Fire in the Valley (2014)
Chapter 2
The Voyage to Altair
You can’t deny that Ed Roberts started the industry.
Mark Chamberlain, MITS employee
The personal-computer revolution came from a company started in a garage, but not the company and not the garage that most people know about. By the early 1970s the desire among engineers and electronics hobbyists to own their own computers was like the pressure in a champagne bottle. And when Ed Roberts and his unlikely company, MITS, popped the cork, the party began. MITS launched a personal-computer industry, complete with stores, publications, conferences, user groups, software piracy, and debates over open vs. closed standards. Virtually the whole history of the personal computer appeared in miniature at MITS, where Bill Gates said, “every idea was half executed.”
Uncle Sol’s Boys
Ed Roberts? You gotta give him credit for doing the first one. But give the guy [who] published him as much credit: Les Solomon.
-Chuck Peddle, computer designer
Bill Gates, Paul Allen, and other computer enthusiasts relied on the hobbyist electronics magazines like Popular Electronics and Radio-Electronics to keep up with the latest technological developments. In the early 1970s, what Gates and Allen were seeing in the pages of those magazines served to frustrate them as much as excite them. Most of the magazines’ readers knew something about computers, and many knew a lot more than that—and every one of them now wanted to own a computer. The computer aficionados who read Popular Electronics and Radio-Electronics were an opinionated lot; they knew precisely what they did and didn’t want in a computer.
What these enthusiasts wanted most often was more control over the machines they used. They resented having to wait in line to use the very tool of their trade or to engage in their favorite hobby. They wanted immediate access to the files they created on a computer, even if they were off somewhere on a business trip. They wanted to play computer games at their leisure without someone telling them to get back to work. In short, what these enthusiasts wanted was a personal computer. But in the early 1970s, the idea of someone owning his or her own computer was no more than a wild dream.
Your Very Own Computer
A big step toward the realization of the personal-computer dream happened in September 1973, when Radio-Electronics published an article by Don Lancaster that described a “TV Typewriter.” A prolific contributor to electronics magazines, Lancaster later published his groundbreaking idea in book form. His proposed uses for the TV Typewriter were nothing short of visionary:
“Obviously, it’s a computer terminal for time-sharing services, schools, and experimental uses. It’s a ham radio Teletype terminal. Coupled to the right services it can…display news, stock quotations, time, and weather. It’s a communications aide for the deaf. It’s a teaching machine, particularly good for helping preschoolers learn the alphabet and words. It also keeps them busy for hours as an educational toy.”
Lancaster’s TV Typewriter, for all its visionary appeal, was nevertheless only a terminal: an I/O (input/output) device that would link to a mainframe computer. It didn’t constitute the personal computer that the electronics hobbyists desperately wanted.
Figure 16. Art Salsberg and Les Solomon Salsberg, editorial director of Popular Electronics (left), and Les Solomon, its technical editor, pose with the historic January 1975 Altair cover.
(Courtesy of Paul Freiberger)
At the time Lancaster’s article was published, Popular Electronics technical editor Leslie (Les) Solomon was actively seeking a computer story for his magazine. Solomon and editorial director Arthur Salsberg wanted to publish a piece on building a computer at home. Neither of them knew if such a thing was even possible, but in their bones they felt that it should be. They didn’t realize that their competitor, Radio-Electronics, was already preparing to publish an article on just such a topic.
If a home-built computer design was possible, Solomon figured that it would most likely come from one of his “boys”—those young, technically sharp Popular Electronics contributors, like Stanford graduate students Harry Garland and Roger Melen, or Forrest Mims or Ed Roberts.
Designs were being submitted to Popular Electronics, but they weren’t from these stars, and Solomon and Salsberg found them unimpressive. Solomon described what he was seeing as “a rat’s nest of wires,” and Salsberg agreed: “They were terrible designs. They were tinker toys. They were kludges.” Solomon wanted to feature a really good design that he could develop into a groundbreaking story. So he encouraged his guys to send him their best designs, and they took his request seriously.
A colorful and ebullient editor with a wry New York wit, Les Solomon was known as Uncle Sol to his contributors. He developed a close relationship with them, carrying on lengthy telephone conversations and visiting their labs and workshops whenever he could. Solomon regaled his contributors with far-fetched stories and entertained them with magic tricks, most notably a stunt that involved levitating a stone table. Part of what made Uncle Sol fascinating was in trying to figure out what was for real and what was a complete put-on. But one thing was certain: he was serious about getting the best material for his magazine.
The avuncular Solomon was always willing to give his boys advice. When Garland and Melen submitted one of their designs, Solomon told them that they needed to find a distributor. He put them in touch with Ed Roberts, president of an Albuquerque-based company called MITS.
The Man Who Launched a Revolution
Solomon had met Roberts some time earlier. The Popular Electronics editor was vacationing with his wife in Albuquerque and stopped off to visit one of the magazine’s contributors, Forrest Mims. Mims took an immediate liking to Uncle Sol and brought him to meet his business partner, Ed Roberts. Solomon and Roberts also hit it off immediately. The meeting would prove to be a significant event in the development of the personal computer.
Roberts, like Solomon, enjoyed tinkering with electronics. He had played with electronics since his childhood in Miami and had managed to build a crude relay computer while still in his teens. Although he had originally wanted to become a doctor, Roberts decided to join the Air Force to get training in electronics. In 1968, while stationed in Albuquerque, Roberts, Mims, and two other Air Force officers started a small electronics company out of Roberts’s garage. They called their company Micro Instrumentation Telemetry Systems, or MITS, and sold radio transmitters for model airplanes by mail order.
Roberts soon involved MITS in other types of projects. For a time, MITS was building and selling a digital oscilloscope for engineers, but Roberts pushed to take on something more daring, closer to the technological cutting edge. His three partners objected to some of his wildest ideas, so he bought them out. By 1969 Roberts was alone at the helm, which was how he liked it.
A physically imposing man who became accustomed to giving orders while serving in the Air Force, Roberts ran a tight operation at MITS and would brook no nonsense from his employees. In every sense, MITS was Roberts’s company. By 1970, MITS had moved out of the garage and into a former restaurant whose name, The Enchanted Sandwich Shop, still hung over the door. Roberts began manufacturing calculators.
The calculator market was a dizzying place to be in the early 1970s. In 1969, when Intel got the contract to produce calculator chips for the Japanese firm Busicom, the cost to build and sell a calculator nearly approached the cost to build and sell a low-end minicomputer. By the early 1970s, semiconductor technology had changed the calculator market so radically that Ed Roberts could reasonably consider wrapping some electronics and a case around some chips from Intel and selling the results for a small fraction of the purchase price of a Busicom calculator.
What Roberts really wanted to do, though, was both more and less ambitious than building and selling conventional calculators: he was determined to produce programmable calculators and sell them unassembled in kit form. A calculator kit was the ideal product to be featured in the hobby electronics magazines, and Roberts made sure to publicize it there.
The calculator kits sold well for some time among enthusiasts, and Roberts made the fateful decision to invest the bulk of the MITS capital and development efforts in commercial handheld calculators. The decision turned out to be disastrously ill-timed.
Killed in the Calculator Market
Two trends in semiconductor technology reached critical stages in 1974 and helped to create the climate in which the microcomputer was born. The semiconductor companies began to produce and market applications of their technology—in particular, calculators—directly against Intel president Robert Noyce’s dictum that the chip manufacturers shouldn’t compete with their own customers. In addition, the early, crude microprocessor chips were being refined with better-thought-out designs and more power.
The first trend brought MITS to the brink of bankruptcy; the second yanked it back.
In the early 1970s, the semiconductor houses, racked by fierce technological and price wars, noticed that some of their customers had much healthier profits than they themselves did. One example was Commodore, a Canadian electronics company that had moved into Silicon Valley from Toronto and sold calculators that were assembled around a Texas Instruments (TI) chip. Commodore was raking in money from a product that was little more than a TI chip in a plastic case.
The demand for calculators seemed endless, and great profits were made in meeting the need. By 1972, TI had entered the calculator business, and other semiconductor manufacturers soon followed suit. “They just came in and ripped everybody to shreds,” according to semiconductor designer Chuck Peddle. TI’s attack on the industry was characteristically aggressive: it burst upon the market and immediately undercut everyone else’s prices.
Calculators quickly became smaller and more powerful, prices dropped dramatically, and profits shrank almost as fast. With a nationwide recession hobbling many businesses, 1974 was not a good year for the calculator industry. Peddle, who was working on microprocessor design at Motorola, recalled, “The market went to hell that year. Supply started catching up with demand. Everybody that year lost money in the calculator business.” Calculators went from being a high-end purchase to a sidewalk giveaway. The average price for a consumer calculator in 1974 was $26.25. A year before it had been $150.
One of the firms stricken by the recession and lackluster profits was MITS. In January 1974, MITS was selling a simple eight-function calculator kit for $99.95, and couldn’t bring the price any lower. Texas Instruments was offering a comparable, fully assembled calculator for less than half what MITS was asking. The tiny firm couldn’t swim in those waters. Ed Roberts lay awake nights trying to figure out where he had gone wrong.
The other pivotal development in semiconductors happened in April 1974, when the successor to the Intel 8008 microprocessor was completed.
Intel had really created the brain of a computer in the 8008. However, the 8008 was, in the words of Art Salsberg, “a kludge and a monster.” Everything was there, but not in the right places. It handled vital operations in a slow, roundabout way and demanded a contorted, awkward form of programming and design. Even engineers inside Intel disagreed about whether the 8008 could actually function as the brain of a viable, commercial computer.
The electronics hobbyists weren’t waiting for the semiconductor engineers to make up their minds.
Going for Broke
Why don’t you call it Altair? That’s where the Enterprise is going tonight.
-Lauren Solomon, daughter of Popular Electronics editor Les Solomon (like other Solomon stories, possibly apocryphal)
Ed Roberts made a decision of his own that spring—he was going to build a kit computer. He had been toying with the idea for some time only to find that by early 1974 the chips, so to speak, were down. MITS’s calculator business had blown away like desert sand, leaving the company heavily in debt. Faced with the likelihood of going under, Roberts decided to go for broke. He would build a product that had essentially no precedent or defined market, a product most people considered fanciful at best. But the specter of bankruptcy only gave urgency to his decision. Roberts always cared more about technological challenges than about any business risks they presented. He would have gone ahead with the kit computer under any circumstances.
Les Solomon Flies to Albuquerque
Roberts studied Intel’s chips—the early 4004, the 8008, and a third Intel product called the 4040—and rejected the 4004 and 4040 as too crude. He was considering building a machine around the 8008, until a programmer told him that he had tried to implement the BASIC programming language on the 8008 and had found it to be an excruciating process. The 8008 carried out the BASIC instructions far too slowly to be useful.
Then a new product caught Roberts’s eye: the Intel 8080. By this time, Motorola also marketed a microprocessor, the 6800, and Texas Instruments and other companies had similar products. But Roberts had studied the chips carefully and concluded that the 8080 was technically superior to the 6800. It had another even more significant advantage.
Intel normally charged $360 for an 8080, but Roberts got the company to knock the price down to $75. It was unheard of: nobody else was getting that price, but for a very good reason. The Intel contract required him to buy in volume, and each computer needed only one processor. He’d be committing to a business model that would require him to sell a lot of computers.
That was fine with Roberts. After the calculator fiasco, which Roberts said was “something you don’t want to go through twice in a lifetime,” his operation was geared up to sell plenty of product—which he would need to do to salvage the company. He was going to have to think big or give it up entirely.
Meanwhile, Popular Electronics was narrowing its search for a computer project it could publicize. “We got in a bunch of computers,” Art Salsberg recalled. “We wound up with two models and decided it was going to be a choice of one or the other. One amounted to no more than a promise. The promise was, I can get the chips at a lower price and make this whole thing feasible. That was from Ed Roberts. The other choice was a microcomputer trainer by Jerry Ogdin.” The model from Ogdin was actually more a tool for learning about computers than an actual computer.
Roberts offered only a concept, whereas Ogdin’s device actually existed and Salsberg and Solomon had seen it. They were both inclined to support a tangible machine over the mere promise of one, even though Ogdin’s machine was built around the 8008 chip, which was about to be phased out. “It looked like it was a go with the microcomputer trainer,” Salsberg said.
Then the July 1974 Radio-Electronics hit the newsstands with an article by Jonathan Titus on building an Intel 8008-based computer called the Mark-8. The article generated a lot of excitement among hobbyists, even if not a lot of orders. The editors at Popular Electronics realized immediately that this put a crimp in their plans. On reading the Radio-Electronics article, Salsberg announced, “That kills the trainer.” Solomon agreed: Ogdin’s trainer was just too similar to the Mark-8. Popular Electronics had to up the ante. An article on an 8080-based microcomputer would do just that.
Solomon promptly flew to Albuquerque to meet with Roberts and work out the details. Salsberg wanted the computer packaged like a serious commercial product. Roberts spent many late nights hashing out the exact components of a desktop computer that could sell for under $500.
This presented an enormous challenge. The Mark-8 sold for about twice that price, and when you added up the cost of the components that any computer needed, it was hard to get the price much lower. But Roberts did have the advantage of the volume pricing from Intel.
In the end, Roberts promised to meet the price and to deliver the first machine to Popular Electronics as soon as it was built, and Popular Electronics promised to publish a series of articles on it, including a cover story.
When Salsberg agreed to go with Roberts’s machine, he was taking a risk. This was to be the cover story for the issue. If they promoted this computer and it turned out to be a bomb, the magazine would look bad. No one at MITS had ever built a computer before. Roberts had only two engineers on his staff, and one of them had his degree in aeronautical engineering. Roberts had no prototype and no detailed proposal. But Uncle Sol kept assuring Salsberg that Roberts could deliver the goods. Salsberg hoped he was right.
Roberts was just as edgy about Popular Electronics’s promises. However much he liked and respected Les Solomon, he was wary of Solomon’s cheerful assurances. And the more he realized how crucial a cover story in Popular Electronics was for MITS, the more nervous he became. His company’s future was in the hands of a man who levitated tables.
The Mark-8 wasn’t the first computer built around the Intel 8008, although Roberts had no way of knowing that. That distinction belongs to the Micral computer, built in 1973 by André Truong Trong Thi, a French Vietnamese entrepreneur. Thi sold 500 of the machines, all in France. Later that year, he demonstrated an 8080-based computer at a major computing conference in the United States. Whatever impact the demonstration had on the engineers and computer scientists who saw it apparently didn’t extend much beyond that conference. The same fate could easily befall Roberts’s machine.
Designing the Personal Computer
Over the summer of 1974, Roberts had sketched out the machine he wanted. As his ideas took shape, he passed them along to the two guys on his engineering team, Jim Bybe and Bill Yates. A quiet and serious man, Yates worked long hours on the layout of the main circuit board for the machine, planning how each electrical signal would get from one point to another in the computer.
Roberts wanted this computer to be expandable, like a minicomputer. He wanted the user to be able to install other circuit boards for particular functions, such as controlling an I/O device or providing extra memory. Roberts wanted the boards designed to plug easily into the computer, a capability that required not just a socket, but also specific, defined data paths. If different elements of the computer were to reside on physically distinct circuit boards, the boards had to be made to communicate with each other. This communication, in turn, required certain engineering conventions. For instance, one board needed to send information when and where it was expected by another board. Almost by default, a bus structure for the computer evolved.
A bus is a channel through which computer data or instructions travel. Typically, a bus is a parallel channel with several different signals passing simultaneously. The MITS computer had 100 separate channels, or paths, and each had to have a stated purpose. Added to that were the physical and electrical constraints that sometimes dictated the design of the layouts. For instance, electrical cross-talk—interference between wires—makes it unwise to place channels for certain kinds of signals too close together. But Roberts allowed Yates no time to address such niceties of design, because the creditors had already begun to bay. Wherever the data channels fell, that’s where they stayed. The bus design did the job, but it wasn’t pretty.
While Yates laid out the boards, another MITS employee, technical writer David Bunnell, was casting about for a name for the computer. His favorite of all the candidates was “Little Brother,” but he wasn’t altogether comfortable with the name. Bunnell wasn’t really comfortable with the whole notion of computers, Roberts recalled. But Bunnell kept his skepticism in check, given Roberts’s lack of patience with dissent.
Bunnell had been with MITS since 1972. He and Roberts had coauthored articles for Popular Electronics, and they were writing a series of tutorials on digital electronics for the magazine at the same time they were toiling in the MITS workshop developing their computer.
Getting Financing
Despite their efforts, it was beginning to look as if the computer was destined to die in the workshop. MITS owed around $300,000 to its creditors. With Les Solomon’s constant reminders that the article’s deadline was imminent, Roberts made a grim trek to the bank. It was mid-September. He was out of money, needed another loan, and fully expected the bank to turn him down. Given his current credit rating and his depleted assets, he doubted anyone would lend him the $65,000 he needed to keep the company’s doors open.
The officers of the bank listened patiently. He was going to build a kit computer? And what exactly was that? And who, did he think, would buy such a product? Electronics hobbyists, sight unseen, from ads in magazines? And how many of these kit computers did he think he could sell in the next year to these electronics hobbyists through advertisements in magazines?
With a straight face, Roberts told them 800. “You won’t sell 800,” they said, thinking he was being unrealistic. Roberts was indeed fantasizing. Still, the bank officers saw no advantage in bankrupting companies with outstanding loans. The loan officers figured that if Roberts could sell 200 of the things it would help MITS to repay the bank something. They agreed to advance him the $65,000.
Roberts did his best to hide his surprise. He was glad he hadn’t mentioned the informal market survey that he had just conducted. Trying to get some sense of how the machine would be received, Roberts described it to some engineers he knew and asked if any of them would buy it. They all said no.
Although Roberts never considered himself a good businessman, he knew instinctively when to ignore market research. He took his $65,000 and, with Yates and Bybe, worked feverishly to complete the prototype to send to Popular Electronics. It was going to appear on the cover, so they made sure it looked especially attractive.
Figure 17. The MITS Altair 8800, assembled The default input and output for the Altair computer were the switches and lights on the front panel. (Courtesy of Intel Corp.)
Because Bill Yates was doing most of the design, he worked with Roberts on the article. While Roberts and Yates were scrambling to finish both the computer and the article, they realized they still didn’t have a name for their machine. They figured Solomon would put a Popular Electronics name on it if they didn’t, so they beat him to the punch by calling it the PE-8. It was Roberts’s last small hedge against Popular Electronics’s scuttling the project. But that wasn’t the name by which the machine became famous.
According to Les Solomon, his 12-year-old daughter Lauren was the one who came up with the name that stuck. She was watching an episode of Star Trek when her father walked into the room and said, “I need a name for a computer. What’s the name of the computer on the Enterprise?” Lauren thought for a moment and said, “Computer.” Her father didn’t think much of that name, so Lauren suggested, “Why don’t you call it Altair? That’s where the Enterprise is going tonight.”
Some of Solomon’s friends told a different story of the naming, but Altair it was. “I don’t give a damn what you call it,” Roberts told Solomon. “If we don’t sell 200, we’re finished.” Solomon reassured him that things were going well and selling 200 was entirely possible. Solomon wasn’t just being polite and trying to soothe the raw nerves of a man who’d been flayed in the calculator-market crash. He was confident that the Altair had the potential to far outstrip the Mark-8.
The Mark-8 was an experimenter’s toy, a way for the engineering hobbyist to learn about computers firsthand. But the Altair was, for all its limitations, a real computer. Its bus structure would make it possible to expand the machine’s capabilities by allowing the user to plug in new circuit boards. And the 8080 chip was a far better “brain” than the 8008. The Altair had the potential, at least in miniature, of doing everything a large mainframe computer could do.
Solomon was convinced of it and told Roberts as much. But he didn’t voice his concern that the message might not get across to the Popular Electronics readers. Art Salsberg told him that Popular Electronics had to offer its readers more than just instructions for building the device. To prove that the Altair was a serious computer, Popular Electronics had to also offer one solid application, a practical purpose for the Altair that could be demonstrated right away. What that application might be, Solomon had no idea.
Delivering the Goods
The deadline arrived for Roberts to deliver the prototype computer to Solomon. Roberts told him that it was coming by Railway Express, a troubled parcel-shipping service that would cease operations later that year, and to watch for it.
Solomon waited. No computer arrived.
Roberts reassured him that the computer was in the mail and should be arriving any day. Days later, the prototype was still a no-show. Solomon, in turn, tried to reassure Art Salsberg at Popular Electronics that the machine was on its way, but now everyone was getting nervous. Roberts flew to New York to demonstrate the prototype, confident that it would arrive by the time he did.
But it didn’t. Railway Express had apparently lost their computer. This was a catastrophe, both for MITS and for Popular Electronics. The magazine had committed to a cover story, and now it had no computer to put on the cover. For weeks, Roberts had lain awake nights, static buzzing away in his brain. Now he felt that his worrying had been justified. His engineers couldn’t possibly assemble another computer in time to meet the deadline. They were sunk.
Unless, of course, they faked it.
Yates could slap together a box, poke little lights through the holes in the front, and ship it to New York. Les Solomon didn’t like the idea. Art Salsberg hated it. Ed Roberts was embarrassed. But when the January 1975 issue of Popular Electronics went to press, it featured a flashy cover photo of an empty metal box masquerading as a computer.
Between the time the issue was wrapped up and the time it hit the streets, Solomon finally got his hands on an Altair computer. He immediately set it up at his desk, but the noise from the Teletype machine he was using as an I/O device made him instantly unpopular in the Popular Electronics offices. So he took the system home and set it up in his basement. It was there that Roger Melen first saw it.
The day after Roberts and Yates’s piece on the Altair appeared, an article came across Solomon’s desk that caught his attention. Harry Garland and Roger Melen, the two Stanford graduate students Solomon had once hooked up with Ed Roberts, sent in a description of a digital camera they had designed. The Cyclops, as Garland and Melen called it, reduced an image to a rectangular grid of light and dark squares and provided a low-cost visual system for a digital computer.
In December 1974, coincidentally just before Popular Electronics’s Altair issue came out, Roger Melen decided to fly to New York. His trip ultimately led him to Les Solomon’s basement.
Melen reminded Uncle Sol of Ed Roberts in a way. Both were well over six feet tall and heavyset, and both were inveterate engineers/hobbyists, but the Air Force-trained Roberts was older and tougher. Melen was quiet and soft-spoken, the product of one of the top engineering schools in the world. Nevertheless, the two would see eye-to-eye, Les thought, chuckling to himself at the unintentional joke. Trying to look nonchalant, he led Melen through his basement to a strange-looking apparatus. “What’s that?” Melen asked. “That, sir,” Solomon told him, “is a computer.”
Figure 18. Les Solomon’s basement Popular Electronics technical editor Les Solomon showed the as-yet-unannounced Altair to an astonished Roger Melen in this basement. Here the basement features a Sol-20 and other historic personal computers. (Courtesy of Les Solomon)
When Solomon told him what the Altair was and how much it cost, Melen politely demurred. There must have been some mistake. Melen knew for a fact that the microprocessor chip alone cost as much as he claimed this whole computer did. Solomon suppressed a smile and assured him that the price was correct. Roberts was actually going to sell this computer for $397. Delighted at Melen’s reaction, Solomon picked up the phone, called Roberts in Albuquerque, and checked the price as Melen stood there. Yep, it was still $397.
Melen was stunned. As he and most hobbyists well knew, Intel was charging $360 for the 8080 chip alone. When Melen left New York that day, instead of flying directly back to San Francisco he took a side trip to New Mexico. Melen sensed that something big was happening, and he wanted to be a part of it.
Roberts greeted Melen enthusiastically at the Albuquerque airport that evening and drove him over to MITS. There Melen was in for another surprise: far from being the large company he had expected to see, the MITS office was in a strip mall, wedged in between a massage parlor and a laundromat. The MITS headquarters must have looked as odd to Melen as it did to the suburban shoppers who strolled past its doors that winter.
“It was obviously the skeleton of what used to be a company, because they had lots of equipment around,” Melen later recalled. “But they only had, I think, 10 employees at that time. They had been very successful in producing calculators, but that was a fad that had passed. He [Roberts] saw this as his big chance for success—his second shot to pull him out of his predicament.”
Melen recognized a mutual opportunity and proposed attaching his Cyclops camera to the Altair. Roberts was interested, and after a brief tour of MITS, the two men sat down to work. Melen studied the Altair schematics, gathering all the information he thought he would need to design an interface between the two devices. He and Roberts talked about computers in general and the Altair-Cyclops interface in particular until dawn, when Melen hurried back to the airport to catch an 8 A.M. flight to San Francisco.
Soon after the meeting between Melen and Roberts, Solomon wrote to Garland and Melen suggesting a television adapter for the Cyclops. They replied that it would be prohibitively expensive, and instead described their plan to link the Cyclops device to the Altair for use as a security camera. Solomon was gleeful. The security camera was the practical application that Art Salsberg had wanted. He incorporated the idea into Garland and Melen’s article on the Cyclops.
The brainstorming session with Melen was not to be Ed Roberts’s last sleepless night. His future, his company, everything hung on this article in Popular Electronics and on a positive response from the magazine’s readers. He kept his enthusiasm in check, despite Les Solomon’s cheery encouragement. Roberts felt that Solomon could scrap the project even on the eve of publication. If that happened, MITS was through. Already hundreds of thousands of dollars in debt, Roberts had borrowed heavily to finance this computer venture. He had purchased enough parts to build several hundred machines—and he still had to pay for advertising. At $397 for one machine, he would need to sell the 800 machines he’d glibly mentioned to the bank and a lot more just to break even. He began to suspect that he’d made a terrible mistake.
All Hell Breaks Loose
PROJECT BREAKTHROUGH! World’s First Minicomputer Kit to Rival Commercial Models…ALTAIR 8800
-Popular Electronics cover, January 1975
Ed Roberts was still worried about his investment even as the first orders came rolling in. But within a week, it was clear that whatever problems MITS would face in the immediate future, bank foreclosure would not be one of them. In just a two-week period, Roberts’s tiny staff had opened hundreds of envelopes and read with giddy excitement orders for all the computers they had ever hoped to sell. Within a month, MITS had gone from one of their bank’s biggest debtors to a fiscal hero. MITS’s bank balance went from $400,000 in the red to $250,000 in the black in a few weeks. Just processing the orders seemed to be a full-time job for everyone.
They Were Buying a Promise
No one had realized just how primed the market was for a personal computer. The January issue of Popular Electronics signaled to thousands of electronics hobbyists, programmers, and other technophiles that the era of the personal computer had finally arrived. Even those who didn’t send in checks saw the Altair article as a sign that they could now have their own computers. The Altair was the fruit of a technological revolution that dropped straight into the hands of a hungry population. They went crazy for it.
Roberts, who had gambled his company’s life on the existence of any market for the machines, was amazed at the magnitude of the response. His experience at selling $99 kit calculators was of little value in predicting the number of buyers for a $397 computer. In addition to the significant price difference, the calculator had a well-defined and obvious function. By comparison, it wasn’t yet clear what the Altair could actually do. Despite Salsberg’s artfully vague promise in Popular Electronics of “manifold uses we cannot even think of at this time,” it was not at all obvious what those “manifold uses” were. That didn’t stop Roberts’s phone from ringing almost nonstop.
Electronic hobbyists were happily buying promises.
Figure 19. The MITS Altair 8800, unassembled Early purchasers of the MITS Altair received a bag of parts and assembly instructions. (Courtesy of David Bunnell)
One of the promises customers bought was delivery in 60 days. Faced with an immediate backlog of orders, Roberts determined that he had to establish priorities or they would never make any deliveries. He issued a no-frills edict: initial production would include only the bare machine. All the bells and whistles, such as extra memory, the clock board, and the interface boards to allow the computer to be connected to a Teletype machine, would have to wait. MITS would ship the box and central processing unit (CPU) board with 256 bytes of memory, the front panel, and nothing else until the backlog was cleared. As delivered, the Altair was no more powerful than the Mark-8. Only its possibilities were greater.
A few orders were filled early in 1975. Garland and Melen, working on Cyclops in the guest bedroom of Melen’s Mountain View, California, apartment, were MITS’s first computer customers. They were not your typical customers. The average order went out only after it inched to the head of the queue, which took time. Garland and Melen received Altair No. 0002 in January. (The first Altair, lost in shipment to New York and never seen again, was unnumbered. Les Solomon got No. 0001.) Garland and Melen immediately set to work on the interface board that would allow the computer to control their Cyclops digital camera.
Despite MITS’s promise of 60-day delivery, orders were not filled in any quantity until the summer of 1975. One hobbyist, Michael Shrayer, who went on to write the first personal-computer word-processing program, described his experience with MITS: “I sent away my $397. Many phone calls later, the computer finally came. It took forever. At that time, I received a big, empty box with a CPU card and 256 bytes of memory. No terminal, no keyboard, nothing. To put anything in it, one had to play with the switches on the front panel and put in minor programs. A lot of peripherals were being promised but not delivered.”
“Minor programs” was a generous description of what you could feed the early Altair. Programs had to be written in 8080 machine language and entered by flipping switches, with one flip of a switch for every binary digit. And once they were entered, the programs could do little except make the lights on the front of the box blink. One of the first programs written for the Altair was a simple game. It caused the lights to blink in a certain pattern, which the player was supposed to mimic by flipping switches.
Figure 20. Steve Dompier He was so eager for a computer of his own that he flew from San Francisco to Albuquerque to check on his Altair order. Here he pays a similar visit to Processor Technology. (Courtesy of Bob Marsh)
The Altair buyer faced another problem after delivery. The computer was sold as a kit, and assembling it took many hours. The odds of the computer eventually working depended on the skill of the hobbyist and the quality of the parts. Most of the first machines simply didn’t work, despite the skill of the user. Steve Dompier, a young building contractor in Berkeley, California, was surprised to find that some of MITS’s advertised equipment didn’t even exist. He recalled sending in a check for $4,000 with a succinct request for “one of everything.” When half his money came back with an apologetic note from a beleaguered MITS secretary saying that they “didn’t have all that stuff yet,” Dompier boarded a plane for Albuquerque.
Flying from San Francisco to Albuquerque over a delay in filling an order for hobby equipment might seem overzealous to some, but not to Dompier. “I wanted to see if they were really there. I rented a car and drove past the place about five times. I was looking for a big building with the letters MITS on it and a front lawn. It turned out it was in a tiny building next to a laundromat in a shopping center. There were two or three rooms. All they had was a box full of parts.” He picked up some of those parts and returned to San Francisco.
On April 16, 1975, Dompier reported on MITS at a meeting of the Homebrew Computer Club, a pioneering microcomputer club in Menlo Park, California. Dompier drew an attentive audience. MITS, he told his listeners, had 4,000 orders and couldn’t even begin to fill them. The thousands of orders, more than anything else, sparked people’s interest. What they had been waiting for had happened. They were going to have their own computers.
But calling the Altair a computer took some imagination. By mid-1975, when MITS was delivering product on a regular basis, the assembled machine was no more than a metal box containing a power-supply unit bolted next to a large circuit board. This board was called the motherboard because it was the main piece of circuitry in the machine. It contained 100 strands of gold that connected the motherboard to 18 slots into which other circuit boards could be plugged.
Those 18 slots were a symbol of both the Altair’s expandability and an owner’s frustration at not being able to use most of them. Regardless of whatever a customer may have ordered, what was shipped was a machine with only two of the slots filled. One slot would have a board containing the CPU (basically, the Intel 8080 chip and supporting circuitry), and the other slot would have a board that contained 256 bytes of memory.
The Altair package also included a front-panel board that controlled the lights and switches on the front of the box. These lights and switches were the I/O, the means by which users communicated with the machine. It was up to the customer to attach the front-panel board to the motherboard by hooking up dozens of wires—a task requiring hours of tedious work. But these three boards, comprising a CPU, some memory, and an I/O unit, meant that the early Altair—barely—met the minimal definition of a computer.
Maybe all that the Altair could do was blink its lights, but for the Homebrew members just the fact that it existed was enough for them. They would take it from there.
“They made the business happen,” semiconductor designer Chuck Peddle said of these early hobbyists. “They bought computers when they didn’t work and when there was no software for them. They created a market, and then they turned around and wrote the programs that brought other people in.”
The early purchasers of the Altair had no choice but to write their own programs. MITS initially supplied no significant software with the machine. The typical response of a computer hobbyist to the Popular Electronics article was to first send for an Altair, and when it arrived (and had been successfully assembled) begin writing software for it.
Enter Gates and Allen
Two programmers in Boston decided to skip Step One.
Paul Allen was working for Honeywell in Boston. Bill Gates was a freshman at Harvard, where he had customized a curriculum that allowed him to take graduate mathematics courses. On weekends the two would get together to brainstorm about the microcomputers they just knew were coming, and these microprocessors that would surely power them. “We were just trying to figure out something we could do with them,” Allen recalled. Gates and Allen sent out offers on their old Traf-O-Data stationery to write implementations of PL/I (Programming Language One) for $20,000. They also considered selling Traf-O-Data machines to a company in Brazil. In the middle of a Boston winter, they were spinning their wheels.
While walking through Harvard Square one day, Allen spotted the Popular Electronics cover that featured the Altair. Like many other computer enthusiasts, he realized at once that the Altair was a tremendous breakthrough. But he also saw it as something of personal interest. Allen ran to tell Bill that he thought their big break had finally come. Bill agreed.
“So we called this guy Ed Roberts,” Gates recalled. “We had a fairly aggressive posture. We said, ‘We have a BASIC. Do you want it?’” In 1975, Allen and Gates were pioneers in the industry practice of preannouncing products that didn’t yet exist. Later, this type of thing would come to be called “vaporware.”
Roberts was justly skeptical. He had heard from many programmers who claimed they could write software for his computer. He told Gates and Allen what he told everyone else: he would buy the first BASIC he saw actually running on an Altair.
Unlike the others, Gates and Allen delivered. About six weeks later Allen flew to Albuquerque to show Roberts their BASIC. The demonstration was a success even though their BASIC initially did little more than announce its presence. The Traf-O-Data company, newly renamed Micro-Soft (later changed to Microsoft) had made its first sale as a microcomputer software house.
In March, Roberts offered Paul Allen the position of director of software at MITS. Frustrated at Honeywell and eager to work in what he saw as a tremendously promising field, Allen accepted immediately and flew to Albuquerque with all the cash he and Gates could lay their hands on. The title of MITS software director, as it turned out, was not quite the illustrious post Allen had imagined. Upon arriving in Albuquerque, he discovered that he was the software department.
Figure 21. MITS Altair ad David Bunnell wrote the copy for this early Altair ad, which ran in Popular Electronics and Scientific American. (Courtesy of David Bunnell)
Putting It Together
Every good idea was half-executed at MITS.
-Bill Gates, cofounder of Microsoft
Compared with mainframes and minicomputers, the Altair was seriously deficient. It lacked any means of permanent storage. Users could put information into the machine and manipulate it, but once they shut off the power, the information disappeared. Even temporary storage was extremely limited. Although the Altair had a memory board, its 256 bytes of memory wouldn’t have allowed enough space to hold this paragraph.
As an I/O system, the front-panel setup was awkward to use and required a tedious series of steps. To enter information, users had to flip tiny switches on and off; one flip of a switch equaled one bit of information. To read output, a user had to interpret a series of flashing lights. Entering and verifying a paragraph’s worth of information might take several minutes, even with practice. Until paper-tape readers and Gates and Allen’s BASIC came along, Altair owners had to communicate with their machines in machine language bit-by-bit via this switches-and-lights routine.
Machine language was the native language of the Altair’s microprocessor, the Intel 8080. A machine language is a set of commands, in the form of numeric codes, that elicits a response from a computer’s CPU. The code causes the CPU to execute one of its elementary functions—for instance, copying the contents of one specified location in memory onto another location or adding the value of 1 to a stored value. Some programmers, the “true hackers,” preferred to work in machine language because of the intimate and immediate control the language gives them over a CPU’s operation. But all programmers agree that programming in a higher-level language is vastly easier than having to work with machine language. Altair BASIC was a higher-level language. Unfortunately, It took up 4096 bytes of memory. This is remarkably little memory for a high-level language, but it was 16 times the amount of memory that MITS provided in the Altair.
By filling most of the Altair’s 18 slots with 256-byte memory boards and entering Gates and Allen’s BASIC into the system—a tedious process that involved flipping the front-panel switches more than 30,000 times without an error—users could theoretically get a high-level language running. However, the amount of memory left for their own programs would be minuscule. Moreover, the BASIC would have to be reentered every time the machine was turned back on. Two improvements were needed to make the BASIC, and in fact the Altair, useful: higher-capacity memory boards and a method for entering programs quickly. MITS was at work on developing both of these. But MITS was at work on a lot of things.
Getting It to Work
When Paul Allen arrived in Albuquerque, MITS’s biggest hardware project was a 4K memory board that Ed Roberts had designed and technician Pat Godding was attempting to build. In computer jargon, the letter K, short for kilo, represents 1,024, the number closest to 1,000 that is a power of two. Therefore, 4K equals 4,096. Because digital computers use a binary number system, in which every number is expressed as a sum of powers of two, exact powers of two are especially easy for a computer to work with. Computer capacities, such as amount of memory or the largest displayable integer, are generally expressed in powers of two. The new MITS memory board could hold more than 4,000 bytes of information, so Altair BASIC could fit comfortably on it.
Because the 4K memory board would make it possible to run Gates and Allen’s BASIC on the Altair, Allen was particularly concerned that the board should work reliably. It didn’t.
Or rather, it didn’t when combined with other boards. The problem wasn’t just the board itself, but also the performance of two or more boards together. “It was almost analog circuitry,” Allen said. “Things had to be calibrated so exactly.” Bill Yates and the other MITS engineers came to dread Allen’s visits to their work area. In order to test the enhancements he was adding to his BASIC, Allen had to try them out on a working Altair with functioning 4K memory boards.
Unfortunately, none of the 4K memory boards were working. Allen would bring in his latest modification to a program and key it into the machine, whereupon all the panel lights would turn on, the Altair’s way of throwing up its hands in confusion. When technical changes failed to correct the 4K boards, engineers went the redundancy route. At one point, MITS was keeping seven Altairs running constantly just to have three reliable machines at any given time. “That 4K dynamic memory board was atrocious,” Roberts later admitted.
At least Allen didn’t have to key in all of the BASIC every time he wanted to use the machine. The workshop Altair had some secret capabilities that MITS wasn’t yet ready to release to customers. Its programs and data could be stored on paper tape and then loaded back into memory later on. When Allen first demonstrated BASIC to Roberts, he brought it to MITS on paper tape. For a while, paper tape was the major means of distributing the language. Bill Gates would later curse those paper tapes because they provided the medium for widespread illegal copying of their BASIC.
But paper tape had some serious drawbacks as a storage medium for microcomputers. Paper-tape readers and punches were expensive, considerably more expensive than the Altair computer itself. Paper-tape systems were also not terribly fast or efficient.
MITS recognized the need for an inexpensive storage method and was considering using audiocassette recorders. Many computer users already owned cassette-tape recorders, and if a recorder could double as an Altair storage device, all the better. But like paper tape, cassettes were a slow and clumsy way of storing data. By comparison, IBM had long used disk drives for data storage on its large computers. Disks, although expensive, made data storage and retrieval quick and easy.
Roberts was convinced MITS should put disk drives on the Altair. Paul Allen agreed. In 1975, when Bill Gates also made the move to Albuquerque to work on MITS programs, Allen asked him to write the software that would allow the Altair to communicate with a disk drive. But Gates was currently occupied with other tasks, and he put off writing the disk code.
MITS had no shortage of either hardware or software projects. The company was working on interfaces to Teletype machines, printers, and cassette recorders, as well as looking for ways to link a simple terminal to the Altair. MITS was also developing the programs to control these devices, new versions of BASIC and enhancements to the language, and application programs. In addition, all these items needed documentation. On top of all this, MITS undertook PR projects such as a user conference and a newsletter.
Taking It on the Road
One unusual promotional gimmick was the “MITSmobile,” also known as the Blue Goose. An outgrowth of Roberts’s fondness for recreational vehicles, the Blue Goose was an advertising tool designed to spark interest in microcomputing. Gates recalled touring on the Blue Goose: “It was one of those GM motor homes. We’d drive around the nation, and everywhere we’d go, we’d get somebody to start a computer club. I was part of the song-and-dance for one of the tours.” The Blue Goose, like many other MITS innovations, inspired imitators. Utah-based Sphere, one of MITS’s first competitors, soon thereafter sent a Spheremobile roving about the land.
The Blue Goose promotion proved effective. One of the clubs it helped initiate was the Southern California Computer Society, which in turn published an influential early microcomputer magazine, SCCS Interface.
There were good reasons for starting computer clubs. The equipment in these early days didn’t always work or work properly, and software was often unusable or nonexistent. Although buyers were typically engineering hobbyists, few of them had all the skills necessary to fully understand a microcomputer. The clubs encouraged a synergistic sharing of knowledge among the sophisticated-but-stymied users of the machines. Without this interaction and mutual aid, the industry would not have blossomed as it did.
MITS no longer depended on local initiative. By April, MITS had its own nationwide computer club that held design contests and published a newsletter called Computer Notes. David Bunnell started up the publication and Ed Roberts contributed a semiregular column called “Ramblings.” Throughout much of the newsletter’s publication history, Gates and Allen wrote a sizable portion of its contents.
The Altair club offered free membership to Altair owners or those who could pass as owners while they awaited delivery from MITS. Meanwhile, other clubs were springing up that bore no particular allegiance to MITS. The Southern California Computer Society and the Homebrew Computer Club in Northern California, although filled with actual and prospective Altair owners, were also made up of technically sophisticated hobbyists who soon contemplated building computers of their own.
The Homebrew Computer Club members were especially interested in this challenge, and from the club’s ranks there quickly emerged a true competitor to one of MITS’s most important products.
The Competition
There was no competition until Processor Tech came out with the memory cards.
-Ed Roberts, founder of MITS
MITS was a catalyst.
Perhaps more by chance than by design, the MITS operation inspired the creation of an entire industry. But that also meant that MITS spawned competitors, and from Roberts’s perspective, competitors were poaching on his territory. You could see it in the way he reacted to the memory situation. When MITS began delivering its 4K memory boards, it didn’t take long for customers to notice what Paul Allen already knew: the boards didn’t work. “I don’t think I’d trust an Altair memory board to do anything,” one MITS executive later admitted.
Defective Memory
Although Roberts eventually acknowledged the board’s design was awful, at the time he brooked no complaints about it, as Bill Gates soon learned. Gates was using a memory-test program he had written to check the boards as they were completed. “Every one that came off the line wouldn’t work,” Gates said, and he told Roberts as much. The resulting confrontation between the slight 18-year-old and the burly Air Force veteran permanently damaged their relationship. Roberts considered Gates a teenage smart aleck and simply ignored him. “I think that was a fundamental failing of Ed’s,” another MITS employee said. “If he said the memory boards worked, they worked.” Unfortunately, they didn’t.
When Bob Marsh, an out-of-work Homebrew Computer Club hobbyist, started a company called Processor Technology in April 1975 and began selling 4K boards that did work, Roberts took it as a declaration of war. MITS was making little or no profit on the Altair computers and desperately needed the memory-board sales that Processor Technology was cutting in on.
Roberts retaliated by using Gates and Allen’s software as a weapon. The BASIC language was a popular item; the MITS 4K board was not. So MITS resorted to a venerable marketing ploy: it tied the price of BASIC to the purchase of the memory board. Customers who bought MITS boards paid $150 for BASIC. Those who didn’t buy the boards paid $500 for BASIC—more than the price of the machine.
The tactic backfired, and the effect on the market was dramatic. Hobbyists, seeing the 4K boards as worthless and BASIC as overpriced, bought the Processor Tech memory and made their own paper-tape copies of BASIC, distributing the copies for free. By the end of 1975, most copies of BASIC in use on Altair computers were pirated.
Processor Technology survived the BASIC price ploy and developed more Altair-compatible products. Other companies also began to produce memory boards that could be used in the Altair. Roberts railed at those he regarded as squatters in his territory. The memory-board companies responded by crashing David Bunnell’s first World Altair Computer Convention. When Roberts denounced certain memory-board firms in his newsletter, calling them “parasites,” two Oakland, California, hobbyists christened their new memory-board company Parasitic Engineering.
The only board company to win approval from MITS was Garland and Melen’s Cromemco (named for Crothers Memorial Hall, their graduate dormitory at Stanford). Garland and Melen had gotten sidetracked from their plan to connect the Cyclops digital camera to the Altair. The interface board that was intended to perform this feat had taken on a life of its own, and had become a video interface board for displaying text and pictures generated by the Altair on a color television. The Dazzler, as they called the board, neatly solved the Altair’s I/O problem. Roberts saw it as noncompetitive (MITS had nothing like it), and displayed it prominently with his Altair computers at a conference the following spring.
The First World Altair Computer Convention
Figure 22. Harry Garland and Roger Melen Garland (left) and Melen were the first customers of MITS and the first to deliver an Altair-compatible third-party product. Later they developed their own line of personal computers under the Cromemco name.
(Courtesy of Roger Melen)
The first World Altair Computer Convention, held in Albuquerque in March 1976, was the first of the microcomputer conventions. Hundreds of people attended this event, but it was intended strictly as a MITS Altair affair. Every one of the dozen or so speakers and presenters was there at MITS’s invitation, including one who demonstrated a backgammon game he had written for the Altair. Cromemco was the only hardware company invited. Garland and Melen were there in person, the burly Melen a match for Roberts in size but far more reticent, the diminutive Garland bubbling with enthusiasm.
Figure 23. World Altair Computer Convention The sign announcing the first “World” Altair Computer Convention captures the grandiose aspirations and amateurish execution that typified the early days of personal computing. (Courtesy of David H. Ahl)
A number of uninvited companies sent out representatives to walk the floor and pass out circulars inviting viewers to see competitive equipment on display in hotel rooms upstairs from the conference center. Among that group were reps from Bob Marsh’s Processor Technology, whose memory boards were threatening to eat into Roberts’s profits.
The presence of the show-crashers irked MITS management. David Bunnell was so perturbed by the crashers that he went around tearing down their signs.
MITS had more to worry about than the board companies who were competing against MITS’s components. Other firms were springing up that challenged MITS’s core product, its computer. Don Lancaster’s Southwest Technical Products and Sphere were both working on computers built around Motorola’s recently released MC6800 processor.
Roberts had proposed building a 6800 machine, too. But some of his employees, including Paul Allen, opposed this new venture as a distraction.
“No, Ed,” Allen objected. “We’ll have to rewrite all our software for the 6800. We’ll have two instruction sets to support. That just doubles our headaches.”
Roberts prevailed. MITS did develop a 6800 machine, starting work on it late in 1975. Named the Altair 680b and attractively priced at $293, that computer was substantially different from the original Altair 8800. Components from the 8800 could not be used in the 680b, nor could the original Altair BASIC.
When the new computer magazine Byte unveiled Southwest Tech’s 6800 computer in November 1975, the announcement was soon followed by MITS’s announcement of its 680b. Additional engineers were hired to work on the new design, and new employees were added. The struggle to keep up with the orders for the 8800 and the determination to rush out the 680b had swelled the ranks of MITS employees from 12 to more than 100 in just a year.
The Role of Software
One of MITS’s new employees was Mark Chamberlain, a quiet University of New Mexico student with a knack for understatement and a taste for assembly-language programming. Chamberlain had worked on a Digital Equipment Corporation PDP-8 computer, probably the closest thing to a microcomputer that most universities had at the time. “I had done a lot of assembly code…and got so turned on to it that they just couldn’t keep me out.” When a professor mentioned that a small company named MITS was looking for programmers, Chamberlain made an appointment to talk to its software director, Paul Allen.
Allen wasn’t sure where MITS was headed and wanted Chamberlain to know the risks involved. Allen had willingly accepted these risks but wasn’t about to inflict them on the unaware. He hired Chamberlain but warned him, “If it doesn’t work out—well, it doesn’t work out.” Chamberlain appreciated Allen’s candor and commenced writing software for the 680b, a machine that “was not enormously successful,” Chamberlain recalled dryly. They had already encountered serious difficulty with the product. “Lots of [the 680b machines] were ordered, but when I came on board at MITS, the whole project was already in trouble. They had to go through a complete redesign.” Despite the revamping, the 680b never really took off. But Chamberlain found plenty of other work to do at MITS. Roberts had other machines in mind, and each of them required new software.
Meanwhile, Allen and Gates were putting increased effort into their own company, Microsoft. Throughout 1975, Gates, Allen, and Rick Wyland, who was hired to write 6800 BASIC, were branching out with their versions of BASIC, including developing versions for other companies. The relationship between Microsoft and MITS was becoming less clearly defined as the two companies grew.
The fact that Bill Gates had yet to write the disk code for the Altair 8800 didn’t help matters, especially because Gates, on leave from Harvard, was considering returning to school. Paul Allen, in his role as MITS software director, nagged Gates about finishing the code. According to Microsoft legend, in February 1976 Gates checked into a motel with some pens and a stack of yellow legal pads. When he came out, he had finished the disk code.
Figure 24. Bill Gates Gates temporarily abandoned his glasses while speaking at the first World Altair Computer Convention in Albuquerque in 1976. (Courtesy of David H. Ahl)
By 1976, the switch from dynamic memory to static memory (two means of maintaining information in memory) seemed to have solved the vexing problem with the memory board, but MITS still had to either troubleshoot the dynamic boards already in the field or buy them back. Early in that year, MITS revamped its quality-control procedures in an attempt to increase efficiency in manufacturing. MITS was already shipping the 680b and planned to ship the upgraded 8800 by midyear. A rudimentary disk operating system written around Gates’s disk code was scheduled to be released in July 1976.
Anyone who owned an Altair had probably written a program for it at one time or another. Mark Chamberlain was now maintaining a library of software submitted by Altair users, thereby setting a precedent for the industry. Chamberlain was distributing such programs as widely as possible throughout the community of users, which was a smart move. Sharing of software vastly increased the value of the machine. In particular, he sought software for the new 680b. When Paul Allen announced the price for the 680b BASIC, customers recognized an already familiar tactic. The BASIC cost nothing with the new 16K memory board, but $200 when purchased without it.
The S-100 Bus
By the middle of that year, the competition that Roberts had long feared was becoming a reality. A new company named IMSAI imitated the Altair design and brought out its own computer, the IMSAI 8080. PolyMorphic Systems introduced what looked like a serious competitor to the Altair, the Poly 88. And, in July 1976, Processor Technology grabbed the front cover of Popular Electronics with its Sol computer, named after the magazine’s editor. Even MITS’s loyal board supplier Cromemco was developing a CPU board designed around the new Zilog Z80 microprocessor as the successor to the Intel 8080 chip that was the heart of the original Altair computer. The Z80 was designed by Federico Faggin, who had left Intel to start his own semiconductor company after his work on the Intel 4004. This new microprocessor was catching a lot of attention among the high-tech hobbyists.
None of the new microcomputer companies represented an immediate threat to MITS’s market share for microcomputers. In that arena, MITS reigned uncontested. But all of the machines from these start-ups could, in principle, use the same circuit boards as the Altair. They all had the same 100-line bus structure and, as Roberts viewed things, that bus was the key to compatibility in that it allowed competitive boards to be plugged into the Altair. He typically referred to the system as the “Altair bus” and wanted others to do the same. When some didn’t comply, David Bunnell suggested sarcastically that they call it the “Roberts bus.” The bus-naming story typifies the curious mix of competitiveness and camaraderie in the nascent computer industry. The bus became a major point of contention between MITS and the rest of the microcomputing world.
Roberts’s position was simple: he and Yates had designed the bus just as they had designed the Altair. Therefore, it was the Altair bus. His competitors preferred not to share his view. The advertised name for the device grew to absurd lengths in order to credit just about every manufacturer. It was billed as the “MITS-IMSAI-Processor Tech-PolyMorphic bus.” Garland and Melen talked about the bus-name problem on a flight from San Francisco to Atlantic City, where PC 76, an early microcomputer conference, was held in August 1976. Garland and Melen were about to release a CPU board for the Altair bus and were reluctant to refer to it by a lengthy list of competitors’ names. They agreed about two things: the name of the bus should not favor any one company, and it should suggest an item that’s been engineered. For instance, the name could consist of a letter and some numbers. They liked the name “Standard 100,” and in keeping with their theme, shortened it to “S-100.” That, they thought, sounded sufficiently official.
Their next goal was to secure the approval of other hardware vendors. Melen recalled, “On the same airplane were the people from Processor Technology, specifically Bob Marsh and Lee Felsenstein. I had a can of beer in my hand, and in the course of our discussing the Standard, the airplane hit a little bump, and I spilled my beer on Bob. He agreed [to the new name] very quickly, to get rid of me and my beer can.” The name S-100 became the common coin, although MITS and Popular Electronics stubbornly clung to the name “Altair bus” for a long time. Seven years later, Ed Roberts was still adamant about it: “The bus was used by MITS for two years before anybody else was producing a computer. It’s the Altair bus. Calling the Altair bus the S-100 bus is like calling Mona Lisa ‘Tom Boy.’ I’m the only one in the world who’s irritated by that, but I’m irritated.”
In addition to the S-100 companies, MITS was witnessing disturbing signs of competition from other, even more unnerving sources. MOS Technology, a semiconductor company, was doing well with Chuck Peddle’s KIM-1, a low-cost hobbyist computer built around its own bargain-basement 6502 chip. This fact alone may have occasioned no immediate alarm, but two months later, in October 1976, Commodore bought MOS Technology. For the first time, a large and well-established company with extensive channels of distribution for electronics products would be selling a microcomputer.
Roberts was justly worried. He remembered how Texas Instruments had stomped all over the calculator business.
But an even more ominous threat was looming. Tandy Corporation, having “just gotten through killing off Lafayette [Electronics],” as Peddle put it, was casting about for a computer to sell in its hundreds of Radio Shack stores. “What Radio Shack wanted to do was to come up with a packaged machine,” Peddle said, “because they knew their guys couldn’t support, and couldn’t design, this kind of thing.” Radio Shack, with its stores all over the country, could sell thousands of personal computers at rock-bottom prices.
With semiconductor companies and electronics distributors getting into the act, the competition was gearing up.
The Fall
Q: Did you think it was going to go under?
A: All the time, all the time.
-Bill Gates
The troubled MITS had more to worry about than its competition. The company had grown too big too fast. “We had too many irons in the fire,” Roberts admitted later. “We had a lot more things going than a company the size of ours should have had.” The faulty memory boards that were still out there in the field were just one of the problems. Quality control was not particularly effective, and customers were complaining. Projects were often launched despite the reservations of many MITS employees. A number of products failed.
Problems and Changes at MITS
“The high-speed paper-tape reader is a good example,” Mark Chamberlain recalled, “because I know we only sold three of those.” The “spark printer” was another: MITS bought a printer from a manufacturer, rebuilt it and repackaged it, and eventually had to charge considerably more than the supplier’s retail price for the original unimproved item. Naturally, the MITS version didn’t take off. Sometimes an entire major product line was clearly a mistake to everyone at MITS except Roberts. Paul Allen objected strenuously to the 680b.
MITS’s difficulties ran deep. “It really gets into a study of personalities,” Mark Chamberlain said. “I don’t know if it’s possible to understand the situation without understanding all the aspects that were [a result of] people’s different personalities.” One thing is clear in retrospect. The channels of communication between upper-level employees and the president were not always open. “Ed isolated himself,” Gates said. “He didn’t have a good rapport with other people in the company, and didn’t know how to deal with the growth.” Roberts later acknowledged that a problem existed: “I was worried about so many things at the time that I felt like everything was a threat.”
A number of changes occurred at MITS toward the end of 1976. By that time, Roberts had brought in his childhood friend Eddie Curry as executive vice president and Bob Tindley (from the bank that had financed MITS) to help with management. But Roberts was soon to lose an important employee. Paul Allen was restless. Microsoft was becoming a more serious enterprise, and Allen was eager to take control of his own destiny. Convinced that MITS’s best days were past, he and Bill Gates began focusing all their attention on their own company.
Mark Chamberlain moved up to replace Allen as MITS’s software director.
Chamberlain found that the job bore unexpected challenges. He quickly encountered upper-level dissension over which products to build and which projects to undertake first. Chamberlain, along with his hardware counterpart Pat Godding, did not always agree with Roberts on critical decisions. In holding on so tightly to the reins of his company, Roberts may have been trying to shield others from all the uncertainty and vulnerability of the fledgling industry. He took all the responsibility upon himself rather than allowing others to share it. As Gates acknowledged, “Nobody really knew what was going on. So many things would have obviously needed to be done if you’d had the vision back then. Nobody had the view of the market.”
“He did have ideas,” Chamberlain said of Roberts. “But we didn’t fill out the product line; we didn’t provide proper support. I think that the early pioneers who used the Altair in business were up against a hell of a lot of frustration.” Among the most frustrated were Chamberlain and Godding. Too many times, they had proposed projects that they knew would generate significant income with just a small expenditure of time, only to have Roberts say that absolutely no more time would be spent on them. Convinced that their ideas had value and equally certain that Roberts would continue to reject them, they often went ahead on their projects anyway. “He didn’t know we were doing it,” said Chamberlain.
Although MITS grossed $13 million in 1976, the company was losing its edge. Its products were not regarded as anywhere near the best, deliveries were slow, and service was poor. Most other microcomputer companies at the time had similar problems to some degree, but MITS’s position in the industry led people to expect more from the firm. Furthermore, MITS had established an exclusive dealership program early on. Retailers who wished to be the only Altair dealers in their area could sell no other brands. But the knife cut both ways, and MITS began to have trouble finding dealers willing to agree to the company’s terms.
Retailers and customers alike were dissatisfied. It was not that MITS was in imminent danger of going out of business, as it had been in 1974, but its prospects were not good and the competition was getting serious. By now, more than 50 hardware companies had entered the market. At the first West Coast Computer Faire held in San Francisco in the spring of 1977, Chuck Peddle was showing Commodore’s PET, a more serious machine than the MOS/Commodore KIM-1 and a formidable competitor to the Altair machines; also, Apple introduced its Apple II amid fanfare that signaled a change in the market.
Selling the Company
On May 22, 1977, Roberts sold MITS to Pertec, a company then specializing in disk and tape drives for minicomputers and mainframe computers. “It was a stock swap,” Roberts said. “They bought MITS for essentially six million dollars.” Whether Pertec got a bargain or a dud depends on the degree to which Pertec management was responsible for MITS’s ensuing slide into oblivion.
Roberts had talked to other companies, especially semiconductor companies, before deciding on a buyer. Pertec had offered him not only personal stock in the company, but his own private research-and-development lab and the freedom to use it exactly as he pleased. The opportunity to work on new products and to somehow tie his fortunes to MITS undoubtedly meant a lot to Roberts. But he mostly just wanted to climb down off the nose cone. The bust of the calculator venture still haunted him, and he knew a similar disaster could very well happen with personal computers.
“Once you’ve been there,” Roberts said, “staying awake every night wondering whether you’re going to make payroll the following day…you’re pretty gun-shy, and you’re making decisions that aren’t terribly logical.”
The Pertec sale led to acrid fighting over ownership of the software. Gates and Allen had no intention of handing their BASIC over to Pertec. They had written the core of the BASIC before even meeting anyone from MITS, and, unlike Allen, Gates insisted that he had never been a MITS employee. “Pertec thought they were buying the software as part of the whole deal,” Gates recalled. “And they weren’t. We owned the software. It was all under license.”
Suddenly, the whole deal was in jeopardy. Gates later recalled the head of Pertec telling him that if the software were not included in the transaction, Pertec would back out of the deal. If that happened, MITS would fold. The pressure on the boys was tremendous.
“They sent out this big-time lawyer,” Gates recalled, and the matter went into arbitration. When it was all over, Gates and Allen had prevailed. The software belonged to Microsoft. Fortunately for MITS, Pertec went ahead with the purchase anyway.
Ed Roberts always held that the arbitration decision was dead wrong. Years later he felt bitter and betrayed. MITS’s agreement with Gates and Allen, he insisted, stipulated that they would receive royalties on the software up to a maximum of $200,000 and then the software would belong to MITS. His company had paid them that amount and therefore owned the software. Roberts was convinced that the arbitrator misunderstood clear issues of fact. “It was a fluke,” Roberts maintained. “It was just wrong as rain.”
Roberts blamed Gates for the outcome. “Our relationship really went to heck,” Gates said. “Ed really got his feelings hurt.” Having won in the arbitration and with no ties holding them to Albuquerque, Gates and Allen moved Microsoft to their native Washington.
Pertec didn’t back out of the MITS acquisition because of the BASIC ruling, but under Pertec, MITS gradually fell apart. Even before the acquisition, the company was losing its dominant role in the very industry it had created. But MITS didn’t start its dramatic decline until the Pertec management teams walked onto the scene.
The Suits
Figure 25. Ed Roberts After launching the personal-computer revolution and riding the roller coaster of entrepreneurship, Roberts sold the company, went back to school, and settled in rural Georgia to live out a second dream as a country doctor. (Courtesy of Ed Roberts)
The Pertec people managed to alienate virtually all key MITS personnel. “They kept patting us on the head, saying we didn’t understand the business,” Roberts recalled. The MITS regulars didn’t respond well to the Pertec management teams. The standard line on them was that they were “two-bit managers in three-piece suits.” The epithet was used so frequently it was shortened to simply “the suits.”
Pertec treated MITS as if it were a big business in an established industry. Before agreeing to buy MITS, Pertec executives asked Roberts to show them his five-year marketing forecast. At the time, MITS advance planning “consisted of where things would be on Friday,” Roberts said. To please the buyers, Roberts and Eddie Curry invented projections they figured would make the Pertec managers break out the champagne. They told Pertec that sales would double each year and provided a pie-in-the-sky guess of how many machines the company could move. Pertec bought it all. Over the following year, managers came and went at Pertec in extraordinary numbers. “People based their careers on trying to live up to that [forecast],” said Curry.
Mark Chamberlain had no use for the Pertec suits who’d invaded MITS: “They sent in team after team. Each team came in to knock off the previous team. Any given team had about 60 to 90 days to turn the mess into something good, but it wasn’t enough time. It was just long enough for the people to come in and switch from a position of trying to understand the problem to becoming a part of the problem. After 60 to 90 days, you were definitely part of the problem. And they’d send in the next guy to fire you.” Chamberlain left to go to work for Roberts in his lab. “I wanted out of that Pertec thing like right away,” he said. “That thing was crazy.” For a while, Chamberlain worked with Roberts on a low-priced computer based on the Zilog Z80 chip, but he soon left to pursue other opportunities.
Others were defecting from Pertec’s MITS group. Bunnell departed at the end of 1976 to start Personal Computing, one of many significant personal-computing magazines he would eventually launch. He published it from Albuquerque throughout 1977 with contributions from Gates and Allen. Andrea Lewis took over as editor of Computer Notes and changed it from a company-written newsletter to a slick magazine with outside contributions. Eventually she accepted an invitation from Paul Allen to move to Bellevue and take over Microsoft’s documentation department. Sometime after that, Chamberlain also joined Microsoft.
Several engineering people left Pertec to work for a local electronics company. Even Ed Roberts, after five months, became fed up with Pertec. “They told me I didn’t understand the market. I don’t think they understood it.” Roberts bought a farm in Georgia and told everyone he intended to become a gentleman farmer or go to medical school. Eventually he did both, with the same concentrated energy he had brought to MITS.
Pertec gradually came to regard the MITS operation as a bad venture and eventually abandoned it. According to Eddie Curry, who stayed on longer than any other MITS principal, Pertec continued making Altairs for about a year after the acquisition, but within two years MITS was gone.
It would be hard to overestimate the importance of MITS and the Altair to the existence and form of the personal-computer industry today. The company did more than create an industry. It introduced the first affordable personal computer and pioneered the concept of computer shows, computer retailing, computer-company magazines, users’ groups, software exchanges, and many hardware and software products. Without intending to, MITS made software piracy a widespread phenomenon. Started when microcomputers seemed wildly impractical, MITS pioneered what would eventually become a multibillion-dollar industry.
If MITS was, as writer David Bunnell’s ads proclaimed, number one in the business, the scramble to be number two was won by one of the most idiosyncratic of the early microcomputer companies.
Copyright © 2014, The Pragmatic Bookshelf.