Personal Computers and Workstations
Early personal computers
Minicomputers were followed by microcomputers, which soon became known as personal computers. They were made by a new group of companies, such as Tandy, Commodore, and Apple, and Acorn who built the very popular BBC micro.
The first personal computer that I owned was a Terak. It was a graphics workstation with an elegant Pascal interpreter. The Terak had a 16-bit processor based on Digital's LSI/11 chip set, but most of the early computers used lower cost 8-bit microprocessors. A typical computer of this generation had 4K to 32K of memory, a keyboard, a simple monitor, and two 5¼ floppy disk drives for backing store. The early monitors were black and white only, but low resolution color monitors followed rapidly. Dot matrix printers from companies such as Paper Tiger and Epson completed the configuration. Tandy's TSR-80 was an early success with hobbyists, but the Apple II was the machine that captured the imagination of schools and universities.
Most Apple IIs had two floppy disk drives but no hard disk. Floppy disks could be used to distribute software and a software market developed. The computer is running a popular flight simulator.
Photograph by John Miranda
The standard programming language was a simple version of Basic. They were easy machines to program but their success came from the rapid development of a market for software applications. This included general purpose applications, such as word processing, databases, etc. An enormous number of simple programs became available for every level of education. Computer games such as Pacman were a great success, but the application that made the commercial marketplace accept personal computers was the first ever spreadsheet, Visicalc for the Apple II.
Most of the established computer companies missed out on the personal computer and many of them went out of business. IBM was late but when they entered the market they did so with a bang. The original IBM Personal Computer (PC) was only an incremental advance technically, but IBM put its full marketing power behind it. Organizations that had been reluctant to buy personal computers from upstart companies were comfortable buying from IBM. Software developers made the safe choice of developing their products for IBM.
The IBM salesmen had been trained to throw scorn on personal computers and had a hard struggle adapting. At the first Dartmouth presentation one of our staff took pity on them and fielded questions from the floor while the salesmen took notes. In contrast the Apple representative was always welcome because she was continually showing us neat things that she did on her own computer.
This is a typical configuration of an early IBM Personal Computer. It has two floppy drives and a dot-matrix printer. The addition of a hard disk with the later model XT transformed the usability and performance. Notice the very basic user interface.
German Federal Archives
Most of the early makers of personal computers could not compete with this onslaught, but a new group of manufacturers emerged selling IBM compatible computers, popularly known as clones. In their hurry to bring out the PC, IBM used components that they bought from others companies, such as the Intel 8088 processor and a primitive operating system from Microsoft, a small company that was previously known for its Basic interpreter. I think that Compaq was the first company to build an exact IBM clone. Any software that would run on an IBM PC would run on a Compaq. For example, the dominant spreadsheet for many years was Lotus 1-2-3. The very first release stated on the box that it would run on IBM PCs and Compaqs. Although IBM dominated the personal computer market for many years, they never controlled it. When, in 1987, they introduced an incompatible replacement, the PS/2, it was too late. Microsoft, Intel, and the clones controlled the market.
Lotus 1-2-3 is often credited as being the killer application that made the success of the IBM PC. I think that is an over statement, but it was very important in forcing the clones to be completely compatible. Previously, every manufacturer had its own operating environment and software vendors were forced to create separate versions for each of them. I recall visiting Informix, who had a relational database system that ran on all types of computer. In the warehouse they had dozens if not hundreds of different versions packed up ready to ship. It must have been a nightmare. When Bill Gates announced that Microsoft would release a product only if they expected to sell at least 100,000 identical binary copies, the standardization was complete.
This is a Lisa II. It has a built-in hard disk and a single floppy. It was a commercial failure, and deserved to be, but compare its modern windowing interface with the text based interface on the IBM PC.
Courtesy of Apple Coputers
The user interface for these early personal computers were much worse than the good timesharing systems. Some applications, such as Lotus 1-2-3, were well designed but they were controlled by a crude command language. The first personal computer with a modern interface was the Apple Lisa, introduced in 1983. The Lisa was over-priced, unreliable, and woefully slow, but it had a few superb applications, such as the LisaDraw graphics program. Four models were built and I had them all.
In the mid-1980s there was a rush to build large personal workstations. They were jokingly called "3M" machines, because they had a megapixel display, a megabyte of memory, and a processor that executed one million instructions per second. The early models also had a 10 or 20 Mbyte hard disk and cost about $10,000, which is a million pennies.
The first sales were to engineers and scientists, but prices fell rapidly and sales grew fast. New companies emerged such as Apollo, Silicon Graphics, and Three Rivers Corporation, but they often floundered because of over ambitious software plans. For instance, the Three Rivers Perq had a beautiful bit-mapped interface but no applications.
Digital had a fast microprocessor, the microVAX, but the company hesitated. Their profits came from minicomputers and they were reluctant to undermine sales of the VAX minicomputers with much cheaper workstations. Unix was developed on Digital computers, but VAX/VMS was Digital's pride and joy, and their support for Unix was always lukewarm. Their later workstations were well engineered, fast, and reasonably priced, but they misjudged the market. Digital, which had been the second largest computer company in the world, never recovered and went out of business a few years later.
The company that came to dominate the workstation market was Sun Microsystems. One of the founders, Bill Joy, was also a principal developer of BSD Unix. Since BSD included the definitive implementation of the Internet protocols the Sun workstations ran well on the emerging networks. In the early days Sun's business model was to be first to market with advanced hardware and innovative software, often at the expense of reliability. When I joined Carnegie Mellon in 1985 my office computer was a Sun 2. It crashed frequently. This was mystery until a colleague pointed out that it failed when the sun fell on it in the early afternoon. Later Sun workstations were much more reliable. They were powered by Sun's own SPARC processors.
The Andrew project at Carnegie Mellon, funded by IBM, was built on BSD. Every program that ran on Andrew had to run on standard BSD. When the project began, IBM was not yet ready to ship its own Unix workstation. To everybody's surprise, IBM allowed Carnegie Mellon to buy more than 100 Sun computers for the project.
IBM made a marketing mistake in delaying the announcement of their Unix workstation until their engineers were comfortable with the reliability of the hardware and had reworked parts of Unix. As a result they were late to market and the operating system was not quite compatible with BSD. A further mistake was not to offer a high-resolution display with the first release. The display in the photograph was an experimental display and not available commercially. While IBM was resolving these problems, Sun was consolidating its position as the market leader.
Photograph by Kenneth Andreyo
Steve Jobs's NeXT computer was a glorious failure. When he was forced out of Apple in 1985 he set out to build a computer with the power of a workstation but the usability of a Macintosh. His target market was education and the business model did not depend on the economies of scale of mass manufacturing.
Photograph by John Miranda.
This photograph is from http://www.johnmirandaphoto.com/next.htm,
which has an excellent description of the NeXT computer.
Many of the technical concepts were influenced by the several visits that he paid to Carnegie Mellon in 1985 and 1986. When my family and I moved to Pittsburgh we lived for a few weeks in a large house that belonged to Carnegie Mellon. For some of that time, Jobs was also living there. Later I was the first member of the NeXT advisory board, but his main contacts were with members of the Computer Science department. From them he was persuaded to use Carnegie Mellon's new Unix kernel (known as Mach), and to include a digital signal processor, which made the NeXT computer particularly suitable for applications such as speech recognition.
Technically the NeXT computer was a powerful Unix workstation with a large bit-mapped screen, but it was full of features that were characteristic of Job's creativity. It was packaged as an elegant black cube, with few cables visible. When I first visited NeXT in Palo Alto, much of the time was spent in showing off the audio quality and color animations, which were far beyond other machines of the time. Unfortunately, the NeXT machine suffered from another of Job's characteristics. It was very late. He would tell you, "It's worth waiting for", but the market did not agree.
Many features of the NeXT computer are now standard but its importance lies in the software. Jobs eventually returned to Apple and the software came with him. Today's Mac OS X is a direct successor to the NeXT operating system. The Interface Builder used to program iPhones and iPads is a descendant of tools built at NeXT to overcome the difficulties that developers had faced in writing programs for the early Macintosh.