A History of Silicon ValleyTable of Contents | Timeline of Silicon Valley | A photographic tourHistory pages | Editor | Correspondence Purchase the book These are excerpts from Arun Rao's book
|
4. The Lab Magicians: Xerox PARC and the Innovation Machineby Arun RaoThe Creation of a Corporate Research Lab Xerox Palo Alto Research Center
(PARC) was the US’s most successful corporate research lab
in the 1970s. Researchers invented the personal computer, the graphical user
interface (GUI), the laser printer, and Ethernet networking
technology. Many agree that the secret sauce that made PARC so successful was
its highly talented employees. Six factors brought these people together in a
creative environment. First was Xerox’s seeming endless pool of cash devoted to
research. Second was a buyer’s market for talent. PARC was started in a weak
economy when after the Vietnam War the federal government was cutting back on
research staff. Third was the state of computer technology, which was at an
inflection point due to Moore‘s Law. Fourth
was its quality management, which knew how to hire the best researchers, give
them a broad mandate, and then let them play without directives, instructions,
or deadlines. Freedom to experiment was invaluable. Fifth were the premium
salaries that Xerox paid its researchers, about $30,000-35,000 in 1970, a nice
amount for a new PhD. Sixth was a paucity of startup opportunities; when PARC
was started, computer science researchers couldn’t easily find funding for a
startup, though that would change. While Xerox never
commercialized all the wonderful technologies at PARC, the company did earn billions from these
innovations, and so made its money back many times over. A handful of people
deserve credit for starting PARC. Jack Goldman, Xerox’s Chief Scientist,
submitted in May 1969 his proposal for an “Advanced Scientific & Systems
Laboratory” to pursue research in computing and solid-state physics. As Goldman
told Xerox execs: “If you hire me, you
will get nothing of business value in five years. But if you don’t have
something of value in ten years, you’ll know you’ve hired the wrong guy.” Xerox’s CEO, Peter McCullough, had the vision
and long-term good sense to approve and champion it. In 1969 McCullough had
Xerox purchase Max Palevksy’s Scientific Data Systems (SDS) for $920 million in
stock. It was a computer company with a second rate minicomputer product that
Xerox would divest years later. Yet McCullough wanted the company to explore in
that direction and he had Jack Goldman take the lead for PARC to create “the
office of the future.” Goldman first recruited some star managers. The most
important was Bob Taylor, a former ARPA director. Next, coming in early 1970,
George Pake accepted the job of director PARC and persuaded
Goldman to locate it in Palo Alto, California, near Stanford University.
Taylor had a gift for finding and cultivating talented researchers in
the computer science field. After the GI Bill paid for Taylor’s study of
psychology at the University of Texas, he eventually joined JCR Licklider at
the government-run ARPA. Around October
to December 1970, Arpanet, the precursor to the Internet, became formally
operational, with four nodes up and running. ARPA had a $14 million budget for
computer science research, more than the top 5 other grant-givers combined.
Taylor eventually became Ivan Sutherland’s deputy at ARPA, and then soon began
running the Information Processing Techniques Office (IPTO). Taylor would
approach the best computer science programs in the country and work with PhD
students and junior faculty to find cutting edge projects to promote, many in
the field of human computer interaction. Taylor was important because at ARPA
he funded the country’s first computer science grad programs at Stanford, CMU,
and MIT. He knew all the young researchers in the field and had their trust. He
knew enough to ask good questions and direct them, but was candidly not a
specialist and would not micromanage research. So Taylor built one of the best
professional networks in the field, and met people like Alan Kay, who said in 1972 that “90% of all good things that I
can think of that have been done in computer science have been done funded by”
ARPA. The ARPA model was to find good people, give them a lot of money, and
then step back. If the researchers didn’t deliver in three years, they were dropped.[22] Alan Kay was one of the
spiritual leaders of PARC. In July 1969, Kay’s doctoral dissertation, “The Reactive Engine,” was
accepted at the University of Utah (he only got into the PhD program because Don
Evans, the director, never looked at grades). Within Kay‘s paper were early descriptions of his “Dynabook”
personal computer, basically an early laptop.
Kay was a non-stop
idea machine; half of which were brilliant and unworkable, the other half could
be tested and be revolutionary. He had been a child prodigy and pure motion -
he could never sit still. Kay hated the
time-sharing computer terminals that everyone had to use at that point. Whether
it was a mainframe or a minicomputer, you had to share it and they had blinking
green text and were only accessible to a nerdy few. Kay wanted an
interface children could use, more like finger paints and color TV. As PARC took off, the
1970s was a tough decade for Xerox. In 1970 IBM brought out its
first office copier, ending Xerox’s historic monopoly and introducing a period
of painful retrenchment at Xerox. Also as some execs tried to later kill PARC,
Xerox board member and Nobel laureate John Bardeen (co-inventor of
the transistor) fought to save PARC in board meetings, believing the $1.7
million budget was worth it. Douglas Engelbart and SRI’s
Augmentation Research Center Before delving more into PARC, it’s important to understand its neighboring
institution, the Augmentation Research Center (ARC), and Douglas Engelbart. Near the end of World War II, Engelbart was midway
through his college studies at Oregon State University when the Navy drafted
him. He served two years as a radar technician in the Philippines, where, on a
small island in a tiny hut up on stilts, he read Vannevar Bush‘s 1945 article “As We May Think.” Bush wrote about computing and a future when
a “memex” device would augment human intelligence. A human could use it to
store “all his books, records, and communications, and which is mechanized so
that it may be consulted with exceeding speed and flexibility.” Engelbart’s experience as a radar technician convinced him that
information could be analyzed and displayed on a screen. He dreamt of knowledge
workers sitting at display “working stations,” probing through information
space and harnessing their collective intellectual capacity to solve problems.
Engelbart returned to
complete his Bachelor’s degree in Electrical Engineering in 1948 and he got a
PhD from UC Berkeley in 1955. After
a year of teaching at Berkeley as an Acting Assistant Professor, he took a
position at the Stanford Research
Institute (SRI) in Menlo Park. In October 1962, Engelbart published a key
document about computing and his beliefs on the modern workplace: “Augmenting Human Intellect: A Conceptual Framework.” At SRI, Engelbart had a dozen
patents to his name and he proposed research to augment the human intellect
using computers. ARPA, a US government research agency, funded him and he
launched the Augmentation Research Center (ARC) within SRI. ARPA gave the team
funds to explore Man-Computer Symbiosis, plus technology for “time sharing” of
a computer’s processing power between a number of concurrently active on-line
users. Engelbart and his team
developed computer-interface elements such as bit-mapped screens, the mouse,
hypertext, collaborative tools, and precursors to the graphical user interface
in the mid-1960s, long before the personal computer industry did. At that time,
most individuals were ignorant of computers; experts could only use mainframes
with proprietary systems and difficult-to-master text interfaces. After two
years of unproductive work for ARPA, Bob Taylor at ARPA funded
a project to experiment and evaluate various available screen selection
devices, or pointers, for use in on-line human-computer interaction. Taylor’s ARPA grant led to the modern computer mouse.
Engelbart conceived of
the device and Bill English actually built
the first wooden prototype. In 1967, Engelbart applied for a
patent (with Bill English) for a wooden shell with two metal wheels: a computer
mouse (US Patent 3,541,541). They
described the device as an “X-Y position indicator for a display system.” No one at the lab remembered who gave it the
name “mouse,” but someone did because the tail came out the end. Sadly, Engelbart and English
never received any royalties for the mouse. SRI held the patent but had no idea
of its value; it later licensed the mouse to Apple Computer for
about $40,000. A year later, Engelbart gave the
“Mother of All Demos.” On December
9, 1968, Engelbart and his group
of 17 researchers gave a 90-minute, live public demonstration of their work. It
was at a session of the Fall Joint Computer Conference held at the Convention
Center in San Francisco, attended by about 1,000 computer professionals. A
number of experimental technologies that have since become commonplace were
presented. It was the public debut of the computer mouse, hypertext
(interactive text), object addressing, dynamic file linking, video
conferencing, teleconferencing, email, and a collaborative real-time editor
(where two persons at different sites communicated over a network with audio
and video interface). A year later, Engelbart‘s lab became the second node on the Arpanet (the
predecessor network that evolved into the Internet). On October 29, 1969, a
link was established between nodes at Leonard Kleinrock’s lab at UCLA and Engelbart‘s lab at SRI. Both sites would serve as the backbone
of the first Internet. In addition to SRI and UCLA, UCSB and the University of
Utah were part of the original four network nodes. By December 5, 1969, the
entire 4-node network was connected. Engelbart‘s ARC lab soon became the first Network Information
Center; it managed the directory for connections among all Arpanet nodes. One
could say that Engelbart‘s lab in Palo Alto was the
physical home of the most important Arpanet/Internet node for its first few
years. During his time at SRI, Engelbart developed a
complex philosophy about man improving through technology, a sort of
co-evolution through human-computer interactions. Engelbart was strongly
influenced by Benjamin Lee Whorf’s hypothesis of linguistic relativity. Whorf
argued that the sophistication of a language controls the sophistication of the
thoughts expressed by a speaker of that language. In parallel, Engelbart believed that
the state of current technology controls people’s ability to manipulate
information. Better manipulation led to more innovation and new, improved
technologies. People could even work in groups, where the collective IQ would
be larger than the sum of the parts (witness the modern laptop, created by
teams of specialists using other computers to design and prototype a laptop’s
different components). Engelbart pithily stated
to Reader’s Digest: “The rate at which a
person can mature is directly proportional to the embarrassment he can
tolerate. I have tolerated a lot.” He
was paid more by Reader’s Digest for this quote than for his many inventions. By 1976, Engelbart slipped into
relative obscurity. Some of his ARC researchers became alienated from him and
left to join Xerox PARC. Engelbart saw the future
in collaborative, networked, timeshare (client-server) computers, while younger
programmers preferred working on personal computers (individual machines that
would not be shared and controlled by a centralized authority). Eventually
funding from ARPA stopped by 1977 and SRI transferred the lab to Tymshare,
which tried to commercialize some of Engelbart’s software. However Engelbart was
marginalized and relegated to obscurity. Management, first at Tymshare, and
later at McDonnell Douglas (which took over Tymshare in 1984), liked his ideas
but never committed the funds or the people to further develop them. Engelbart retired from
McDonnell Douglas in 1986 and in 1988 founded the Bootstrap Institute with
modest funding to promulgate his ideas. Hiring the Best Computer Scientists Around On July 1, 1970, Xerox’s Palo Alto Research Center
(PARC) officially opened its doors at 3180 Porter Drive,
near Stanford University. For
the location, Yale’s New Haven was the first choice, but Goldman was put off by
the snobbery of the university and its hostility to enterprise next door to it.
Berkeley had no
dedicated real estate near the campus, and Santa Barbara had no large airport.
The physical and cultural climate in Palo Alto helped. Pake had hired Bob
Taylor to help him
staff the Computer Science Lab. Taylor forced the researchers to build things
they could use daily and avoid prototypes and playthings that just sat on a
shelf. He described his position at Xerox like this: “It’s not very sharply
defined. You could call me a research planner.”
Taylor made two key hires for PARC. First, in November he hired the engineers of the
failing Berkeley Computer
Company, including Butler Lampson, Chuck Thacker, and Peter Deutsch. Second,
Taylor raided Doug Englebart’s lab at SRI’s Augmentation Research Center, where
there was no desire to make a product or prototype, but just to search for
knowledge. Bill English, a brilliant hardware engineer, left for PARC and
other Englebart protégés followed. PARC’s first big project came from a corporate squabble.
The researches decided to build a clone of the DEC PDP-10, which
was the standard minicomputer machines of the time that all the researchers
wanted. Xerox had tried to
force them to take an inferior SDS machine because Xerox owned SDS. Instead,
the PARC researchers lost the battle to buy a PDP-10 but won the war by just
ordering parts and putting together a PDP-10 clone. It was a great bonding
exercise and a waste of one year. They called it the MAXC as a comeuppance to
Xerox management and the poor products that Max Pavelksy’s SDS made. The PARC researches were
tinkerers and hackers. They liked to make things. Generally the office had a
feeling of collegiality and a grad school environment. It had lots of informal
collaboration or “Tom Sawyering,” with someone proactively setting forth an
idea or project and then convincing others to join to attack it. If the problem
or project got momentum, the ad hoc team could spend 3-6 months on it; if not
everyone dispersed and looked for something else. One project was to make
replicas of the expensive Bose 901 speaker systems, where a set cost $1,100.
They reverse-engineered the speakers and made 40 pairs for the team at a cost
of $125 per set. Alan Kay once said “a
true hacker is not a group person. He’s a person who loves to stay up all
night, he and the machine in a love-hate relationship.” Hackers were nerdy kids who were smart but
un-interested in conventional goals. Computing was ideal because no credential
or PhD was required and coders could be independent artisans, selling directly
to customers based on the quantity and quality of output and not pedigree or
something else.[23] One PARC institution was
“Dealer,” a weekly meeting in a lounge with sofas and bean bag chairs at lunch
time, usually Tuesdays. Attendance was mandatory for the computer science
researchers. It began with housekeeping, and then one person would be the
“dealer” and take over, to set a topic for discussion and rules of debate.
Topics were unconstrained, like how to take apart and re-assemble a bike, how
programming algorithms are similar to kitchen recipes, or a presentation on the
sociolinguistics of the Nepalese language and culture. Discussion and blunt
talk were common, with people calling each other out with ejaculations like
“bullshit” and “nonsense,” not to mention denunciations like “That’s the
stupidest thing I’ve heard” or “It’ll never work.” It was a feral seminar, a match of intellects. By the summer of 1972, Kay and a
hand-picked team completed the first version of their revolutionary
object-oriented programming language, Smalltalk, which would heavily influence
such modern programming systems as C++ and Java. Kay had the idea in
a shower in Building 34 on the Xerox campus for an
entirely new syntax of computer programming based not on data and procedures,
but on “objects” that would be discrete modules of programming. Object-oriented
languages are easier to code in because as a program becomes more complex, much
complexity is kept within an object. So a programmer can manipulate the program
more easily and stick to the big picture rather than getting lost in the
granular code. Because anything could be an object, like a list, word, or
picture, Smalltalk did well for a graphical display. It was the language that
enabled the Alto to be really
useful. Around that time, Stewart Brand wrote an article about
PARC titled
“Spacewar.” It was about
a game called Spacewar on their computer which joined computers and graphic
displays. It was play, part of no one’s scheme or theory, and just done for
competitive fun. Yet it encompassed many of the things the researchers were
trying to do for computing. As Brand noted, it was interactive in real-time,
used live graphics, served as a human communication device, was on stand-alone
computers, and was quite fun in a way that only games could be.[24] The PARC researchers
would go on to make numerous devices that lived up to these principles. The Miraculous Inventions of PARC In early 1971 Gary Starkweather transferred from Xerox’s other research lab in Rochester to PARC, bringing with him the concept of the laser printer.
Starkweather was a scientist outcast at the other lab in Webster, where he
created a laser technology to “paint” an image onto a xerographic drum with
greater speed and precision than ordinary white light. In November 1971
Starkweather completed work on the world’s first laser computer printer. He had
modulated a laser to create a bit-mapped electronic image on a xerographic
copier drum. The commercial project was approved and killed three times, saved
only by Jack Lewis, a Xerox executive who ran the printing division and ignored
orders. In 1972, the Lawrence Livermore Lab in Berkeley put in an order
for the printers, which Xerox declined to fulfill (too low production
run-unwilling to create an early adopter market). A corporate committee decided
to delay for three years until a conventional high-speed printer, the 9000
series, was made and sold. The Xerox 9700 laser printer only came out in 1978,
and that was after Burroughs showed it in a demo at the Hanover Messe. The
laser printer and its successors would generate billions in sales. In September 1972, after MAXC was completed, Thacker
and Lampson invited Kay to join their
project to build a small personal computer. The machine would be known as the
Alto, and have a keyboard, screen, and processor in
portable, suitcase-sized package (it would later have a mouse and GUI interface). The
idea was that processors would be cheap enough in 5-10 years for every person
to have their own “personal computer” instead of sharing time on an office
computer. In November 1972, Thacker began design work on the
Alto. The original plan was to make 30 units for the PARC computer science
lab. The screen would be 8.5x11” to mimic paper and the projected cost was
$10,500 per machine. In the end, Xerox made 2,000
Altos at a cost of about $18,000 per machine, which fell to $12,000 after a
high-volume program was put in place. There were some technical innovations
like micro-parallel processing (to shift the memory access problem to the
microprocessor) and a new high-performance display that used less memory (and
so allowed the user to actually run apps). Meanwhile, in June 1972, Bob Metcalfe encountered a
technical paper by Norman Abramson describing Hawaii’s ALOHAnet, a radio
network. Metcalfe would use several principles in that paper while designing
the first Ethernet, a computer networking technology for local area networks
(it’s how most office Internet networks are connected even in 2010). A month later, Bob Metcalfe wrote a patent
memo describing his networking system, using the term “Ethernet” for the first time.
Metcalfe had come from Harvard after they rejected his doctoral thesis on how
networks transmit data in digital packets because it was “insufficently
theoretical.” He would later use the
concepts in that thesis to build a multi-billion dollar company and transform
the networking industry (he also resubmitted his thesis with more math, and it
was accepted). Metcalfe had a huge advantage over many researchers because he
was the Arpanet liaison or “facilitator” at MIT in 1971, and so saw the early
networking technical issues and he had valuable personal connections with
people on Arpanet. Instead of getting a university position after graduation,
Metcalfe chose Xerox for the high
pay, beautiful weather, and pure research freedom with no teaching
responsibilities or worry about tenure. Metcalfe hooked up MAXC to Arpanet, but
other local network proprietary systems were too expensive. Taylor had set
specs for a local area network linking the Altos whose cost was no more than 5%
of the computers it connected and was simple, with no complex hardware, and
that was reliable and easily expandable (didn’t want to splice cable all the
time). Metcalfe used Abramson’s paper, adapting it for Altos and building in
redundancy (a string of verification bits known as “checksum”) and an algorithm
to deal with interference. It would also require a physical line and Metcalfe
called it the Ethernet. Initially none of the Alto users wanted to
use Ethernet at a $500 cost, and it competed with “sneakernet,” that is, people
using hard disks and walking between labs with sneakers to transfer data. But
when an early version of Starkweather’s laser printer was connected to the
Ethernet, the “EARS” system was too valuable. Ethernet for the network, the
Alto for the personal computer, a Research character generator for early word
processing, and a Slot machine (the name for the laser printer) to make
professional paper documents. On March 31, 1974 Metcalfe filed a patent for
Xerox (awarded two
years later). He then quit for a job at Citibank, where he got higher pay and a
chance to work on its electronic fund transfer system. He was the first top
researcher to leave PARC. Bob Metcalfe of 3Com (2010) The Alto is the First
Personal Computer (PC) In April 1973, the first Alto became
operational, displaying an animated image of Sesame Street’s Cookie Monster.
The Alto was described in a memo in 1972 by Butler Lampson (himself inspired by
the “Mother of All Demos” of Doug Engelbart); Chuck Thacker was the main designer of the Alto.
Lampson’s memo had proposed a system of interacting workstations, files,
printers, and devices linked via one co-axial cable within a local area
network, whose members could join or leave the network without disrupting the
traffic. The Alto was
revolutionary because it was a personal workstation for one, not a room-sized,
time-sharing computer for many, meant to sit on a single desktop. It is
credited as being the first “personal computer” (PC) in a world of mainframes
(note that some would argue for other PCs being first, like the Olivetti P101).
The Alto had a bit-mapped display, a graphical user interface (GUI) with windows and icons, and a “what you see is what
you get” (WYSIWYG or “wizzy-wig”) editor. It also had file storage, a mouse,
and software to create documents, send e-mails, and edit basic bitmap pictures.
Also in April 1973, Dick Shoup’s “Superpaint” frame buffer recorded and storeed
its first video image, showing Shoup holding a sign reading, “It works, sort
of.” It was the first workable paint
program. The Alto got better as
PARC’s programmers built apps for it. In fall 1974, Dan Ingalls invented
“BitBlt,” a display algorithm that later made possible the development of key
features of the modern computer interface (overlapping screen windows, icons,
and pop-up menus which could be manipulated with a click of the mouse). This
was the desktop metaphor used by 99% of personal computers around the world
even in 2010. At the same time, Charles Simonyi, Tim Mott, and Larry Tesler began work on two
programs which would become the world’s first user-friendly computer word
processing system. The Alto, BitBlt, and Bravo basically created the modern
industry of desktop publishing, used by office workers around the world.
Ordinary people at home or work could turn out professional quality
newsletters, magazines, books, quarterly letters, and so on, faster and more
easily. Bravo, the word processor, has a fascinating story.
Charles Simonyi, a Hungarian computer science student who defected to
the US, was a key actor. His defection, as a side note, caused the Hungarian
government to fire his father from a teaching job at a Budapest engineering
institute, showing how the vaunted “Soviet science” system devoured its best
talent for idiotic political reasons. Simonyi built on Burt Lampson’s ideas for
holding an entire document in memory using “piece tables” to create an app
called Bravo. It was the first “what you see is what you get” WYSIWYG word
processor on a computer at a reasonable speed – a useful application. People
started coming to PARC to use it for
personal stuff like PTA reports, letters to professional bodies, resumes, and
so on. Their friends writing PhD theses wanted to use it. Larry Tesler and Tim
Mott improved the Bravo user interface to create something similar to the
menu-based interface people use in MS Word in 2005. It had features like “cut,”
“paste,” and so on, after watching how non-engineers actually interacted with
the interface. In early
1975, Xerox established the
System Development Division, as a stronger attempt to commercialize PARC technology.
More than five years later, SDD would launch the Xerox Star. Meanwhile, a Sante
Fe startup called MITS was selling the Altair 8800, a hobbyist’s personal
computer sold as a mail-order kit. It made the cover of Popular Electronics and
caught the attention of a generation of youthful technology buffs—among them,
Bill Gates and Paul Allen. In February 1975, PARC engineers
demonstrated for their colleagues a graphical user interface for a personal
computer, including icons and the first use of pop-up menus. This concept would
later be stolen by Steve Jobs and Bill Gates and be
developed into the Windows and Macintosh interfaces of
today. A month later, PARC’s permanent headquarters at 3333 Coyote Hill Road
formally opened. Others Commercialize on PARC Technology Due to one bad corporate decision, a billion dollar
product was lost. In August 1977, Xerox shelved a plan
to market the Alto as a commercial
project. It closed the door to any possibility that the company would be in the
vanguard of personal computing. If Xerox had followed through with its plan, it
would have released a PC in mid-1978, beating the IBM PC by three
years with a much better machine. The project was killed because Xerox
President, Archie McCardell, was an accountant who didn’t get technology. Also,
because of Xerox’s poor organizational structure, the Altos would have to be
made by a Dallas manufacturing facility that made typewriters. The managers in
Dallas just wanted to keep making the same product and get their highest
short-term bonuses. Xerox’s top execs
just didn’t get the Alto or PCs. They were used to a leasing business model
where customers leased a copy machine and paid annual fees for the copies used
based on the meter. Their fear was that if there was no print copy, “how would
Xerox get paid” over and over again? However, Xerox did sell some
early Altos running Bravo to the Carter White House in 1978, and eventually to
Congress for their offices. John Ellenby tried to more aggressively push the
sales of Altos. But after senior management interfered too much (over the
course of 3 years), Ellenby quit Xerox in 1980. He started his own company,
Grid Systems, making some of the world’s first laptop computers. At the same time, during a “Futures Day” at the Xerox World
Conference, Boca Raton, Florida, personal computers, graphic user interfaces,
and other PARC technologies
were introduced to a dazzled sales force. Other than the laser printer,
however, few reached market under the Xerox name. In June 1978, PARC scientists
completed the Dorado, a high-performance PC, and Notetaker, a suitcase-sized
machine that became the forerunner of a generation of laptops. The next month,
PARC made a mistake by starting a program in silicon-based integrated circuits
and building an expensive fabrication lab. Building a fab lab briefly put Xerox in competition
with Intel, a hardware component company which Xerox had no
business competing against (and Xerox never made money in that business). Xerox
was attempting to do something internally that it could do much better and more
cheaply by sourcing externally. In December 1979, two key events occurred. First,
Stanford University
Professor James Clark designed the “Geometry Engine,” the first 3-D computer
graphics chip and later the foundation of his billion-dollar company, Silicon
Graphics, Inc. He had used design principles formulated at PARC. The company’s chips allowed the computer-aided
design of cars, aircraft, roller-coasters, and movie graphics like “Jurrasic
Park.” Clark’s first test chip was built
by Lynn Conway at PARC, who came from IBM in 1972. She
had written a book with Carver Mead on VLSI chip design (how to pack more
circuits into a microprocessor). PARC then offered professors at a dozen
schools the use of PARC’s lab to create their own specialty microprocessors.
Clark moved to PARC’s offices and focused for 4 months in the summer of 1979 to
create his chip. At the same time Carver Mead went to Xerox headquarters to
suggest they do a better job of commercializing PARC technology. He
suggested Xerox set up an internal venture capital arm to fund startups with
technology made by their scientists. Xerox would take an equity role and have a
strategic position, while incentivizing entrepreneurial scientists to run
companies. Xerox declined. The second big event in December 1979 was when Steve
Jobs and a team of
Apple Computer
engineers visited PARC twice and took
copious notes. They came because one of Jobs’ key designers, Jef Raskin, had many relationships with PARC researchers and he
was impressed with their work. Jobs had signed a
deal with Xerox letting Abe
Zarem’s Xerox Development Corporation, a subsidiary, invest in Apple pre-IPO in
exchange for “marketing help.” It turned
out that the technology demos were much more important, and they gave Jobs a demo that no
other outsider had received at that point. After observing its hardware and
software in action, Jobs and his team
took steps to incorporate Alto‘s design principles and the GUI into the Apple
Lisa and Macintosh. Jobs even poached
some PARC talent, like Larry Tesler, who would eventually become Apple’s Chief
Scientist. In September 1980, PARC finally
released its first invention to the world for commercialization. Along with
Intel and Digital
Equipment, Xerox issued a formal
specification for the Ethernet and made it publicly available for a nominal
licensing fee. Ethernet quickly became
the networking technology of choice. PARC scientists also worked on an Internet
Protocol standard, called PARC Universal Packet, or “Pup,” which eventually
became a crucial part of the Arpanet standard known as TCP/IP. It became the
standard for much of the data passing through the Internet. At the same time,
John Shoch invented an early computer virus, a “worm,” which temporarily shut
down the entire network and all the Altos at PARC one day in 1978. Xerox did have a new
computer product; it just wasn’t a good one for the market. In April 1981,
Xerox unveiled at a Chicago trade show to wide acclaim the Star workstation as
the Xerox 8010 Information System, with a beautiful GUI and desktop
metaphor. It was the commercial offspring of the Alto and other PARC technology.
However, the Star was slow and cost $16,600. Moreover, customers needed to buy
2 to 10 at a time, and had to install Ethernet and a laser printer. The costs
were daunting. By August IBM unveiled its
Personal Computer, forever altering the commercial landscape of office
computing and making the Star obsolete. IBM’s machine only cost $5,000 and
didn’t have the pretty GUI. It didn’t have icons, windows, a desktop metaphor,
e-mail, or Internet; it crashed
randomly. Yet it was good enough for basic business tasks and apps and
it sold very well. Only 30,000 Stars were sold, compared to millions of IBM
PCs. PARC’s talent was frustrated and wanted to leave. Earlier
that year, Charles Simonyi was thinking
about next steps. Bob Metcalfe suggested he
talk to a 22-year old kid named Bill Gates who ran a
startup called Microsoft. Gates and Simonyi hit it off right away with
high-bandwidth conversations on the nature of computing, the role of
technology, and future product ideas. Simonyi felt the Xerox corporate brass
didn’t know much about technology and didn’t care – they were bean-counters,
ex-Ford finance people that McCardell had hired to run the company. Gates on
the other hand was a visionary and a first-rate, cut-throat businessman. As
Simonyi put it, “you could see that Microsoft do things one
hundred times faster, literally.” So
Simonyi left PARC for Microsoft, where he became a “messenger RNA of the PARC
virus.” Within six years, the market
capitalization of Microsoft was higher than
Xerox’s, and Simonyi plotted a strategy to exploit a range of markets that
Xerox fumbled on: word processors,
spreadsheets, e-mail, and voice recognition. Simonyi especially helped on the
project to create Windows, a first-rate GUI operating
system that competed with Apple. Another dispirited engineer who left PARC in 1981 was
Chuck Geschke, who was frustrated that Xerox wasn’t
commercializing their work. He went on to found Adobe Corp., a billion-dollar
company that used postscript, a typesetting language, to help computer users
make crisp, printable, presentable, and professional documents with text and
graphics. The company’s technology became the de facto standard of computer
typesetting and held that position in 2010. By May 1983, Apple introduced the
Lisa, a personal computer with a graphical interface based
on principles developed at PARC. Jobs joked that
Xerox couldn’t
compete with his scrappy startup because Xerox’s cost structure was too high.
The company was fat and bloated. As one Xerox engineer joked, “If we built a
paper clip it would cost three thousand bucks.” In September 1983, Bob Taylor resigned from
PARC under pressure.
Within a few months many of the center’s top computer engineers and scientists
resigned in sympathy. Many went to Taylor’s new employer, the DEC Systems
Research Center. In January 1984, Apple introduced the
Macintosh, the popular successor to the Lisa and the most
influential embodiment of the PARC personal computer, with a striking
“1984”-style television commercial during the Super Bowl. Did Xerox PARC Blow It? Why did PARC invent so many
great technologies and then fail to commercialize them? The first part of this chapter listed factors
leading to success. Now we turn to why Xerox failed at
commercialization. As Steve Jobs said in a
speech in 1996: “Xerox could have owned
the entire computer industry… could have been the IBM of the
nineties… could have been the Microsoft of the
nineties.”[25] One reason is that the company’s decision-making on
dozens of occasions was not about new technologies and opportunities, but about
personalities, politics, and short-term incentives. The second was that the company’s managers saw it as a
copier company, not as a computer or a publishing company, let alone an enabler
of the “office of the future.” The
managers were fixated on the leased copier business model, and the sales force
was trained on copiers and typewriters, not new office technology. Also the
purchasing managers for computers were professional IT people, not the managers
who ordered copiers. A third reason was that Xerox wouldn’t allow
entrepreneurial scientists to do spinouts and avoid the corporate bureaucracy.
New ventures had to be led by people running established divisions, people who
hated risk-taking. So Xerox lost talent like Clark, Simonyi, Geschke, Metcalfe,
and others who did startups that became billion-dollar companies much bigger
than Xerox. Finally, the fault lay with PARC itself, which
often acted as a pure research center. The scientists were generally far away
from customer development, sales, or intrapreneurial development. The few Xerox execs (not PARC
researches) who tried to commercialize products were crushed by the corporate
bureaucracy. So while PARC was a success at an innovation, it was mostly a
failure at commercialization. |
History pages | Editor | Correspondence |