The Tyranny of the Keyboard

by Jay Hersh

Hardly anyone would place the modern computer keyboard in the rank of medieval torture devices such as the rack or iron maiden, and yet today there are tens of thousands of people in whom the keyboard strikes the same terror as felt by those of that heinous era. These modern individuals are people crippled, or in the processed of being crippled, by the enormous stresses placed upon their bodies by the daily utilization of this device. Many have lost their livelihood. Others confront the daily dilemma of stoically facing down the pain inflicted upon them by the keyboard use that their jobs require, or abandoning those jobs. These jobs are often well paid professional careers, making them difficult to leave. Often individuals who find themselves in this unenviable position toil on in soldierly fashion until the pain and medical toll becomes too much to bear. By that time however it is often too late, serious and long lasting physical damage is often the consequence.

With the increased prevalence of computers in the workplace and classroom, the population potentially subject to such injuries grows daily. Recognition of the problem, development of alternative computer interface technologies, and integration of such technologies into the workplace and classroom proceeds slowly. An understanding of the history of this seemingly benign device and the market forces which underlie its development will demonstrate that while not born of conspiracy the keyboard's dominance as a computer interface paradigm, and the effects of that dominance, comprise a form of tyranny towards those who have been disabled by its use. It will also demonstrate that the time has come to rethink the modes in which we interface with computers, and to begin to effect a transition from the outdated mode of keyboard input, in which the human is forced to adapt to the machine, to new modes of input, in which the machine's power is harnessed to adapt it to the user.

The origin of the modern computer keyboard began humbly with the invention of the typewriter by Christopher Latham Sholes in September of 1867. The device was patented the following year. It was improved and marketed by Remington beginning about 1877. By the beginning of the 20th century the typewriter was finding its way into businesses.

Other devices for inputting alphanumeric data were also in use by the beginning of the 20th century. Punched paper tape was experimented with for use with the telegraph as early as 1881 although Morse code was used almost exclusively until about 1917 when some of the main circuits, especially those used by wire news services like API and UPI were converted to using mechanical printers. Punched card systems, used for many years to control looms, also began to be employed as a means to store alphanumeric data for use with mechanical tabulating machines at this time.

In the 20th century a confluence occurred. Teletype machines, which married the abilities of the typewriter (which served as both an input and a printing device) with the telegraph, were developed and introduced during the 1930s. Initially these machines printed on narrow paper tape, but by the end of the '30s page printers appeared as well. The punched paper tape technology developed in 1881 also provided a means by which the user could store messages and then play them back through a mechanical reader.

Similarly punched card systems were interfaced with typewriter like keyboards that facilitated input of numerical data. Such punch card systems, referred to as keypunches, were the basis of numerical tabulating systems of which IBM was selling over one million dollars worth in 1931. Such was the prevalence of both the punch card and teletype technologies by the beginning of World War 2 that even the invention of computers would not eliminate them as input technologies for over thirty years. Instead, as we shall see, they became important components of the interface with these new devices.

World War 2 created a dramatic need for computing devices that could rapidly perform calculations for creation of numerical tables used in such activities as weapons targeting and code breaking. Even before the War however companies such as Bell Laboratories had begun work on computing devices. In September of 1940 G. R. Stibitz demonstrated the Model 1 which utilized input from a teletype keyboard or paper tape and presented output on them as well. This demonstration was additionally unique in that it was accomplished in remote fashion over telephone lines, something that would not be duplicated for another decade.

Perhaps the next major milestone was the development of the ENIAC computer in 1946. This computer utilized an IBM punched card reader as its input device and presented its output on a card punch controlled by the ENIAC. This machine was followed by the BINAC in 1948 which utilized a typewriter keyboard to input data directly onto magnetic tape. The tape was read into the computer that performed its calculations and printed out its results onto an electromechanically controlled typewriter. This mode of operations was known as batch mode computing since each set of data and instructions was submitted to the computer one batch at a time.

The '50s saw the beginning of the commercialization of computing. Computers such as the UNIVAC utilized keyboards as control consoles but their primary purposes were for numerical calculation. Input was accomplished via batch mode, predominately through the use of punched cards. Output was performed via line printers which had evolved from typewriters to provide dedicated high speed output printing. The development of commercial electromechanically controlled typewriters such as the IBM Selectric contributed to the commercialization of these main frame computers by providing a technology that allowed the computer operator easy control of the device as compared to computers of the prior generation with their numerous switches and complex wiring.

Even as batch mode computing was commercialized universities such as MIT began experimenting with multi user systems, known as time sharing systems, where each user could simultaneously share the computer's resources via their own teletype terminal. The earliest of these experiments began in the late '50s. By 1964 MIT, Bell Laboratories and General Electric had developed such a time sharing system called MULTICS. The development of time sharing systems created a need for better interfaces than the awkward teletype. It fueled the development of video display terminals (a.k.a. VDT) technologies in the late '60s. The VDT joined the input capabilities of a typewriter with features of the cathode ray tube (basically TV tubes). This allowed text to be displayed in a non printed manner that was faster and could be directly edited.

As early as 1967 Control Data and Sanders were manufacturing VDTs. Harris was selling them in 1969, and Hewlett Packard by 1972. The development of the VDT used in cooperation with time sharing computer systems represents a major turning point in the utilization of the keyboard as an input device. Prior to the advent of the VDT the keyboards used to access computers were either on teletype machines or on punch card keypunches. These machines had the inherent limitation of relatively slow input speeds due to the nature of the electromechanical linkages from which they were constructed. The VDT had no such inherent limitations. The switches connected to the keys sent electronic impulses directly to the computer, with no mechanical linkage involved. The only limitations were the speed at which the computer could buffer the input sent to it. Initially of course computer speeds were relatively slow so that the limits of the human user's typing speed were not surpassed by the ability of the machines to process the alphanumeric input. As we shall shortly see all this was to change with the coming personal computer revolution.

Almost simultaneous with the commercialization of multi user systems in the late '60s was the introduction of minicomputers by companies such as Data General, Prime Computer, Hewlett Packard and others. These computers were small single user machines typically utilized via a teletype console. Their use became more widespread during the '70s, and teletype interfaces were quickly abandoned in favor of VDT technology as the interface for these small systems.

The '70s saw a rapid expansion of main frame and time sharing systems in universities and private companies. These systems were utilized via batch mode for the main frames, or VDT for the time sharing systems. By the late '70s to early '80s however the use of punch cards had plummeted. IBM, whose success in the main frame business was due at least in part to moving their pre-computer era punch card customer base onto main frames, discontinued selling them entirely. The VDT had achieved dominance of the computer interface paradigm. This alone was not significant in regard to the expansion of repetitive strain injuries caused by its use. The next factor was to be the enormous proliferation of computers and their rapid increase in speed.

By the end of the '70s computer use had expanded significantly however it was still mostly the province of universities and businesses due to the significant costs involved in purchasing and operating these machines. The expansion of minicomputers however led to increased availability of the components used to assemble them. This led electronic hobbyists to experiment with assembling systems of their own. It was these such hobbyists who in 1977 founded Apple Computer to market a small computer system which cost $1200 upon its introduction, a cost that made the systems affordable to a significantly larger number of people. Shortly thereafter, in 1981, IBM followed with the introduction of the personal computer, or PC for short. Other early producers included Tandy Corp. With the $300 TRS model computers first available in 1977, and Osborne which produced a 28lb portable computer for $1800 in 1981. By 1985 over 3 million microcomputers had been shipped to businesses. Today there are hundreds of millions of personal computers in use.

The rapid expansion of computers saw little in the way of consideration for the human interface of these devices. The initial computers of the '40s and '50s employed teletypes or electromechanical typewriters as their interface since these components were readily available. In the same manner the personal computer revolution of the late '70s and early '80s employed electronic keyboard and VDT technology as the interfaces since these components were readily available. Perhaps the most attention paid to the human interface was by IBM in 1984, and was in regard to the compatibility of the keyboard layout of the PC/AT with respect to the layout of its predecessor the Selectric typewriter. Completely overlooked in the computer revolution was the impact of introducing machines that could easily process the input of even the fastest of trained typists, to a population of users with predominately little or no typing skills. The result of this has been a near epidemic of the incidence of repetitive strain injuries in some user populations, especially among those who make extensive use of computers without the benefit of training in typing skills.

One issue that has not yet been mentioned is the layout of the keyboard itself. The industry standard keyboard layout that possesses a virtually complete monopoly on computers today is referred to as the QWERTY keyboard because of the arrangement of the keys in the upper row. This layout is in fact different from the original layout designed by Sholes in the late 1800s. When initially introduced in 1877 the first typewriter models were utilized by typing with only two fingers. The development of 10 finger typing is attributed to a Mrs. L. V. Longley in 1878. Shortly after the concept of "touch typing" was introduced (attributed to Frank E. McGurrin, a federal court clerk in Salt Lake City), whereby typists would type without looking at the keys, having memorized their locations. These new techniques, and some celebrated typing competitions, demonstrated the worth of the new machine and led to continuing increases in sales. The subsequent superiority of typing speeds over handwriting speeds (the latter rated at 20 words per minute) was in itself problematic as proficient typists easily caused the typewriter mechanisms to jam.

Sholes himself redesigned the keyboard layout into the now familiar QWERTY layout we see so predominately today. The QWERTY layout was designed to maximize the separation on the keyboard of the most frequently used keys in order to reduce jamming of the mechanism that arose as a result of the increased input speeds achieved by touch typists. As a result of this rearrangement the keys that were used most frequently were not as easily accessible to the typist. Thus the QWERTY layout effectively reduced the speed at which human users could type, thereby preventing their jamming the mechanism too often.

Other keyboard layouts exist. The most notable one is named Dvorak after its inventor, August Dvorak who was an early Ergonomics researcher at the University of Washington. In 1936 he analyzed the English language to determine which letters were most frequently used. He then rearranged the keyboard layout so that these keys were positioned on the home row, that is, the row under the fingers of a typist in the rest position. By grouping the keys such that those most used were closest to the typist's fingers reaching was minimized and typing speed was increased. Unfortunately the introduction of this alternate layout, despite its efficiency, was not successful. The QWERTY layout had become a defacto standard and no typewriter manufacturer wished to introduce a product that would require its users to have to retrain the manner in which they worked. Thus it is that today the standard computer keyboards still utilize a layout designed for a device first marketed over a century ago.

As an aside it is also informative to take a look at other pieces of the computer human interface that have become dominant, and some newly emerging alternatives. Pointing devices (by this I mean a device for selecting a location on the VDT screen), most notably the mouse, invented in 1968 by Douglas Engelbart, are a major part of the interface today. The mouse was not the first pointing device to be invented. Light pens, which utilize a light sensor and took advantage of the manner in which the VDT refreshes the screen, and track balls which utilize mechanical mechanisms similar to the mouse were also in existence at about the same time. The light pen was a popular early pointing device since it allowed screen positions to be selected directly by simply pointing at them. This facilitated input for mechanical design software such as many of the early CAD/CAM systems. Prime Computer Corp. was among those vendors who provided such an interface in their IMLAC system of the late '70s and early '80s. IBM also had similar systems for mechanical design during this period. The track ball was used similarly but like its relative, the mouse, it did not directly access screen locations. Instead the cursor had to be "dragged" across the screen to the desired location, thus it was slightly more cumbersome to use for mechanical design.

The introduction of the Macintosh Computer in the middle '80s by Apple Computer Corp. launched the beginning of the mouse as a widespread input paradigm. The Macintosh was the first mass produced computer to incorporate a graphical user interface. In the IBM PC models, and Apple's first offerings, the user interfaced with a command line upon which they entered instructions for the computer to perform. These instructions allowed the user to manipulate files in the file storage system and to execute operating system commands or programs. Programs would often perform screen control functions displaying text or images upon the VDT's display. The Macintosh changed this mode of interfacing with the computer by eliminating the need for the user to learn and memorize operating system commands.

Instead of a command based interface, a mouse was provided, along with a graphical user interface, which performed two functions, screen control and program execution. Files in the file storage system, including executable programs, were displayed as images or in the jargon of graphical user interface designers "icons". By positioning the mouse over the icon and clicking one of its buttons the user would, in effect, be instructing the system, via the intervening graphical user interface, to execute the program. The simplicity of this interface paradigm increased the popularity of small computers by eliminating the need for the user to learn the arcana of how to manipulate the computer's operating system.

While Apple was first to the market with a computer based upon this paradigm the idea itself was not unique to them. The idea was pioneered by Xerox in their Palo Alto research labs, and rival companies in the computer industry were also developing comparable interfaces. The most popular of these was Microsoft's Windows product. Another entrant was the X Windows system, developed at MIT under the sponsorship of a number of leading companies in the UNIX based scientific computing market including Digital, HP and Sun Micro Systems. Again, as with the keyboard, the ease of use of the mouse paradigm was paramount, but little or no attention was paid to the ergonomic impact that such a device would have upon its users.

The early light pen systems were experimented with for handwriting input, a paradigm in business use significantly longer than the keyboard or mouse, although not necessarily more well suited for the human anatomy. While digitizing of handwriting was quite feasible, handwriting itself was of no practical use in controlling computers. This is because computers are based upon digital logic which uses numerical codes to represent alphanumeric characters. The most common representation used is known as ASCII. For handwriting to be practical as a computer interface the computer would need to recognize what the user had written and convert that into the ASCII representation. This has however proved to be a complex task. Much work has been done along these lines and the 1990s have seen the release of several systems that are capable of performing handwriting recognition with reasonable accuracy. Among these are the Newton from Apple Computer, a small hand held device that was the first entrant in a class commonly referred to as personal digital assistants or PDA for short. Other systems, which are full functionality computers in the PC class, include the Grid by AST, the Versa by NEC, and IBM's Thinkpad.

The '90s have also seen add on pen interfaces become available for use as pointing devices or as combination handwriting input and pointing devices. Track ball and joystick input devices have also become more widely available although predominately as add on devices or computer game interfaces. Certainly now as the decade begins to draw to a close the recognition of a need for a wider range of pointing devices with which to manipulate the graphical user interface has begun and far more variations of pointing devices are available on the market.

This raises the question of why a proliferation of alternate pointing devices has already gained reasonable momentum while alternatives to the QWERTY keyboard have been slower to gain prominence in the marketplace. The most obvious reason is because of the simplicity involved in pointing devices. As long as a common standard exists by which these alternative pointing devices can interface with the computer the information that these devices send is rather straightforward, a two dimensional coordinate and button press events. Perhaps more importantly is that because of the simplicity of the devices themselves it is quite easy for a human user to quickly adapt to a different pointing device from one to which they are accustomed.

The keyboard however is significantly more complex in that it typically possesses over 100 keys generating approximately 120 different symbols or control characters (not counting capitalized forms). This makes it significantly more difficult to commit to memory knowledge of the device's layout, and thus also more difficult to change between alternative layouts. The QWERTY layout is so predominant that despite the existence of common cabling standards within the PC compatible and Macintosh communities of hardware, respectively, it is only recently that new keyboard designs have become available. Another factor may also lie in the continued proliferation and maturation of the computer industry in that it is becoming more common for individuals to now have a computer dedicated for their exclusive personal use versus computer resources shared among groups within a home or office environment. This would have the effect of allowing individuals to customize the computer interface to better suit themselves without having to worry about the impact upon other users.

The Dvorak layout is among these new designs and is beginning to gain wider acceptance. More novel designs have also emerged utilizing keyboard cabling standards and geared around finger motions but in a significantly different manner. Chord based "keyboards" which rely upon the user pressing combination of keys in a manner similar to playing a musical instrument to generate characters are now marketed. Another design utilizes several switches placed in close proximity so that the user may strike them with small movements of the fingers up, down or sideways.

In addition to these more novel means for manipulating switches to generate characters for the computer more ergonomic designs have been developed for the traditional QWERTY keyboard. These designs focus on reducing the awkward body positioning involved in utilizing a keyboard. Split keyboard designs that eliminate the need for the user to rotate their arms into a palm downward position are beginning to emerge. Variations on this design that eliminate the need for the user to reach forward to access the keyboard (a source of neck and shoulder strain) have also begun to emerge.

Alternate keyboard and pointing devices are however only an incremental modification on existing computer-human interface paradigms. Both of these still rely upon some type of mechanical interaction between the human and the computer be it for handwriting, pointer positioning, or manipulating a mechanical device to generate character and control input into the computer. More recently the increased speed of computing devices has made possible a new means of inputting information into the computer, namely speech recognition. The use of speech as a means of human communication clearly predates the use of formal written language. Additionally speech rates which typically top 120 words per minute exceed the communication rates of either handwriting (20 words per minute) or typing (80 to 100 words per minute for well trained typists), thus making it perhaps the most natural means for humans to communicate with computers.

Speech recognition is just transcending its infancy. Presently several products are available for the home and office computer market. These products such as DragonDictate from Dragon Systems, Voice from Kurzweil, Voice Type from IBM, all for PCs, and Power Secretary from Articulate Systems for the Macintosh all utilize a mode of speech referred to as discrete speech. Discrete speech requires the user to speak with short pauses between their words. While this is not necessarily a natural mode for human speech, input rates of as fast as 90 words per minute with accuracy rates exceeding 95% are reported to be achievable on these systems. Continued increases in computing power and algorithmic efficiency will likely lead to development of products capable of performing continuous speech recognition for most users in the near future.

Speech recognition has been a godsend as a means by which to interface with the computer for individuals whose ability to manipulate mechanical input devices has been limited or eliminated by the development of repetitive strain injuries developed as a result of long term usage of these devices. In its present state speech recognition still comprises a synthetic alternative to the keyboard. This is because most applications written today are written around the keyboard input paradigm. Applications are also sometimes written around the mouse/pointer device as an input paradigm. Thus for speech recognition systems to be compatible with applications written today need to work in such a manner that they can synthesize the functionality of the keyboard and the mouse. Most entry level computers being sold in 1996 not only possess the speed necessary to accommodate speech recognition systems, but they also possess multimedia capabilities. This means that from this point in time forward the capability for speech recognition to become a user interface paradigm tightly coupled with the capabilities of the computer is present. However it will still require a change in the mindset of user interface and application designers before it is possible for speech recognition to become a native input methodology the way the mouse and keyboard presently are.

The future of computer/human interface paradigms is not written in stone. It has seen a constant evolution during the short history of the computer. It will surely continue to evolve. A fundamental change will occur however as it does evolve. For the first time the power of the computer itself will begin to be harnessed aggressively towards making the machine itself adapt to its user, rather than the user adapting to the machine. Touch screen and pen pointing devices, handwriting, and speech recognition interfaces will continue to make inroads into the computing interface. Devices where the keyboard is of little significance or possibly not even physically present will gain greater acceptance. Such devices exist today (this article was written with one) although they are generally more expensive and not sold as mainstream products. As these new interface technologies prove their worth they will become more tightly integrated into computer systems. Application software design paradigms will shift to accommodate these technologies.

In addition standardization of cabling and data transmission mechanisms will provide significant benefit to all users, especially disabled ones. Such standardization (possibly via keyboard cabling standards presently found on PC compatible computers) will allow individuals to possess their own small portable computers for use as personal input devices that they can connect to any computer they might need to utilize. In some respects the common PC cabling standard for keyboards is spawning exactly this kind of behavior. Some individuals now have an input device (for instance customized ergonomic keyboards) for their own particular needs that they carry with them and simply attach to any computer (well most any computer anyway) they need to. The promise is there for existing technologies like laptop computers with handwriting and/or speech recognition systems to perform a comparable function as a personal handwriting/speech recognition engine trained to the individual user, but capable of serving as an interface to any computer the individual would need to access.

By virtue of its overwhelming prevalence the traditional QWERTY keyboard layout has dominated, and continues to dominate, as an input paradigm. It's days as such are numbered however. The continued physical toll in mounting injuries to its users will inevitably highlight the history of this device. The glaring fact that it was never born of any rational design with regard to the anatomy of its intended users, but rather simply evolved from a century of happenstance invention, will eventually cause its role as an interface technology to be reevaluated. Those injured by this device, a group composed predominately of computer literate individuals, will create the core constituency who will lead the charge for its demise. Aided by the increasing power of computing, and entrepreneurs who will harness that power in increasingly ingenious ways, new interface technologies will evolve. We must only hope that the type of market forces that suppressed acceptance of the Dvorak keyboard layout for 60 years will not also doom these new technologies. If so it would surely prove the keyboard a tyrant.