History of Computers
Originally written for Spider Magazine 2005 annual cover story.
The Idea that Changed the World
The creation of the modern computer has changed the face of the planet. Today, there are more devices fitted with a microchip than there are human beings. The idea of a “computer” cannot be attributed to a single person. Rather, it has been the combined contribution of many innovative and forward-thinking scientists, mathematicians, philosophers and engineers that has brought us to what we now refer to as “the computer age”. This is their story…
It is no coincidence that the decimal number system rolls over after a count of 10. During a time when numbers did not yet exist, fingers, along with twigs and pebbles, were the most convenient way of tracking quantities. Around 500 B.C., the Babylonians made advances in accounting and devised the abacus — a small contraption with a few rods and free-moving beads. This device operated by accepting data (quantities), a set of instructions (add or subtract) and in return provided an answer — it captured the essence of a computer. The abacus is an example of the earliest known computing device: a primitive calculator that possesses some sort of innate intelligence allowing it to translate instructions into meaningful answers.
A century later, the Arabs invented the decimal numbering system — the basic language of mathematics. Armed with this language, it became possible to learn the answers to more complex problems such as 100+100 without having to visualise 200 pebbles. It also gave way to pre-computed answers in the form of lookup tables — the most trivial form of computational engineering. Lookup tables such as the ones used for multiplication can instantaneously provide answers at a glance without any mental effort.
The model of the abacus integrated the knowledge of the decimal number system and evolved into a mechanical calculator. During the sixteenth century, Leonardo Da Vinci, master painter and inventor, designed an adding machine that was a complex clockwork of mechanical cogs and rods with a few dozen buttons for adding and subtracting numbers. Though clunky, the machine was a testament of engineered design. The design went undiscovered until 1967 when it was found in Da Vinci’s cryptic notebooks transcribed in his usual mirror script.
Unfortunately, the adding machine performed only one mathematical function at a time. Charles Babbage devoted his life’s work to remedy that. Starting in 1837, Babbage worked on the design of a general purpose computer called the Analytical Engine. His efforts eventually earned him the title “father of computing”.
A working model of the Analytical Engine never materialised due to financial constraints but the design appeared to be sound. Were it ever to have been constructed, the Analytical Engine would have been some 30 metres long and 10 metres wide, powered by a steam engine and accept not only data (for example, the dimensions of a triangle) but also programs or functions (for computing the area of that triangle). The input would be fed through hole-punched cards — an idea borrowed from the textile industry which used such cards for automatically guiding patterns in weaving machines. The results were designed to be output through a curve plotter or a bell. The Analytical Engine was the first programmable computer and as such, it laid out the fundamental principles found in any modern computer: input, a program, a data store for storing intermediary answers, output and an arithmetic unit that performed the basic functions underlying all computation (add, subtract, divide and multiply).
Babbage’s principles found a purpose in the 1890s. The US Census Board realised that manually counting the results of the current census would take more than 10 years, by which time another census would be due. As part of a competition to come up with a solution, Herman Hollerith, an employee of the census department devised a machine that took in punch cards similar to Babbage’s Analytical Engine and added the results for over 6.2 million people in only six weeks. The idea for the system came to Hollerith from railroad operators who punched tickets in a certain manner to indicate whether the ticket holder was tall, dark, male, et cetera.
Mechanical computational engines continued to evolve and devices such as chronometers or watches became the marvel of mechanical orchestration and miniaturisation. Containing anywhere from a few dozen to a few hundred moving parts comprising clutches, cogs, gears, springs, coils and so on, these contraptions could keep a heartbeat for years or even decades with millisecond precision. The complex orchestration of these parts also explains the higher price tags on some of the more sophisticated of these watches, compared to the much cheaper digital counterparts of the modern era. Mechanical contraptions, however, have a physical limit to how small they can get and succumb to the number and size of parts, friction, weight, portability, power requirements and precision.
Fortunately, science in typical fashion made a leap during the mid-1800s. Thomas Alva Edison’s pioneering work in the field of electricity allowed it to be harnessed for practical use for the first time. With the control of electricity came the radio, the light bulb, wires and other invaluable electrical inventions.
As physics paved the way for electrical innovation, scientists discovered in electrical charge a way to represent data. The beads of the abacus were replaced by bits in the modern computer — essentially a bit or ‘binary digit’ is a small electrical charge that represents a 1 or 0. The creation of the bit marked a transition from the decimal system for humans (10 primary numbers from zero to nine) to a binary system for computers (only two numbers, zero and one).
Binary arithmetic provided the foundation for operating with bits. It was the contribution of Gottfried Leibniz, a prodigy in mathematics, philosophy, linguistics, law, politics and logic. In fact, he posited that every argument could be deduced to numbers using binary logic. Hence, 1s and 0s in binary arithmetic are also referred to as “true” and “false” (or “on” and “off” due to their application in electronic switches).
George Boole, a mathematician and philosopher, relied on binary arithmetic to advance his theories of logic, at the time still a branch of philosophy. The field would later evolve into Boolean algebra for managing and calculating the outcome of a number of bits interacting with each other. The simplest of Boolean logic might take the following form: a light switch toggles a light bulb — flipping the switch turns the light on if, and only if, it’s off, and vice versa. In the modern computer, however, a hybrid of few million such switches are attached to any single circuit and flipping a combination of these switches can achieve results that can only be managed using the techniques of Boolean algebra.
At its core, a computer is doing just that, switching a galaxy of bits on or off. A ballet of bits is constantly playing out and each flip of the switch results in a domino-like chain reaction. Each ballet of the bits is used to represent an outcome and must be orchestrated with absolute accuracy.
During a 3D game for example, the tiniest movement of the mouse turns the ball, which turns a wheel that is being monitored by a chip whose purpose is translating this movement into an electronic signal. The chip changes a few thousand bits and causes a chain reaction down the wire of the mouse connected to the computer. The reaction eventually ends up in the computer’s main processor, which in turn tells the graphics processor that the mouse has moved one millimetre. The graphics card does a few thousand mathematical computations to calculate the shadow, lighting, shading and angle of light, and generates a new image corresponding to the movement of the mouse. While it does all this, the memory in the computer does the job of remembering the previous position based on which the next image is calculated. The graphics card renders the new image on the monitor by changing the state of a few billion bits on the screen and producing a massive collage of a few million pixels — all this within a fraction of a second.
The language of bits was not always the language of choice. The idea came from the early days of telephone companies when they used switches with “on” and “off” states to connect or disconnect a circuit. Human operators made the connections by manually operating the switches. For long distance calls, a local operator connected to a foreign telephone exchange, which in turn connected to its own local exchange and created a link between the calling parties. A computer uses the same principle of using switches to control bits and direct the flow of information.
In his 1937 Masters degree thesis at the Massachusetts Institute of Technology, Claude Shannon proved that management of a large number of these switches could be simplified using Boolean algebra. Inversely, he also proved that switches could be used to solve Boolean algebraic problems. This meant that if a set of bits interacted in a particular way, they would magically result in the answer — in the same way that mixing red and green paint results in yellow.
How this magical interaction happened was the pioneering work of Alan Turing, father of modern computing. In 1936, a year before Shannon’s thesis, Turing laid out the fundamental theoretical model for all modern computers by detailing the Turing Machine. Its basic idea is quite simple: in a perfectly well choreographed ballet, for example, a dancer does not need to keep track of the entire ballet. Instead, she may need to keep track of only a few simple cues: step forward if dancer to the left steps back; spin synchronously with the lead dancer; stop dancing when the dancer in front stops dancing. Each dancer in the ballet follows their own set of cues, which creates a chain reaction among other dancers. The ballet is initiated (or brought to a halt) by a single dancer responsible for starting the chain reaction.
Similarly, bits react to cues and influence each other. When the ballet of bits concludes, the new state of bits (for example, 111, 001 or 010) represent different results. Turing’s contribution is remarkable due to the nature of the pioneering work and his thought experiments that led him to develop such a system.
Turing’s work added to centuries of advances and breakthroughs in engineering, mathematics, physics, logic and an endless pursuit of human spirit that would manifest themselves in the form of a 30-tonne machine called the Electronic Numerical Integrator And Computer (ENIAC).
The ENIAC was the first fully programmable machine capable of solving almost any mathematical problem. Built by the US Army in 1946, the ENIAC was capable of adding 5,000 numbers per second. It was powered by 18,000 vacuum tubes, 6,000 switches, around five million hand soldered joints and took three years to build. This marvel of a machine, however, was specifically programmed to calculate in a matter of hours the trajectory of artilleries to hit enemy targets. This was a task that would otherwise have taken days to compute.
The vacuum tubes used in the ENIAC were vaguely similar to light bulbs in both function and form but with metal casings instead of glass. These vacuum tubes functioned to represent data using electrical charge. However, they were problematic at the same time and kept fusing out. The heat and other lights on the ENIAC computer attracted a lot of moths which in turn caused a lot of short circuiting. Computer problems henceforth came to be known as “bugs” and fixing them, “debugging”. Due to these problems the ENIAC could sometimes be down for half a day at a time and required a lot of hands to keep it up and running.
While the input data could be stored on the ENIAC, the program to operate on the input had to be wired through plug board wiring. Programming it was cumbersome and each program required unplugging and re-plugging hundreds of wires. This method of programming was almost as primitive as Babbage’s punch cards. And the limitation meant that computers, although programmable, were restricted by the complexity of the process.
It was the mathematician John von Neumann who, shortly after the ENIAC, introduced the concept of a stored-program computer. Storing the program in the computer memory meant that the system of semi-permanent plug board wiring on the ENIAC could be deprecated. Bits would represent not only data, but also the programs themselves which consumed the data — bits controlled by bits.
The stored-program design had profound implications. Prior to this breakthrough, computers accepted normal input and passed it on to programs which operated on it. However, if the program itself was an input, then operating on this program would require another master program. Turing’s Universal Machine described such a master program and von Neumann provided the implementation which has now become the model for nearly all computers.
Even with the programmable architecture well in place, it was doubtful if vacuum tubes would allow computers to scale. These deficient vacuum tubes set the backdrop for the most important invention of the digital age: the transistor, for which its three co-inventors William Shockley, John Bardeen and Walter Brattain went on to receive the Nobel Prize in 1956.
Transistors are microscopically small in contrast to the finger-sized vacuum tubes, require lesser power and are capable of switching states (1 to 0 and 0 to 1) much faster. Their beauty also lies in their composition: as solid-state semiconductors, they are built from material that has the ability to conduct electrical charge, like metal, or block it, like rubber.
To deliver on the promise of transistors, Shockley would go on to head the Shockley Semiconductor Laboratory in Northern California with his colleagues Walter Brattain and John Bardeen. The two eventually left Shockley due to his paranoid and competitive nature (once, an employee cut her finger which Shockley suspected was actually a plot targeted toward him and to find the culprit, forced a lie detector test upon all his employees).
Along with Bardeen and Brattain, six other scientists quit Shockley Semiconductor. These “treacherous eight” — as Shockley referred to them — went on to form Fairchild Semiconductor in the same region and adapted the more abundant silicon as the semiconducting material of choice. This marked the beginnings of the Silicon Valley which today is the epicenter of computers and high-tech businesses.
Transistors which represent the bits in a computer needed to be wired together for interaction. Common configurations of wiring came together as integrated circuits or microchips, the first of which was invented by Robert Noyce, one of the “treacherous eight”. If transistors are characters of the alphabet, microchips are the words formed by those alphabets and computers are the composition of dozens of these microchips. All digital electronic devices are composed of microchips with many of them sharing the same common subset of chips.
Robert Noyce, along with Gordon Moore would go on to form Integrated Electronics, now better known as Intel. It was at Intel that he oversaw the work of Ted Hoff who invented the greatest microchip of them all. The microprocessor or the Central Processing Unit (CPU) found in all personal computers (PCs) is a single, highly complex microchip that functions as the brain.
Co-founder Gordon Moore meanwhile gained notoriety for his speculation that the number of transistors on a microprocessor would double every two years. Moore’s speculation became Moore’s Law and has held up since it was first posited in 1965. Current Intel Pentium 4 processors have the muscle of over 100 million transistors fitted inside a matchbox-size chip that is capable of adding over 5,000 million numbers per second. Contrast this with the 17,000 vacuum tubes in the 30-tonne ENIAC which could add only 5,000 numbers per second and the significance of transistor technology becomes clear. If the Greeks had an Intel Pentium 4, they could have saved themselves centuries of mathematical labouring.
Intel processors started their legacy in 1975, by powering the first commercial personal computer, the MITS Altair, with an Intel 8800 processor. Microsoft founders Bill Gates and Paul Allen would go on to develop Altair BASIC, its first programming language. Interestingly enough, in the same year, Advanced Micro Devices (AMD) — also formed by a group of Fairchild defectors — reverse engineered the Intel 8800 processor and started the long running Intel-AMD rivalry.
While the Altair was being sold as a hobbyist kit, the Apple I was the first fully assembled computer developed around the same time by hobbyist Steve Wozniak and sold with the help of close friend, Steve Jobs. The two subsequently founded Apple Computers in Jobs’ family garage. Today, 30 years later, Jobs serves as the visionary and CEO of Apple Computers Incorporated.
Developed in 1973, it was the non-commercial Xerox Alto, however, that took the title for first personal computer. The Alto, developed at Xerox PARC (Palo Alto Research Center) in Palo Alto, California, was one of a dozen inventions to come out of the research centre including colour graphics, object oriented programming and wide application of the mouse. After seeing a demo of the Alto, Apple engineers purportedly adopted the concept for their own commercial computer Lisa, which eventually proved to be too expensive and ahead of its time. The lack of commercial demand meant that over 2,000 Lisas would need to be buried in a landfill.
Contrary to IBM chief Thomas Watson’s speculation in 1943 that “there is a world market for maybe five computers,” personal computers found widespread demand in a growing market that has today reached nearly two billion units. This figure primarily represents PCs, but its siblings and cousins (cellphones, PDAs, laptops) far exceed the population of even humans on this planet. Whether in the form GPS tracking devices, rain-sensing windshield wipers or electronic hearing implants, microchips continue to shrink and integrate into our lives.
While the hunger for more powerful and smaller chips is insatiable, Moore’s Law seems to be giving away as the current generation of microprocessors are showing signs of plateauing. Even though the natural laws of physics dictate that bits can be as small as the atoms in which they are stored, we are far from reaching this atomic threshold. The problem lies in the economics of miniaturisation as increasingly expensive fabrication plants for producing smaller chips yield disproportionately diminishing returns. Nonetheless, all hope is not yet lost as scientists are already exploring the frontiers of sub-atomic particles.
Atoms are composed of a set of protons and neutrons orbiting around a nucleus. Removing protons or neutrons changes the charge of the atom to a negative (0) or positive (1), allowing them to act as bits. These electrons and neutrons are in turn made up of three quarks each. Understanding the nature of these quarks and their influence on neutrons and protons will unlock the power to make today’s most powerful supercomputers pale in comparison. If these quantum computers ever materialise, they will in theory be able to compute in a matter of days what would by today’s computing ability take a few million years.
While the shape, form and power of computing devices continues to evolve, a parallel evolution has been taking place in the related field of communication technology.
The first electronic telegraphs (including wireless) were already communicating in 1832, a century prior to the ENIAC. George Stibitz, a researcher at Bell Labs during the 1930s and 1940s, used a teletypewriter (essentially a typewriter hooked up to a telephone line) to communicate with a calculator on the other end and receive results for remote computation. This was the first time a computer had ever been operated remotely over a phone line.
The US Department of Defense, Advanced Research Projects Agency (DARPA) duly noted the missing link in computers and initiated efforts to fill the void. Around 1962, a series of memos about the “Galactic Network” laid the conceptual foundations of the internet. Shortly thereafter, Vinton Cerf received a “request for proposal” from DARPA to design a packet switched network. Cerf’s research efforts lent itself heavily to the design of the first network of computers and earned him the title “father of the internet”.
The resilience of the internet derives from the packet switched network Cerf detailed. In such a model, all information is divided into tiny packets. Each of these packets is transmitted separately and embarks on a journey to find their destination on the internet. Their only strategy to get to the destination is to ask intermediate routers (who conduct traffic on the internet) for directions to the next router that might lead the way and so on until the last router points them to their final destination. Anyone who has ever gotten lost and asked for directions can probably relate to a packet. For these packets, a dozen things can and do go wrong. They often get lost in transit, are captured by a hacker or arrive at their destination out of order with other packets.
The research and prototyping for refining the packet switched network began at the University of California at Los Angeles (UCLA) where Cerf was doing graduate work. By 1969 the Advanced Research Projects Agency Network (ARPANET) would take shape as UCLA, University of California at Santa Barbara, Stanford Research Institute and University of Utah came together to form a network.
Along with the contributions of dozens of other individuals, Cerf would go on to develop the Transmission Control Protocol (TCP) in his new home at Stanford where he had taken up assistant professorship in computer science and electrical engineering. After four iterations, the TCP suite was finalised in 1978 following an exciting demonstration in July 1977 when a packet was sent on a 94,000 mile round trip on the ARPANET without losing a single bit. As a result of its resilient design and infastructure, TCP/IP (Internet Protocol) became the standard for transferring data across networks. Relying on TCP/IP, the ARPANET grew into the internet and has since continued to scale unchecked to become what it is today.
The internet was primarily used for transferring and sharing data. It handled documents but webpages as such did not exist until Tim Berners-Lee, an independent contractor at CERN, became frustrated with the lack of ability to easily share and update information between researchers. He transformed the internet landscape by introducing the concept of hyperlinks — the links on webpages that allow them to point to each other with the click of a mouse. These hyperlinks created a ‘global web’ of linked pages commonly referred to as the World Wide Web (WWW).
As far as communication networks go, the internet overshadows the telephony network, integrates the television, radio and newspapers and challenges even our physical social realm. Its humble beginnings ultimately brought the communication revolution to all its glory not only for humans but also for devices.
Through microchips, electronic devices became aware of their own function. A chip acting as the brain inside a cellphone encodes every bit of relevant information about the host. Relying on exacting communication protocols, these devices suddenly become aware of the existence of other devices made up of similar microchips and can speak to them in a similar language.
This new species of electronic beings are continually evolving and trying to overcome their cultural differences so that an alarm clock can talk to the coffee maker in the morning or a health monitor can check our vital statistics during recovery. These modern-day slaves encapsulate tiny worker atoms which manage for us what our preoccupied minds rather not. The quality of life during our brief welcome on this planet has been elevated because of them and in return for doing everything they are told, they ask for nothing. If we are God’s creatures, then computers are ours: a manifestation of the human spirit and potential.
Further Reading: A comprehensive reference on the History of Computers
Post to del.icio.us | digg