would permit its research arm to develop technology that was not tied to a specific, anticipated product. Not even Xerox. It seems counterintuitive that PARC’s success in producing new technologies, even if Xerox couldn’t properly capitalize on them, didn’t convince other large companies of the value of building their own internal innovation labs. But the risk was perceived as too great, and in most cases, relegated to the world of much smaller start-ups.
Since 2002, PARC has existed as a wholly owned subsidiary company of Xerox rather than simply a research arm. And while Xerox accounts for around 50 percent of its business, it has other clients including Samsung, NEC, and VMware. Many of the engineers and scientists who worked at PARC in the early days went on to populate then unknown companies such as Apple and Microsoft. PARC’s impact continues to be felt, but the lessons drawn from its failures continue to weigh on the corporate version of American tinkering.
Henry Chesbrough of UC Berkeley’s Center for Open Innovation has argued that PARC’s biggest problem was that it nurtured what he calls a “closed innovation paradigm” rather than an open one. PARC in the 1970s sought to create all stages of a new product’s development within the confines of its own corporate structure. It wished to own not only the tinkering environment but also the facility that developed the resulting innovations into products, the factories that manufactured those products, and the teams that ultimately marketed, distributed, and serviced the products. That had been the vertical integration model used by all of the top industrial companies in the post–World War II era.
But what happened at PARC, according to Chesbrough, offers proof that another approach was in order. As he meticulously documented, nearly all of the innovations devised at PARC in its most fertile period were later produced and sold at other, smaller companies populated by former PARC employees. Among the better known of those technologies were the Macintosh computer and the Bravo word processor (which later became Microsoft Word). Of course, many of the other companies built on innovations developed at Xerox failed. But that’s exactly the point. One company, no matter how large or well capitalized, is rarely equipped to handle all of the ups and downs contained in the normal course of development for technologies that are genuinely radical and new.
Xerox was not completely unmindful of the value of technologies developed at PARC; rather it decided not to develop technologies it felt were not worth its further investment. In most cases, the developers of these innovations left PARC with Xerox’s blessing and only after Xerox arranged for a hefty licensing fee.
But the licensing fees could never come close to matching the revenues realized elsewhere by the products that became huge commercial successes. They also did little to capture the value of the ongoing evolution of an innovative idea. Most of the technologies that left PARC during those years were not appealing ones, at least at the time Xerox gave its okay.
Chesbrough used SynOptics as an example. Founded by two former Xerox executives, Andy Ludwick and Robert Schmidt, in the mid-1980s, the company was created to develop and market a technology that provided high-speed Ethernet service over optical cables, an innovation developed by an individual PARC researcher, that allowed for much faster transference of data. The problem was that optical cables were not yet prevalent, and building the necessary infrastructure looked to be decades away. Furthermore, users of the new fiber-optics technology would have to rewire their offices with optical cables as well, to connect computers with printers and other devices. As a consequence, Xerox decided the research was no longer financially prudent to pursue, since the mainstream consumer market was nowhere near adopting it. As a smaller, independent company, Ludwick and Schmidt were prepared to wait until the market caught up with the technology; they amicably brokered an exit from Xerox, with Xerox retaining a 15 percent interest in the new company.
Shortly after breaking free of Xerox, SynOptics discovered that its technology worked nearly as well over the existing copper-wire infrastructure, providing substantially faster transfer speeds without the need for an overhaul of the infrastructure. It was the freedom to keep tinkering without the pressures of a corporate parent that allowed SynOptics to stumble upon this new application. This radically changed the way its product was developed and dramatically accelerated the company’s growth curve. Three years after its inception, SynOptics went public in October 1988. It soon was worth north of $1 billion and merged with a company that later became part of Nortel.
The original technologies developed by an inventor are often not the ones that ultimately make it to market and become commercial successes. Xerox’s research and development process was designed to capitalize on new products that it could market to its existing customer base, and naturally favored ones that related to its copiers and printers. However, those were not the technologies that necessarily had the most commercial potential. Rather they were the ones that best fit into Xerox’s preexisting system for developing new products.
At its heart, Xerox’s PARC incubation process failed to acknowledge some of tinkering’s self-evident truths. First, that innovation most often begins with the vision of an individual and it is only through support for that individual’s efforts that the new technology ever makes it past the early stages of development. Second, managing the technological risks of innovation has little to do with managing the associated economic risks. And third, perhaps most important, no matter how large and well funded a corporate incubator is constructed, it is at best a crapshoot as to what the benefits of that support system will be.
In the United States, this issue is perhaps more protracted than in other nations, since the individual’s urge to innovate is girded by our tinkering history. And yet Americans have built most of the biggest and strongest corporations the world has ever known. The contradiction inherent in these two realities has at times spurred the country to create great products (Apple’s first iPod, which Sony failed to imagine, is a good example).
But more than we’d like to admit, it also has resulted in stagnation, especially when corporations use their firepower to bombard any new ideas they find threatening to their existing businesses. This might explain why some truly radical technological innovations of recent years have emanated from foreign shores.
CHAPTER 9
A TRIO OF ALTERNATIVE TINKERING APPROACHES
CLEARLY, TINKERING IS A KEY TRAIT of the American technological firmament, but plenty of tinkering occurs elsewhere. In the following three parables, I explore some more recent developments I believe offer distinct insights into the ongoing evolution of the tinkering process. Two are about European tinkerers and the other is about a Chicago architect. Together, they illustrate a few elements I believe are largely absent from today’s tinkering landscape in the United States.
Deep in the bowels of the gargantuan Javits Center in Manhattan, a rumpled,
professorial-looking man in a gray suit and tie spoke from a podium in clipped, German-accented English to a room of fifty or so geeky conventiongoers on October 21, 2011. The occasion was the 131st biannual Audio Engineering Society (AES) Convention, but this keynote speech hardly seemed to garner much interest.
The speaker, who had a salt-and-pepper beard and moustache and wire-rim glasses, was Karlheinz Brandenburg, better known as the father of MP3. MP3, for those few who are unfamiliar with it, stands for MPEG-1 Layer 3. MP3 the most common digital format for compressing music so that it can easily and quickly be transported over the Internet. Without MP3 and its related formats, there would be no digital music, no Napster, no iPod, and the music industry would still be successfully selling compact discs under its old business model. By making music incredibly portable, the MP3 format pointed to a future in which information could flit around in the ether via cloud-based or wireless communications systems.
Little more than a decade ago, such concepts seemed fantastical, unlikely to have much of an impact in the near future. Now MP3s are so common and ever present that they seem almost old-fashioned, despite their relatively short history. This might explain the low attendance at Dr. Brandenburg’s lecture at the AES conference. To the scientists, audio engineers, and acousticians assembled in New York, MP3 had become the ordinary, the common language, and a revolution that had come and conquered.
Although some American engineers, including James D. (JJ) Johnston of AT&T Bell Labs, played a role in the development of the MP3 format, the story of how Karlheinz Brandenburg arrived at its ultimate codec is one of painstaking tinkering that might not have been able to occur in the current-day United States. This is not because Germany was some hotbed of human knowledge but rather because the right set of tools were made available to the right set of people for an extended period of time, thanks to receptive financial backers. There was an element of luck involved, as well. But the opportunity to make something out of luck is something that can be planned as much as anything else.
The story of MP3 is very much like that—luck and skill and opportunity operating in unison. As such, it is the tale of a tinkerer’s paradise.
Karlheinz Brandenburg was born in 1954 in Erlangen, Germany, the oldest of three children and the only son. From an early age, he was a classic tinkerer: he loved music, and so got an amateur radio operator license, which allowed him to communicate with others over the airwaves, and soon began to build his own amplifiers. By the middle of high school, he had launched his own business, selling amplifiers to his fellow students.
The second current of his early years was his membership in the Boy Scouts; Brandenburg was a Boy Scout for all of his childhood, and came to love participating in group activities, especially those that involved building something.
His love of music extended to playing musical instruments; he learned to play the recorder, the violin, and piano through formal lessons. He later taught himself to play guitar “for around the campfire,” Brandenburg told me. After showing no exceptional talent on any of the instruments he encountered, however, he solidified his role as a devoted and attentive listener.
By the time this budding tinkerer-entrepreneur-musician arrived at the Friedrich-Alexander University of Erlangen-Nuremberg, he had become something of a pragmatist, and, thus, enrolled in the school’s engineering program. A year later, however, realizing he still had a passion for numbers, he also enrolled in the university’s mathematics program. He graduated with a degree in electrical engineering in 1980 and one in mathematics in 1982. But Brandenburg quickly discovered that pure theoretical mathematics required a focus and dedication he was unwilling to grant them, so when it came time to pursue studies toward a PhD, he took a definitive turn toward electrical engineering.
In 1982, while still deciding what studies he might pursue, Brandenburg was asked by his PhD advisor, Professor Dieter Seitzer, to take over work on a project he himself had been pursuing for a while: how to transfer high-fidelity music over a phone line. Seitzer’s prime interest was not music but improving the sound fidelity of telephones, which use only a quarter of the sound spectrum heard by the human ear. This goal was of little use to Brandenburg; he did, however, become entranced with the idea of how much of an audio signal could be removed without the ear detecting any distortion. Seitzer was impressed when Brandenburg succeeded in completing the task the professor assigned to him.
Seitzer also had an important affiliation: when, in an effort to put Germany at the forefront of cutting-edge microelectronic systems, the Fraunhofer Society—a countrywide German research organization funded partially through the state, but primarily through government and corporate contracts—decided to establish the Fraunhofer Institute for Integrated Circuits (IIS) in Erlangen in 1985, it named Professor Seitzer as its founding director. As a result, Brandenburg almost immediately had access to a wealth of technical resources, including state-of-the-art signal-processing equipment, as well as a mission to help develop products that might one day earn the institute profits via its patents.
Over the next four years or so, he spent his time experimenting with different ways to compress sound by eliminating certain elements of the sound spectrum. The goal was to learn which parts could be removed with little or no discernable degradation of the listening experience. The higher the bit rate, or number of kilobits per second, a file uses, the higher the audio quality. Generally, the problems with music file compression arise in finding a suitable balance between file size (lower bit rates result in smaller, more portable files) and fidelity (which improves with higher bit rates and larger, more cumbersome files). Brandenburg’s innovation was to find the perfect formula for smaller file size and improved audio quality.
“The standard way of doing it was with a bit-allocation algorithm,” said Brandenburg, referring to the mathematical formula that helped translate a large quantity of audio inputs into a reduced set of signals, essentially eliminating certain frequencies in order to shrink the size of the file. “But what I did over time was create a much more flexible system.” It was more flexible because it took into account previous knowledge of speech and musical sounds, providing a more holistic template for compressing those sounds with losing their realistic qualities.
Brandenburg used the “analysis by synthesis” approach of speech perception, which allowed him to focus on how the human ear actually hears music rather than relying on a very generic bit allocation formula. He famously used a recording of folk singer Suzanne Vega singing her hit song “Tom’s Diner” to calibrate precisely the right frequencies that could be removed from a music file with noticeable loss in fidelity. “I picked ‘Tom’s Diner,’” he said, “because it is begins as an a cappella and then adds very metronomic rhythmic elements.” (According to Brandenburg, it is very difficult to conceal deteriorated audio quality of an unadorned human voice.) He also incorporated the clever use of filter banks, which isolate and highlight key elements of an audio signal, and other high-tech frippery to concoct his compressed music.
Despite his success in completing the task he had first been given by Dieter Seitzer, an early attempt to obtain a patent for his sound compression work was met with a good deal of skepticism. The patent examiner apparently told Brandenburg “there is no high-fidelity music at 128 kilobits per second,” he recalled, the common transfer rate of an MP3 digital file, meaning the distortion at that level of compression would render any recording unlistenable.
Brandenburg admitted that, as proud as he was at the time of the elegant programming that went into MP3, he was despondent about its future as a marketable music format.
With good reason. In 1983, the recording industry introduced the compact disc, a digitized-content format that could be played over and over with no noticeable degradation of audio quality. The CD, which packed ten times the fidelity of an MP3 file—1,400 kilobits per second versus 128 kilobits per second—onto a plastic disc with a diameter of 120 millimeters, seemed to have eliminated
the need for an easily transportable music format such as MP3.
Thrilled to have found a way to boost the revenues of an ailing industry, the major labels priced CDs substantially higher than the vinyl records they were meant to replace. Understandably, they had little interest in a competing digital format with lower audio quality.
Indeed, it took until 1989, a full six years after Brandenburg did his pioneering MP3 work, before there was even an inkling that there might be a practical use, and a commercial market, for the industrious student’s work. By then, Brandenburg was already wrapping up final work on his PhD.
That same year, Germany financed the Digital Audio Broadcast project as part of a European effort, known as Eureka 147, to develop a digital standard for broadcasting over the radio airwaves, which turned into a bake-off between the IRT (Broadcast Technology Institute) in Munich and the Fraunhofer IIS in Erlangen. Out of that challenge, which took place from 1989 to 1991, the MPEG-1 Layer 3, or MP3 format, the technology Brandenburg had developed, emerged as the best compromise between file size and sound quality, although the MP1 and MP2 formats continued to have their uses.
Also around that time, Brandenburg began a yearlong stint at AT&T Bell Labs in New Jersey, a descendant of Alexander Graham Bell’s research efforts and ground zero for the United States’ efforts to advance sound compression technology. He says he visited Bell Labs as an exchange postdoctoral researcher in part to see how such research operations conducted business in a country that rarely funded technological innovation to the extent his home country did. Brandenburg spent his time in New Jersey fine-tuning the MP3 codec alongside James D. Johnston, a prominent American audio engineer. He was deeply impressed by what he witnessed during the time he spent at Bell Labs. “It was like a university with famous professors,” he said, “but no students.” Even more surprisingly, despite its corporate underpinnings, he found the research being done at Bell Labs was “even more detached from the realities of the marketplace than at Fraunhofer.”