Resolve and Fortitude : Microsoft's ''SECRET POWER BROKER'' breaks his silence

by Joachim Kempin

  The largest remaining obstacle to a successful cloning of the architecture was a piece of software code called the IBM BIOS. It contained the Basic Input Output System an IBM PC needed to manage peripherals like printers, hard disks, etc. These programming instructions should have been an integral part of two the OSs. Instead, IBM had written the essential software herself. Without it, no one could replicate her PC architecture. Worse, IBM was showing no intent to or interest in licensing it to lurking competitors. To protect it from unauthorized use, she had copyrighted its code. An insurance policy and a company called Eagle Computer soon felt IBM’s legal wrath after copying it without permission. But as with all software, clever programmers can legally overcome such obstacles.

  A small entrepreneurial ISV in Massachusetts called Phoenix Computer Systems was first in line attempting to clone IBM’s code. Reinventing software, even if copyrighted, is perfectly legal as long as the imitator writes a functional equivalent piece of code and refrains from identically replicating the original instructions. This was understood as being one of the vulnerabilities in the software industry and easier to accomplish when the code’s volume is relatively small. To pull this off, developers use only the functional specifications of the targeted software as a starting point. As long as this principle—called a dark-room approach by attorneys—is strictly followed, the newly from scratch created program is deemed legal and will survive court challenges. Phoenix encountered none.

  Once available, companies such as Compaq, Acer, Nippon Electric Company (NEC), Hewlett-Packard (HP), Zenith, Dell, Siemens, Victor, Olivetti, Tandy87 and Osborn took up the challenge and designed IBM PC clones. Under Bill’s directions, MS overtly promoted them with valuable tech insight and advice, which did little to engender warm and abiding friendships within IBM. Energized to compete, the avalanche of new entries accelerated the affordability of PC technology and presented MS with the opportunity to win new customers for MS-DOS. Not to be outdone and with a highly compatible but not identical OS at hand, DRI immediately engaged as well, drawing MS into a fiercely and long lasting battle to win over emerging PC manufacturers.

  All were joined by risk taking ISVs drawn to potentially high volume sales opportunities and, compared to Apple’s, a promising and more homogenous application environment. Business and home users alike quickly recognized the groundbreaking benefits of choice, speed, and efficiency provided by multiple vendors. Accessible and affordable PC power and software had arrived once and for all! On their own desks. In their own homes. As Bill predicted, truly profound pioneering advances in productivity and ease-of-use freed consumers from timeshare, minicomputer, and mainframe dependencies, gutting IBM’s terminal sales, to her detriment. For existing computer paradigms, the wave created by the desire to buy affordable PC clones had just turned disruptive.

  Coinciding with the launch of her PC model AT/33988 in ’86, IBM began preinstalling MS’s OS exclusively on all her PCs. The crucial change of heart signaled the beginning of the end for DRI’s product—and in the swift trajectory of tech history eventually doomed the company herself. The market had spoken and IBM confirmed customer sentiments.

  Not so fast, changing to a pre-installation routine was not caused by an overwhelming preference for MS-DOS alone. Like always, in the IT industry competitive pressures were solely to blame! The PC world had just gotten more hassle free and user-friendly. IBM’s competitors, influenced and encouraged by MS sales reps, forced her hand to deliver out of the box ready to use systems. Costlier or not, the industry leader bit its tongue and gruntingly—after five years—lastly followed competition.

  Apple, the early pioneer, was hardest hit by the onslaught. By ’83, IBM PC clones outsold Apple’s PCs three to one. The coveted number one spot was taken over by IBM and never regained. Adding to Apple’s misery was collective reluctance among the business community to buy IT technology from a dwarf like her. Wasn’t an IBM approved technology a much safer bet for the future?

  PC technology took off like an Atlas V rocket. Before it got introduced, most computing tasks were accomplished by using terminals connected to either mainframes or minicomputers. Their central computing power was shared between several users simultaneously. This often led to bottle necks determined by computer capacity or connectivity bandwidth. End-users at home had no access to computer power at all. With the arrival of the PC later accompanied by network connectivity via regular phone lines, the stone age of computing had finally come to an end.

  Usability and access were not the only obstacles plaguing the IT industry. Applications specifically written for one computer system in native assembler programming language would not run on any other without costly adaptations. The invention of high-end programming languages like FORTRAN89 and COBOL90 improved that. A program written with such programming instructions promised to be written once and then run on any computer system as long as the respective language translator—typically a compiler—was hosted on the targeted system. Its role was to transfer the high-end instructions into system specific code the computer could execute. Recompiling a program was fast and cheap! Problem solved.

  Unfortunately, the availability of these high-end and supposedly standardized programming languages did not mark the end of the incompatibility story. Leave it to the restless human mind to innovate, improve and embellish them with non-standard extensions. Proprietary in nature and therefore reintroducing incompatibilities, programmers used them anyway to optimize programs and gain execution speed. Trade-offs such as these, while well-intended, played right into the hands of computer manufacturers who assumed that any incongruities between systems impaired the possibilities of customers switching vendors. Investments, in particular in databases and other commercial applications, often exceeded customer’s hardware outlays. Having to throw these away when changing hardware vendors constituted, in the eyes of computer manufacturers, a formidable barrier of entry for competitors, casually described as a choke point or legally characterized as a lock-in situation. IBM had created one of the largest with her 360 mainframe system, which was followed by every minicomputer vendor under the sun with its own unique and proprietary approach.

  Customers, though, did change computer vendors. The lock-in chains were always weaker than imagined and feared by advocates and politicians. Entrepreneurial spirit combined with brave risk taking always found a way to break them. When I worked for Digital Equipment Corporation (DEC), I brought the thinking of a considerable number of clients around, switching them from IBM’s to DEC’s gear, guiding them through the process. The highly competitive IT industry contrived multiple disrupting technologies over decades. They offered so many advantages to potential buyers that switching of vendor platforms every so often became an absolutely necessity, if one wished to remain competitive and cost conscious. PC technology was dead on one of them.

  Bill’s personal vision and goal was to fundamentally change the status quo of the nascent PC industry way beyond what the early micros and their successors had already achieved. First, he promoted and supported the cloning of IBM’s PC architecture. Second, he helped establish MS-DOS on most PC systems. (DRI scored on the rest with her relative MS-DOS compatible CP/M-86) The caveat? Hardware brand became less important than software marque and changeover inconveniences disappeared overnight! A properly installed MS-DOS or CP/M-86 bridged any vendor gap perfectly well, serving as the magic glue to make the normally tricky application compatibility issue stick. This allowed the industry to address a set of novel of performance challenges and provide for healthier vendor differentiation. As a result, application programs sold in higher volumes than ever, and to the joy of consumers, they sold at progressively cheaper prices.

  As that notion caught on, Apple’s attractiveness shrunk. I dubbed the resulting and overwhelming creation process a 24/7 phenomenon. The lights at Apple went out each and every day when engineers left heading for home. Not so in the PC clone industry. Around the globe PC vendors, their component manufacturer
s and ISVs never stopped implementing innovative ideas and producing compelling products. Opportunities popped up nearly overnight and attracted groundbreaking entrepreneurs who took their chances. Absolutely amazing to witness, creating not only legions of satisfied fans but more importantly attracting additional buyers. Prices came crashing down as the competition was heating up and everybody was striving for the number one spot.

  Becoming a volume leader in one of the lucrative segments was a coveted rung on the ladder. The exceedingly competitive and never-ending struggle to achieve and defend this was nevertheless not for the faint of heart. Considering the rewards, the enduring mêlée produced frothy and protracted competitive wars. The winners took no prisoners. Characteristically, they took most or all of a certain market segment, something as stressful to defend as to obtain. MS, always under pressure, performed accordingly.

  The economic consequences in regard to the emerging PC business model would later be described as being fueled by network effects. The more people used PCs, the more economical the model turned out. Leading us to the 64 billion dollar question: Would Bill’s groundbreaking realization lead to a computing paradise or a choke point known as MS-DOS compatibility? Answers differ without challenging the following: Achieving and policing MS-DOS and later MS-Windows compatibility created a pro-consumer marketplace. The issue yet to be settled was who would or should ultimately be allowed to control this rising pyro-software empire? MS, the PC manufacturers, or the Feds? This keystone topic, volcanically controversial and conceptually riveting, eventually evolved into the most media-exploited and judiciously contorted battle in recent IT history. Dissecting how unfairly this bloody war was fought and eventually concluded was one of my penultimate inspirations for writing this book!

  Early on, most PCs were sold through retailers consisting of specialized mom and pop shops, large chains such as Best Buy, or anything in between. In 1983/84, at the end of President Reagans first term, the economy finally climbed out of a 16 month recessionary period. Jubilant and hopeful, the industry bet on sky high expectations for accelerating PC demand. Component manufacturers wildly expanded production capacity. New competitive entries further distorted the picture. While PC usage and sales did expand tremendously, the entrepreneurial industry nevertheless overestimated sell-through. The till-then rocketing component producers throttled back—too late. Their over-expansion left lots of goods unsold in volume, piling up in warehouses. Innovation accelerating, the inventoried components had a shelf life of six or nine months at best. By the end of ’84, producers were forced to find a home for their surplus. Either that or begin fearing for their business lives. With their traditional direct customers, the larger PC manufacturers, saturated and at full capacity, they experimented with pushing overstock through a fledgling component distributor network.

  Selling below or close to costs jump-started a whole new breed of start-ups, whose entrepreneurs understood PC demand could further be stimulated by lowering prices. In particular, small business owners were no longer willing to accept the premiums that brand name PCs demanded. Not the only concessions they were looking for—like advanced users, they were on the hunt for customized PCs in configurations most retail stores did not offer. The juicy wave of bargain basement PC components allowed responses to both. Soon, small system-builder shops—less respectfully called screwdriver outfits—sprung up all over the world, specializing in customized PCs at reasonable cost. Most of them found their clientele in their local communities and gained their trust by supporting them well.

  Other emerging stars from these frugal beginnings, notably Gateway and Dell Computer, expanded their horizons in ’85/86 beyond local business opportunities. They envisioned a nationwide direct sales burst, accurately predicting tech-savvy users could be attracted by the ability to order customized PCs by phone. By no means a slam-dunk, however. There were inherent investment risks in launching such a pioneering high-tech marketing venture over the impersonal land-line frontier. Hiring technically knowledgeable sales personnel and using cleverly crafted print and TV advertising, their phones started ringing. And ringing. Buying PCs through mail-order took off and rapidly established itself as a valid procurement alternative. Manning phone banks and streamlining operations were soon the biggest challenges for these newcomers who were coping with unexpected and explosive demand. A new way of acquiring PCs, allowing tech-savvy customers to circumvent the middle men, had passed the test and was there to stay! With increasing technical knowledge this buyer segment continued to expand, eroding the benchmark standard of PC prices. Unburdened by high overhead costs, manufacturers conquering the new terrain were reasonably profitable from the get-go. Established retailers and old-school marketers sat up and took notice. The tides were turning. Business was being swept away from under them.

  The reigning brand-names grudgingly acknowledged the emerging system-builder competition and fast expanding mail-order rivals. Their empires had blossomed into well-established and financially sound business institutions. Goliaths versus the Davids of the new world order! With the old guard deluded by a short history of success and already hobbled by inertia, the new players experienced no meaningful resistance, gaining a substantial foothold. With giddy start-ups nipping at their ankles, the stubbornly entrenched establishment missed a glaring opportunity to aggressively plunge into the fray and counterpunch the new competitors when they were still vulnerable. Helped by Reagan’s tax policies and increased deficit spending, the accelerating economy increased PC demand across the board. Flourishing upstarts discovered plenty of room for unfettered growth. The quantifiers, pundits and hired-gun market-mavens missed their rise. They either disregarded the emerging phenomenon altogether or analyzed it incorrectly for quite a long time, underestimating world-wide PC demand and unit sales. Professional analysts were alarmed to discover that by 1987/88 nearly 25 percent of WW PC volume was sold by system-builders or direct marketers.

  Instead of paying attention foremost to changing trends in PC purchase behavior, retail-driven brand name OEMs91 were pre-occupied with building larger and highly efficient production facilities locally and streamlining their operations. To whoever wanted to hear it, they crowed on at length and in detail, exhorting how efficient their manufacturing process had become and proudly offering tours to inspect their grand, stalled kingdoms. In hindsight, the trend of investing in high-wage burdened local manufacturing plants in the ’80s—dubious at best—diverted management’s time and focus from the increasingly threatening sales channel transformation. A first sign for shake-ups to happen!

  The technology improvements and related performance advances in PC software and hardware from ’81 to ’86 had been truly astonishing. Intel’s Central Processor Units (CPUs), the main performance engines of the IBM PCs, were now at least eight times faster than six years earlier without experiencing substantial price increases. Moore’s law92 was holding up nicely. Mainstream PCs had gained increased memory size and speed, and hard drives of vastly greater capacity and shorter access times had become common. Data access had gained velocity from wider internal bus architectures. They had widened from 8 bit to 16 bit internal computing highways with 32 bit designs knocking at the door. These hardware engineering advances led to new usages. A PC was no longer merely a desktop device. The most powerful ones worked as office servers, effectively competing with low end minicomputers and signaling an end to this approximately 25 year old technology.

  In the expanding universe of Central Processor (CPU) technology, Intel was the undisputed leader followed by AMD,93 which made her mark selling Intel clones at lower prices. Intel had achieved at least 80 percent market share and defended her turf with shrewd marketing tactics, often stopping AMD from making inroads into the top brand name manufacturers. We observed with astonishment how Intel consistently got away with seemingly anti-competitive actions. AMD complained to the Department of Justice (DOJ) and the Federal Trade Commission (FTC) and so did customers, but so far no government action wa
s forthcoming. AMD enjoyed most of her success with European, Asian, Latin American and lower volume US manufacturers. The cheapest PCs in 1986 were still Intel 8086, 8088 or 8018694 based. The bulk of the market had moved to systems powered by advanced 80286 CPUs. In ’86, Compaq stunned IBM by delivering the first 80386 based PC, using an expanded Compaq designed memory bus system including memory protection and 32-bit data transfers. The race to 32-bit computing was on! Constrained by production shortages, Intel played a masterful game—or a dirty one depending on what side you were on—allocating production capacity to her most loyal customers. The industry was at buzz.

  When Intel soon thereafter developed and released a Reduced Instruction Set Computer (RISC) CPU, named i860, as a response to a similar Sun Microcomputer offering, the industry was immediately rife with rumors. How secure was the future of Intel’s current architecture built on Complex Instruction Set Computer (CISC) technology? Would RISC technology prove disruptive enough and hinder Intel to produce a successor to her flagship 80386 CPU? As a consequence, most software, and in particular OSs, would need rewriting from scratch. Fortunately the rumors were short-lived, and as insiders we quickly understood Intel had already started working on an upward compatible successor called the 80486 eventually to be released in ’89. Lucky us!

  In 1986 we released version 3.2 of MS-DOS. Our intense sales and marketing efforts had by now restricted DRI, our main competitor for the IBM clone business, to primarily serve the low end of the PC market. Her prices were now half of ours and counting. Her management had learned from earlier mistakes. Like AMD with CPU sales, DRI had gained ground in developing countries and with a few larger European manufacturers. Most dominating brand name OEMs were for now firmly in our camp. An OEM buying primarily AMD’s CPUs was always a danger sign for me, as they were drawn to DRI like flies to the honey pot.