DOS Development History - Part 1


Prior to the invention of DOS, there was a progression of events in the world of technology that paved the way. Before the transistor was invented, mainframe computers ran on vacuum tubes and their heat emission prohibited the miniaturization of electronics. Without the ability to engineer a small computer system due to the heat dissipation issues, the line in the sand was effectively drawn between bulky, room-sized machines and the introduction of desktop systems. This is the point of departure in tracing the genealogic origin of DOS.

The transistor was invented in December, 1947 by a Bell laboratory team lead by William Shockley. The device was not officially released until its announcement in June, 1948. This early transistor was the point-contact type, meaning it used a Germanium base with two wires. When current was applied to the wires, it created an electric field and stimulated the release of phosphorus, which in turn produced a positive-negative polarity.

Shockley went on to invent the junction transistor in 1951. The junction version was based on an extruded Germanium base doped with negative impurities in one layer and positive impurities in the other. By melting the crystalline material in layers the transistor was manufactured, then sliced into individual blocks with wires attached to the negative and positive materials. This transistor type proved a success, and was licensed to RCA and General Electric as an alternative to vacuum tube technology.

In 1953 Bell labs perfected the extrusion process for pure silicon and managed to create silicon wafers to be used as conductive surfaces. This was followed by the ability to mask impurities onto the wafers using silicon dioxide, which acted as an insulator and template to apply smaller amounts of doping materials strategically to produce specific junction points for current flow. In 1954 Texas Instruments began to manufacture transistors, and in 1955 IBM® released the world's first transistorized calculator.

By now transistor technology was finding its way into mass production with radios, but computers had not changed size and the marketplace for smaller systems had yet to materialize. The problem, despite some advances in transistor design, was that parts were still large and individual, leading to the dilemma known as the "tyranny of numbers", the inhibition of advances to electronic design based on the number of individual components required (resistors, transistors, and capacitors).

In 1956, Shockley negotiated with Beckman Instruments Co. for financing, and went on to form his own company, Shockley Semiconductor Laboratory in Palo Alto, California. He was awarded the Nobel Prize in Physics in the same year. In 1957, eight scientists at Shockley Semiconductor pooled their money and independently developed a technique for multiple silicon transistors on a single wafer. This team, led by Robert Noyce, later went on to form Fairchild Semiconductor.

While Noyce and his staff were working on their design, an independent research effort was under way. In 1958 Jack St. Clair Kilby, an engineer at Texas Instruments, was designing his version of the integrated circuit. His invention was parallel to the Fairchild design for miniaturized integration of transistors into a single chip. Despite Kilby being credited with the invention of the integrated circuit, it was Noyce and Gordon Moore who prevailed with their patent filing and being awarded the patent for the silicon-based semiconductor in April, 1961.

After selling IBM® their first production order, Fairchild continued to develop semiconductor technology as other manufacturers began to emerge offering smaller-scale transistorized products, some mass-produced. The industry was getting its footing and by 1963 the marketplace had integrated circuitry devices for a number of consumer electronics.

In 1964 an invention of great significance to the eventual development of DOS occurred at Dartmouth College in New Hampshire. John Kemeny had been appointed chairman of the Mathematics department and Thomas Kurtz was senior professor. At the behest of Kurtz, who wanted to expand the availability of computing time-sharing services at Dartmouth to students, they worked together to develop a programming language easy enough for students and faculty to use. In May, 1964, the Beginner's All-Purpose Instruction Code was tested successfully and BASIC was officially released to the user community. The invention of BASIC hailed a new beginning in the ability of programmers to interact with the computers of the time, and it was quickly adopted.

As solid-state electronics design took hold, the portability of machines was starting to come of age as large devices were replaced with smaller, IC-based designs. In 1965 TI began development of the Caltech Project, a small, hand-held calculator that saw its prototype debut in 1967. Working with Sharp and Sony in Japan, the use of Integrated Circuit technology was quickly starting to see manufacturing on an assembly line scale. Fairchild Semiconductor became Intel in 1968, and continued its development path on Silicon-based micro-electronics. TI stayed focused on refining the IC design for calculator functions.

Around this time, the future inventor of the first to market consumer PC appeared. Ed Roberts was an Air Force ROTC student at the University of Oklahoma, graduating in 1968 with a degree in Electrical Engineering and assuming active duty. He was assigned to the Air Force Weapons lab at Kirkland Air Force Base in Albuquerque, New Mexico where he worked on a variety of electronics research projects. Roberts and a lab associate named Forest Mims went on to form a small garage business making model rocket and airplane controllers named MITS - Micro Instrumentation Telemetry Systems.

Their business was a part-time operation and sales were poor, prompting Mims and two other lesser partners to resign their interests in the company to Roberts for $100 each. With MITS now a sole-ownership company, Roberts evaluated the business and decided to embark on the production of a hand-held calculator kit. Utilizing the chips from TI, Roberts began to experiment with various designs and soon landed on a prototype he intended to market. After refining the design he began to sell desktop calculators by mail-order. The kits were a success, and soon contributed to the stability and growth of MITS, which Roberts then moved to a larger facility in Albuquerque to increase production.

As Ed Roberts continued to work with integrated circuits building calculators, the second and third significant people in the progression to DOS were about to come on the scene. Bill Gates was enrolled in a private school in Seattle named Lakeside known for its academic reputation. In 1968, the Lakeside Mother's Club, a social fundraising effort, decided to have a rummage sale. The proceeds from the sale were used to install a ASR33 Teletype with purchased time on a GE mainframe (arguably visionary on their part given the cost).

Gates and his fellow Lakeside student and friend Paul Allen both shared an enthusiasm for computer technology. Soon after the ASR33 was installed, the two of them spent as much time as possible with the device. Being incredibly curious and demonstrating a keen business acumen at an early stage, they got on every user group list they found, made up company names to mail off for literature, and paid close attention to computer companies and what they were up to. After they succeeded in using up all the Lakeside mainframe time writing programs in BASIC with loop statements, they convinced a local company named C-Cubed to let them have time on their DEC PDP10 system in return for hunting down bugs in their code.

This proved to be a valuable period for the two, as they reverse-engineered operating system code to learn everything they could, working with programming languages like Fortran and LISP. When the C-Cubed time finally ran out, they found a company in Oregon (Information Sciences, Inc.) running a PDP10 that hired them to write a payroll program in COBOL. This led to the formation of the Lakeside Programming Group as Gates and Allen were joined by two other friends, Ric Weiland and Kent Evans in 1971.

Gates and Allen formed their first company with another friend named Paul Gilbert called Traf-O-Data. Gilbert had designed a device to automate the reading of collected traffic data by counting cars across a rubber tube in the roadway. They went on to generate $20,000 in revenues with the device and landed job offers working for TRW in their software division.

About this time, the Intel 8008, an updated 0.2Mhz 8-bit version of the 4004 processor, came on the market with more registers and the ability to address a memory space of 16k. These processors were designed primarily to be terminal controllers and perform rudimentary binary logic. This was followed by engineers at Texas Instruments developing the TMS1000 single-chip micro controller, integrating a 1k ROM, 32 bytes of memory, and a 4-bit processor on a single unit. The chip went on to become the foundation for the TI calculator line beginning in 1972 and was mass-produced in the TI Datamath 2500 calculator.

As TI production increased and prices dropped, Ed Roberts and MITS were leveraged into a difficult business position and began to lose money to TI's low price and retail availability. By 1973, MITS was starting to hurt badly from TI's dominant market share. Allen went on to Washington State University, and from there he went to work for Honeywell in their Boston office working as a software programmer. Gates had enrolled at Harvard in 1973, with a focus on mathematics and law. Then, in 1974, Intel unveiled the 8080 processor, a revolutionary step forward that laid the groundwork for the personal computer. The 8080 had an 8-bit data bus and a 16-bit address bus with 256 Input/Output ports built in to pre-allocate addresses for peripheral devices. It also had a signal pin that toggled the memory stack allowing it to occupy a separate memory bank. The 8080 was a 2Mhz processor that addressed 64k of memory space.

By this time MITS was in serious trouble and needed a new product technology to save the company. Roberts had been working on the concept of a home computer kit, and together with Les Soloman, the technical editor of Popular Electronics magazine, they designed a system built around the S-100 data bus. The idea for the design was based on the larger mainframe systems of the day where individual boards would perform specific peripheral functions in the computer, and Roberts wanted to parallel the idea in miniature. After many late nights, they came up with the prototype and named it the Altair. After making a deal with Intel, Roberts secured a supply of 8080 processor chips for a low overhead, and factored in the cost of components and manufacturing to arrive at low selling price the beginning of 1975.

With the prototype and documentation in hand, Soloman submitted the article on behalf of MITS to Popular Electronics Magazine, and in January of 1975 the "World's First Minicomputer Kit to Rival Commercial Models" was introduced in a front cover ad. In short order MITS had literally hundreds of orders coming in for its new computer kit from around the country. The initial demand was so big that Roberts had to ship his first systems out incomplete. The kits were manufactured for buyer assembly, requiring wiring and soldering skills. The MITS Altair went on to become the company's savior, and paid its own way for advanced development of future models.

In January 1975, Allen saw the cover of Popular Electronics, picturing the MITS Altair. Immediately realizing the potential, he sought out Bill Gates. They talked it over, and after convincing Gates to write the code (despite having no access to the machine) Allen sent a letter to Ed Roberts at MITS offering to write a BASIC for the Altair. After Roberts agreed to see a demonstration, Allen and Gates had to develop it, not having anything to offer at the time. Working together with their associate Monte Davidoff, they designed and coded the entire project in six weeks time (Allen wrote the non-runtime code, Gates the runtime code, and Davidoff the math routines). Using the Altair's published specifications, they created a simulator on a DEC PDP-10 allowing them to emulate the MITS computer.

With the bootstrap loader for the tape reader the only missing element, Allen crafted it together at the last minute flying to Albuquerque to run the demonstration at MITS. After issuing the command PRINT 2+2 the Altair returned 4. Allen went on to compute square roots, sums, and other calculations. Roberts had said he would buy the first BASIC he saw run on the MITS, and he agreed the demonstration was sufficient to move forward. He offered Allen the job of Director of Software at MITS, which Allen accepted. Gates then dropped out of Harvard and moved to Albuquerque to work with Allen and MITS developing software for the Altair platform.

With their success with MITS and recognizing the potential to port BASIC to other systems, Paul Allen and Bill Gates renamed their independent partnership Traf-O-Data to Micro-soft, on April 4, 1975. Later dropping the hyphen in the name, Microsoft®'s client list began to grow, with NCR and Intel as their early projects. Two of their high school friends - Marc McDonald and Ric Wieland, were hired as Microsoft®'s first programmers. Microsoft® went on to modify BASIC for Citibank, DTC, and General Electric, attracting more clients along the way. In early 1976 Paul Allen quit his position as Director of Software for MITS to join Microsoft® Corporation fulltime. Operations were moved to the Park Central Tower in Albuquerque in the Spring of 1976. At the time there were then seven employees. Gates had gone back to Harvard briefly for the Spring term, but found himself pre-occupied with Microsoft® and directing the licensing efforts for product development, which over-shadowed his studies.

On November 26, 1976, Microsoft® was officially registered with the New Mexico state government as a partnership, although it had been in existence for a year since the previous November. By the end of 1976 it was well on its way to becoming a dominant force in the writing and delivery of computer software. In early 1977, they expanded their development effort into the business and science markets, releasing their versions of FORTRAN, COBOL, and their Assembly compiler, and plans to release a stand-alone copy version of BASIC were finalized and forwarded for manufacture. In February of 1977, Gates and Allen drew up a formal partnership document that gave Gates a 60/40 split of the company over Allen.

Ed Roberts, meanwhile, was beginning to watch MITS slide out of its position as an early leader in the personal computer market. Memory board failures had been due to poor quality control, and MITS dealership agreements were now causing problems as bad hardware supplied to them was further eroding the company's reputation. With the Apple II and Commodore Pet debut, the MITS Altair line was effectively challenged, and Roberts threw in the towel selling MITS to the Pertec company in May of 1977. Gates and Allen refused to allow Pertec to take rights to the BASIC code they had written, which Roberts held the license for when he sold MITS. Gates brought the matter to arbitration, and the legal decision prevailed in Microsoft's® favor in the latter part of 1978.

With their relationship with MITS now terminated, Gates and Allen decided to move Microsoft® back home to Belleview, Washington in December of 1978. Their first International affiliate was coming together when ASCII Japan agreed to market Microsoft's® OEM products as ASCII Microsoft®. By early 1979, Microsoft® now had a commanding presence with their BASIC, having been ported to a significant number of microcomputer brands. By late 1979, the European market was established by affiliation deals with Vector and Phillips. Looking beyond BASIC, Microsoft® formed their consumer products division to expand their portfolio to marketing retail products both original and licensed from third parties.

The market share for the Apple computer was untapped by Microsoft®, and Gates and Allen had a discussion about porting their products to the Motorola 6502 Processor. Realizing it would require too much effort to port the software, Allen came up with the idea of a hardware emulation. The idea was formalized and in 1980 Microsoft® released its first hardware product, the Softcard, that allowed Apple computer owners to install a 8080 processor board that allowed them to run programs based on BASIC-80 and CP/M. In June of 1981, Gates and Allen, along with Steve Ballmer, officially incorporated Microsoft® from their partnership. At this point, Bill Gates was the major shareholder in the corporation. With the departure of Paul Allen from the corporation in late 1981 due to Hodgkin's disease (lymphatic cancer), Gates was left as President and CEO to run the company.