Book Notes: Fumbling the Future: How Xerox Invented, then Ignored, the First Personal Computer
Fumbling the Future
Xerox established PARC in 1970, before there was a market for personal computers. No CD players, no Walkman, no cell phones, digital phones, VCRs.
In 1973, the Alto was made at PARC, the first computer designed for dedicated use by a single person.
In 1979, Xerox spotlighted the Alto with a television commercial. However, they did not opt to continue marketing the system. Once a competitor of IBM to be the leading office product company ten years earlier, it was no longer the same company; fierce competition, government antagonism, and economic recession all contributed to Xerox’s slide from overconfidence to loss of confidence. Internal factors between research, finance, and marketing groups each pursued different visions of Xerox’s future.
IBM introduced its personal computer in 1981.
Apple’s famous commercial was in 1984, and first soldered a computer together in 1973.
Peter McColough became CEO of Xerox in 1968. He had been with the company for years - in 1956, he was manager of marketing and in 1959 general manager of sales. When he joined, Haloid (the company name at the time) was spending all its money on R&D. McColough pushed Joe Wilson, president at the time, to build demonstration rooms, to build out a direct sales force (cut out the commission-based distributors), and service distribution centers.
In 1959, Haloid’s 13 year struggle with the development of xerography yielded the first prototype of the plain paper office copier. Knowing that building a logistics operations around the technology would be expensive and hard, it shopped itself to other better capitalized companies. IBM scoffed at the offer, thinking the number of units it would sell would be quite low. McColough and others didn’t want to sell the machine to other companies like IBM anyway.
Instead, the company issued more stock and later sold its 914 Copier in 1960. The year before, it’s sales were $32m. The year after, sales were $61m. By 1968, sales reached $1.125b. Over the same period annual profits soared from $2.5m to $138m.
Knowing that it was essentially a one-product company, McColough and Wilson looked to expand in three areas: education, medicine, and computers. By the late 60s they had made headway with the first two, but computers remained elusive. They could not find a company to buy that fit their needs.
Finally, they decided to buy Scientific Data Systems in ‘69, the only real game left in town. SDS had earned $10m on $100m in sales. Xerox paid 92x SDS’s earnings, giving them $900m worth of Xerox stock, an extreme number.
Invention, often a stroke of genius, is a hard and fickle. Tough enough, turning an invention into a business, or innovation, is arguably even harder.
Xerox (Haloid) found Chester Carlson’s invention of electrophotography and through years of R&D and working through thousands of problems to bring the invention into a sustainable, consumer-ready product. It took 14 years from the first day Wilson looked at Carlson’s electrophotography machine until the introduction of the Xerox 914.
At the same time, Joe Wilson had to convince the board and employees that xerography was the way to go for the future of the business. For 50 years Haloid had been in the photographic supply business. Xerography was a foreign concept even to employees.
Xerox also produced a variety of creative business solutions. They built a wall of protection through various patents, was one of the first companies to use television advertising for business products, and used a leasing program to amortize the cost of the expensive photocopy machines by charging per copy.
This last point was big: typically in its first year a leased machine would produce enough copies to pay for production, marketing, administration and overhead of the machine. That meant the next years revenues from the machine were mostly profit. And Xerox still owned the machine.
Jack Goldman was head of research at Xerox, arriving shortly after McColugh became CEO. From Ford, Goldman detested the sterile numerical analysis employed at Ford, following WW2. He say it as being a stuffy culture that stifled innovation that didn’t directly correspond with the bottom line. At Xerox, he saw the drive to believe in innovation (the 14 xerography R&D cycle) and the companies growth trajectory.
He was surprised when he visited the Webster, NY lab that they had limited digital capabilities. He was even more surprised when he was not consulted for the gigantic SDS acquisition, given he came with the desire to be in the room when big decisions were made. He figured it was just because he was the new guy. He came to see the acquisition as a testament to how serious McColough was about a new future for Xerox. In turn, he brought up the idea for a research center, which was unsurprisingly green lit.
In 1979, Xerox had 95 percent market share of the plain paper copying market. Peter McColough still wanted to change: patent protections would expire soon, and IBM and Kodak were each coming out with their own copier soon. Employees in the company were not so excited for change, believing that Xerox was diversified enough by having a hand in every industry. Being a copier company was part of their DNA. The numbers may have supported their case: copying revenues exceeded a billion dollars while the largest diversification program before SDS, the Education Group, had yet to reach $100m in sales.
SDS was projected to grow 27-32% compounding from 1968 to 1973, and in order to do so would need to reshape the narrow strategy they formally employed.
At the time, IBM controlled 70% of the computer market. SDS bypassed IBM by selling to technical markets rather than commercial ones. McColough felt that Xerox had the resources in sales, marketing, and cash to go toe to toe with IBM, unlike other companies that had previously entered the computer business. Research had always been part of Xerox’s story, and here it would help develop technology proprietary and unique to Xerox to help it sell to business customers.
McColough saw copiers and computers as part of the same business: delivering information in better, faster, and more powerful ways. He said, “the basic purpose of Xerox Corporation is to find the best means to bring greater order and discipline to information. This our fundamental thrust, our common denominator, has evolved toward establishing leadership in what we call ‘the architecture of information’”.
Research: The Creation of the Alto
Goldman tapped George Pake, a former colleague and friend, to build the research center from scratch. Xerox had a much different take on research from other companies, understanding that any results might take 5 to 10 years. Usually, corporate R&D was more myopic and focused on immediate marketing results and development.
After considering New Haven, Rochester, and Palo Alto, Pake decided to go with Palo Alto as it was close to Stanford and other chip companies in the Santa Clara valley.
JCR Licklider began promoting the concept of “Man-Computer Symbiosis” in response to the batch processing of computers and annoyances of the cycle of programming: punching cards, printing, and then submitting to the machine, and then doing it over again. Licklider eventually became director of ARPAs computer arm, the Information Processing Techniques Office. He promoted timesharing, where multiple users plugged into the same computer, streamlining the batch processing across many different actors.
Bob Taylor was a disciple of Licklider, with a passion for computer interactivity. He eventually worked under Licklider’s successor at IPTO, Ivan Sutherland, at the age of 33. He dreamed of the computer as something beyond an arithmetic machine. He hired Larry Roberts, you designed message switching to solve computer communication and built APRAnet, the first nationwide computer communications network. After becoming disillusioned with Vietnam, Bob Taylor left ARPA.
Pake brought Taylor in to ask advice about computing and PARC. Taylor, having a previous bad experience with SDS, thought it was silly Xerox would focus 1/3 of its lab on developing research for SDS and thought it’d be better to work on capabilities that would enable Xerox, already with a presence in offices, to convert analog to digital and move from a copier company to office information systems. He also thought SDS’s focus would not align with the focus of many top computer scientists, so Xerox wasn’t going to be able to recruit first-rate talent.
Pake hired Taylor to run the Computer Science Laboratory at PARC.
Taylor hired a young, colorful computer scientist named Alan Kay from the University of Utah, versed in Jazz and thesis writer of a computer system and language called FLEX, which discussed all the components of wide spread interactive computing. While originally scheduled to join Taylor’s CSL lab, Kay eventually moved into the Systems Science Laboratory also in PARC, run by Bill Gunning. Taylor made the recommendation for Kay to join SSL because it more aligned with his interests, but was also part of a long con where Taylor hoped to grow influence over that lab.
Taylor also recruited 27-year old Butler Lampson, who ended up writing the operating system for Project Genie. Lampson and other Project Genie alumni set up the Berkeley Computer Corporation, which began to run out of money in the 1970-71 economic downturn; Xerox was able to scoop up a bunch of top computer scientists.
After staffing the CSL, Taylor was not picked to run it by Pake. Given he had no technical experience, Pake thought some else might be better for the job. Taylor schemed up a plan for eventual control by convincing Pake to give him an assistant research director job and hiring in someone he was friendly with and with upward ambition as his boss.
The Alto was inspired by Bob Taylor’s idea of having single, dedicated computers connected by a network rather than time sharing. It was also an alternative to a limited of computers in the early ‘70s; people think fast and computers were slow. The computer and human experience should be integrated.
They decided to go with a raster screen rather than a calligraphic screen because calligraphic flickered, as the screen refresh rate was too slow. The raster bitmap took a lot of memory, nearly half in the first iteration. A decade later the amount of memory would be cheap, but in 1972 it was $7,000.
Another dilemma was that the control unit of the computer had to use cycles for input and output devices. This was tackled by other engineers in the 50s and 60s by moving logic to those devices, outfitting them with their own processors and memory, free the computers control unit. However, all this hardware cost money. To save on the Alto, Thacker invented multitasking, in which rather than the control unit having one input it could have up to 16, and control of the control unit would be based on priority of the input. This decreased hardware requirements by a factor of 10, made costs 60% less than a minicomputer, but caused the Alto to be about 3x slower.
It took under four months for the first Alto to be produced, arriving in April 1973.
The Alto had been the worlds first personal computer, but without an operating system or programming language or application package, the Alto was “no better than a hot rock - interesting but useless.”
Telephone communication used something called circuit switching, in which a series of operators and equipment ran connections from one node to another until a full circuit was formed. This took take a few seconds, an eternity in computer time. “Message switching” was invented, in which the trip was made with sub trips, A to E by A to B, B to C, etc. each node would send a message to the next node, which would then save the message and acknowledge receipt. It would then pass the message on (this was ARPAnet). Unfortunately, this was hard for the local network Alto required. PARC invented the “local area network”, or LAN. Their invention was Ethernet, in which the entire network was connected and messages were broadcast across the entire network, but machines only received dispatches carrying the correct address.
PARC also produced the first printer, an engineering feat by Gary Starkweather - originally from Webster - by converting an old copier machine to use lasers to write 23 million bits to a piece of paper within two thirds of a second.
To print these documents from the Alto, they needed to be composed on the Alto. Bravo was the first WYSIWYG product, a document editor for the printer. This was complicated for non-technical users, as typing involved using “command” syntax (think like vim).
One of the publishers in the Education subsidiary of Xerox was looking for a way to improve the publishing process. After reaching out to PARC liaisons, eventually a word processor without command syntax named Gypsy, a successor to Bravo, was built by two programmers working in overlapping 14 hour shifts. Non-technical folk got it right away, and it represented savings of 20-30%. It was the first time users outside of PARC really got what was going on. This was 1975.
Finance: The Rejection of the Alto
With the explosion in revenue and market power came the government knocking with anti-trust claims. Competition latched on, suing Xerox separately. They sought to end the leasing scheme and have Xerox divest its international join ventures, Rank-Xerox in England and Fuji-Xerox in Japan.
The stock price topped out at $179 and dropped all the way to $50 amid a large recession and high inflation, which are into Xerox’s profits. Their costs for payroll, interest, and materials rose from 60% to 70% of their revenue, costing the company $400m of income. Management did not effectively control costs with layoffs and had not accounted for a reduction in new leases, as the net install rate was about 1% over the year.
The giant SDS acquisition was also failing dramatically. Xerox wanted to go head-to-head with IBM, even though SDS’s strategy was explicitly to not go into the enterprise market but the technical / scientific one. Even though McColough espoused that he did not understand computers and this would have someone else lead, management still shifted SDSs profitable business and directed it to be head on with IBM. After a downturn caused its main market to evaporate and post a $71m loss over two years, McColough integrated the computer and copier divisions into a single, functionally arranged organization.
Aligning the two divisions watered down SDS even more. The executive in charge of the copiers and computers was located in Rochester (SDS was in Cali) and had a background in copiers. The revenues from copiers dwarfed SDS’s losses. The merger was doomed. The hope was that there would be cross-pollination and reduction of costs, but computers and copiers are very different products with very different target groups. With three functional groups (one handling engineering / manufacturing, one handling marketing / sales / service, and one handling planning), rather than per product group, there were often disagreements by the three operating heads.
For example, the Information Technology Group (manufacturing) assigned the SDS plant tasks unrelated to computers, which caused the plant to fail to meet project delivery schedules. SDS lost control of its costs, as it was sliced horizontally. The sales force for computers could no longer play hard, stripped of weapons like discounts in favor of monopoly-supported copier sales tactics. Expectations for computers was at the same level of copiers, leading to low morale. After years of operating losses, the decision to pull the plug on SDS was arrived at in 1975, with the acquisition overall costing the company 1.3b.
McColough was an energetic man, both within Xerox and without. He was deeply passionate about politics and social action, sitting on nonprofit boards and doing community service and taking positions in the national Democratic Party. Some were concerned that he was a slightly aloof CEO, not giving his full concentration to Xerox. Lawsuits also drew McColough away from the role for 6-8 weeks a year.
McColough brought Archie McCardell on as President and COO to run the business. He was an ex-Ford financial analyst, but affable and approachable. He implemented numbers-based strategies. But while the pairing of McColough and McCardell seemed to be a good fit for the company, communication to the troops was not a strong suit, with decisions often made behind closed doors and contradictory messages sent from the top.
In 1973, the pair decided to reopen the question of Xerox’s strategy. The commissioned a team to review the company’s current strategy and operations. The results were grim. The financial systems put in place by McCardell gave the illusion of control, but costs for personnel, materials, and product development were rising beyond reasonable rates. They worked at Ford, but not at a technology company where product development was critical and sales and service orgs dominated.
Maintenance only worked 75% of the time on the first try, causing unnecessary rollouts. 30% of new machine installs were aborted because of failed coordination between sales and service groups. 1/3 of total service visits could’ve been avoid with proper training. These sorts of problems flowed through the entire company across sales, engineering, finance, development, etc. McCardell’s systems were overwhelmed by a company culture of not paying too much attention to the budget, fostered after a decade of incredible growth and near monopoly status.
The group also predicted tougher competition in coming years, FTC notwithstanding. Other companies were getting in the copier game (IBM had 10% market share), and Xerox had continually raised minimum leasing prices to offset costs, new machines for bigger, faster, and more complicated, and operation costs became unwieldy. Customers that didn’t meet the minimum monthly number of copies would like switch, which represented 1/3 of customers.
McColough and McCardell had different objectives for the group. McCardell wanted financial cushion and capital for product development, looking toward insurance or financial services for expansion. McColough wanted another Xerox like product that could “pop”. The group came up 4 alternative strategies, deciding on one that prepared Xerox for the competitive onslaught by getting copier costs under control, and then diving into the office of the future.
Ultimately, McColough and McCardell didn’t choose any of the strategies, and the group was disbanded shortly after without much reasoning.
The plan from the start was that PARC and SDS would join hands to have the former research new ideas and the latter to develop them into products. But SDS was not built for that - it scoffed at research and worked in an industry where research was for their vendors, the hardware suppliers, rather than themselves. PARC’s Bob Taylor had philosophical differences than SDS’s chairman, and PARC as a whole believed SDS’s products to be mediocre.
Chuck Thacker’s multi-tasking design gave Xerox a 1-5 year head start before hardware costs came down enough for personal computer to be inexpensive. But the opportunity was never taken, missed by both PARC manager George Pake or corporate. It was obvious that PARC did not have the chops (or the desire) to commercial products
McCardell’s arrival brought with it more Ford analysts reporting to him. The analysts were put in charge of engineering groups, and not having a technical background, tried to enforce a system of control use numbers and formulas. They were very risk averse in a situation where risks needed to be taken and not all research / product development would work out.
Jim O’Neil, another ex-Ford analyst, was installed as the head of the Information Technology Group, the engineering / manufacturing division. He clashed off with Jack Goldman, head of research and originator of PARC. Their approaches toward product developments were starkly different - Goldman wanting almost startups, where product development mattered over all else, even with a half-baked scheme, and O’Neill who saw Xerox’s role as accelerating viable new products within the mature machinery of the company.
In 1973, McCardell and O’Neill appointed Bob Potter to head up an Office Systems Divison that would work on new non copier office products. Thinking that Rochester’s center of gravity was copiers, they wished to put the office elsewhere. It came down to SF and Dallas. Goldman and PARC lobbied for SF - a wealth of talent in the area, the SDS relationship had broken down, and a chance to connect PARC with an engineering group that could handle product development - but the financial analysis showed Dallas to save Xerox money. This could be pointed to as the one clear “fuck-up” as Dallas went on to develop a culture completely orthogonal to PARC.
Dallas and PARC got off on the right foot. Bob Potter presented the plan for Dallas to PARC, not mentioning software once. PARC scientists were enraged, thinking Dallas was completely missing the point of the future of the office and a waste of resources. Dallas thought PARC to be naive, untether to revenue or profit or costs, etc. Dallas had a charter to produce profit, to provide a return on investment within a year. PARC products would objectively take a few more years to be inexpensive enough for consumers. Rather than cooperation, Dallas and PARC turned into a rivalry.
Bob Potter was also unimaginative, only able to see the office as it was rather than what it could be. His 800 series word processor, though with a new daisy wheel technology that made it twice as fast as IBM’s, lacked other features and sold mediocrely in the field. Xerox convened a task force to study the word processor strategy and came back with a plan to use the Alto (the Alto III). However, a team sympathetic to Dallas reviewed and reversed the decision and went with the next Dallas product, the 850. Politics.
Marketing: The Reaffirmation of the Copier
Xerox had morphed into a large, listless corporation, an unwieldy bureaucracy where competition and politics within was more important than competition without.
Japanese competitors, notably Savin/Ricoh, entered the lower/end part of the market, flipping many of Xerox’s strategies on its head. This did not worry Xerox - while losing out on market share they still made a majority of the copies, which is what they charged by - but it provided a soft underbelly of the market for competitors to eat away at. At the top of the market, Xerox misjudged their marketing strategy and spent lavishly on developing a new high-end project that initially didn’t hit target numbers. Eventually it did, but by pure accident and a little bit of luck.
McCardell was eventually hired away by International Harvester, one of the 25 largest corporations in America. McColough saw an opening to replace him with a salesman, a person who could inspire 100,000 employees, instead of another numbers person.
That salesman was David Kearns, who was brought over during the transition from Haloid to Xerox, and had spent 17 years at IBM. He was seen as the primary combatant to O’Neill for company control. His take was a breath of fresh air, at a company meeting named the World Conference, iterated the important of fixing the bureaucratic bloat and need for honesty across the country. At that same meeting McColough stated that cost leadership, shorter product development cycles, more effective planning, and less bureaucracy were needed for success.
PARC employees hustled and planned for months to put on a big presentation at the World Conference. It went off without a hitch. But the executives that saw the products seemed a little underwhelmed (however their wives loved the products). After the conference, almost nothing happened. The only thing was the charter of the Advanced Systems Divisions, which had to both manage former SDS engineers and “develop markets with pre-products.” The SDD in Dallas was supposedly responsible for commercializing PARC products, so ASD didn’t have that responsibility.
Ellenby joined ASD and continued to push for the Alto. There were some early takers - there were 4 trials set to start in 1978 - but eventually Ellenby went around Elkind, head of ASD, and directly up to the President Kearns with a proposal for the Alto. This was exposed and management came down hard on Ellenby. However, his proposal was considered, but then rejected (likely due to the pull of the management tried that Ellenby had gone around).
Xerox continued to lose market share. It clung to phase program management, even though it was discredited with Xerox’s poor performance, and demanded shorter product development times. This led to worse products, with the 3300 being so bad it had to be recalled. Things got so bad that Xerox began to sell copiers rather than lease them to sacrifice long-term revenues for short-term profits.
In 1980/81, Kearns finally acted on his desire for change. He learned from Fuji-Xerox, who developed products in half the time it took Xerox to, that delegating responsibility to the people closest to the action - engineering to engineers, marketing to marketers, manufacturing to manufacturers - would result in better decision making and organizational change. He rounded up a group of 25 leaders, and worked with them to make change throughout the company.
They eliminated unnecessary jobs, created several general manager positions for strategic business units, replaced phased program planning, and established a series of benchmarks by which the company could compare its performance against competitors and customer demands. They also all receive positions and clout to carry the change through.
Research: The Harvest of Isolation
36-year old Don Massaro ran the Dallas division after Bob Potter left to join Harvester International as its CTO, after many years of over promising and under delivering in Dallas. Massaro had joined Xerox when the company he co-founded was bought. He was wealthy from the purchase and was still early in his career. He was okay taking chances others wouldn’t. He reignited a fire in Dallas, and his track record over the first year was impressive. Eventually, he was pitched the Alto and ran with it.
The Alto was now called the Star, and had a bunch of software written for it. It was originally the basis for an electric copier. By the time it went out to market, however, the personal computer market had already begun to blossom and the Star didn’t quite fit into the current market. It didn’t have a spreadsheet function, already critical by that time, and wasn’t “open”, in that only Xerox engineers could write software for it. This was because engineers did not trust marketing or finance, and so received little input about what was needed in market.
The method in which it was rolled out, in which Xerox offered an expensive hardware / software system with new enhancements and software every year, was a classic IBM strategy of “lock-in.” However, when IBM entered the PC market it broke from its usual tactics and offered an open system with mostly off-the-shelf components.
The Star was hard to market because the sales force did not understand well enough it’s purpose. They were used to comparing specs on “boxes” of copiers. It was also expensive, and the market wasn’t ready to experiment with connected systems on a network. Soon, Massaro’s sales team was merged into the larger copier sales organization. With that, Massaro quit.
Over the years, many notable PARC employees quit due to their dissatisfaction with how Xerox handled their research. Some left for other research jobs. Others found computer companies backed by venture capital.
In late 1979 Steve Jobs visited PARC after one of Xerox’s investment arms contacted Apple about a possible deal. Jobs instantly grabbed onto the Alto. Eventually, many of the features that were introduced into the Alto were included in the Apple Lisa, introduced in 1983.
However, some had stayed, such as Bob Taylor, Butler Lampson, Chuck Thacker, and others who had created personal distributed computing. Eventually, the clashes between Pake and Taylor led to an ever-growing rift; Taylor was a joy to work for if you were on his good side and shared his worldview, but if one didn’t one would be blocked from making an impact at PARC. Taylor also engineered some political coups, installing friendly agents in various groups and running the lab without the title or responsibility.
Politics blossomed, effectively splitting PARC into two - those with Taylor and everyone else. There was real animosity between the two groups.
Eventually, Taylor was read the riot act when it was clear his power grab and continual disparagement of other parts of PARC and the company were not going away. He resigned in a meeting, surprising the researchers. Thacker resigned immediately after. Taylor was hired to build a similar lab at Digital Equipment Corporation, and upon hearing the news many of the PARC researchers left to rejoin Taylor.