The History of the Keyboard as a User Interface

The Keyboard as Interface

The computer keyboard is an interface on several levels, though the basic concept as initiated with the typewriter is easier to pigeonhole. It is a system for transforming literate thoughts into standard representations of letters: handwriting without the interpersonal variance and potential information loss. The typewriter started out as a system to produce legible, professional-quality one-off output by hand (as opposed to the typeset printing press). In this sense it was interface between human and paper, and also a middleman in the recording of thoughts, symbols, and characters. Additionally its typewritten output was further an interface between people--one's words represented unambiguously to another party minutes, days, or years later.

As computers have grown in popularity, the interface has become more complex. Keyboards are now an interface between analog human thinking and digital computer operation, storage, and transmission. It's still a typewriter, but it outputs for the world instead of one person, and the audience has grown to include machines.

We'll deal with the concept of the "typical" keyboard for most of the work here, and that typical keyboard is an interface for dextrous, literate people. People without hands, dolphins, and people who cannot distinguish symbols are not helped by this interface. Individual keys are symbolic; by default they represent letters and characters. But even in yesteryear, the development of the SHIFT and CARRIAGE RETURN keys demonstrated that the character set need not be static. As technology advanced, keys became even more customizable; now we can and do have keys to print the screen, change the brightness, launch programs, and more. So now we can say that the interface has evolved to represent a symbolic transmission of intent--this interface is a way for literate humans to translate an intention into a command that a machine carries out, whether that intent be "B" or "shut down this operating system in an orderly fashion, closing all applications, and finally power it off." And the interface is still evolving here too: For example the optimus keyboard brings the _Diamond_Age_ style future to today: no static text required!

While the keyboard is a versatile human to computer interface, it is not the most versatile that exists. Mice, touchscreens, computer vision, speech recognition all contribute to the "sensorium" of a computer, and each have their benefits and drawbacks, and more and less relevant environments and audiences. Suffice it to say that the keyboard is a basic, multifunctional, and simple interface. Plug it into a computer capable of interpreting the signals sent by a keyboard and you have a programmable interface with shades of sense. For instance, one's typematic habits can be used fairly accurately as a form of biometric identification.

A key feature of the keyboard as interface is that its output is standard. Unlike handwriting, speech, or visual recognition, output has been unambiguous since the invention of the typewriter; it is clear, legible, and unassailably precise. In the computer era, as more and more of our whims are interpreted by machines driven completely "virtually", this lack of ambiguity is key to the keyboard's continued success and use as an interface. At this date, speech recognition is not suitable for mission critical applications, nor is handwriting recognition. Communicate by variable duration, intensity, and differently timed keystrokes, but a computer understands each and every letter transmitted and interprets it as expected. Software may be another matter, but the typed symbol reaches the machine unaltered in all but the most pathological cases of hardware malfunction.

Another feature of the interface is usability. Speed, accuracy, correctness, and effort are all improved over handwriting. The pressed keys are easier than pressing a pen to a page; less arm and wrist motion is required, and typing is far less taxing than writing, as anyone's who's taken a timed writing test will tell you despite typing for hours on end without respite.

As an interface, the keyboard is barely bidirectional; the only feedback it gives is optional. The early typewriter demanded firm pressure and created audible, tactile indication not only of mechanism engagement but also of the physical thump of keystroke on paper. The modern keyboard can provide the same mechanical simulation of threshold effort to engage, spatial travel and an audible "keyclick." Some keyboards even have speakers in them and handle the keyclick entirely without interference from the computer, while others are silent but the computer typically features availability of an audible feedback mechanism. Regardless of mechanism or lack thereof, nearly all modern keyboards provide as much feedback as they ever will instantaneously, in terms of keypress & spring-back.

In the end, the keyboard's interface has not itself changed substantially since introduction, except in cosmetic details, but the interpretation and importance of its output and its configurability have grown greatly alongside the march of technology.

Assessing the Keyboard as Interface

The next question is whether the keyboard is a successful interface, and for metrics by which to judge it we turn to the following design experts: Tognazzini, Nielsen, and Norman. In the Tognazzini (hereafter referred to as Tog) article, First Principles of Interaction Design, Tog lays out a number of behaviors of successful, effective interfaces. Not all are relevant, but many are.

Does an interface anticipate the user's needs and make all functions available instantly? A keyboard is WYSIWYG--symbols laid out for anyone to use without a deeper understanding. Is an interface accessible to people with disabilites? Tog only addresses color-blindness, but the beauty of the keyboard is that aside from complete color-agnosticism, it is also easy to make blind-accessible versions, and finally even people with many forms of limited motor skills can use keyboards with little modification.

Does an interface offer sufficient customization for the user to become comfortable? Do the defaults make sense and are they changeable in case they don't suit a need?

Keyboards are both trivial and impossible to change, making them strange dichotomous gray sheep in Tog's world. From the point of view of the user, the keyboard has physical symbolic manifestations that are indelible, or at best, require substantial, mechanical, individual effort to change.
But wait! The future of keyboards is poised to address this with the optimus LED-keycapped keyboard..
However we must also consider the computer-as-user. The keyboard is a two way interface; the computer's interpretation of the input is as important as the user's intended input as far as providing "correct" functionality. Additionally, from the computer's point of view, a total reconfiguration of inputs (aside from certain mechanical limitations regarding shift and alt keys on standard keyboards) is trivial and can be instantaneous. This means that for some (touch-typing) users, changing keymaps (i.e. DVORAK->QWERTY) or languages via software fits Tog's recommendation perfectly; not only can defaults be changed easily, they can also be changed arbitrarily.

As regards intelligent, responsive input ability, anyone with reasonable manual dexterity can rapidly eclipse the next best method of manual text input (handwriting) with minimal instruction. Once one can read, keyboards are the next best thing to intuitive, aside from some baroque concepts like carriage return and shift (arguably, nearly all necessity for capitalization could be programmed since it follows fairly precise rules).

Cursor keys and function keys are a little tougher, but using a computer as more than a glorified word processor with current state-of-the-art software is not particularly intuitive overall. Here we must make mention of the "standard" US/latin 101-key keymap known as QWERTY. While this keymap was originally designed for mechanical and non-practical considerations, it has not been proven to be vastly inferior to the competing, logical, more intelligent DVORAK standard. Nobody argues against DVORAK being a better layout, and indeed {ref site that measures efficiency} it is provably more efficient for English. However, the lack of efficiency in QWERTY is not much of a detriment in practice especially when one tries to optimize across languages and various uses. 

   '{' is very rare in English, but very common in [the programming
   language] C++" [Poika]
Finally, the electrical impulse is near enough to instantaneous that we have to pass the buck on the judgment of "response" to the software. It is easy to recall situations in which response is near enough to instantaneous as well (we see our characters echoed on the screen faster than we can measure), but one also need not go far to discover a situation in which keypresses take ages to echo, and due to expectations from the aforementioned good interactions this is nearly intolerable. Thus, even though the blame may lie with the software or host computer, the interface can only receive a failing grade.

User efficiency is the next topic in Tog's world. Context switching? No thanks! After a brief learning curve to adapt to the layout of the keys, a user can type an entire novel without having to stop on account of the keyboard. Once touch typing has been achieved (sometime during that novel-writing session), the user is even free to look elsewhere... at the risk of distracting themselves. Does the user stay occupied? The keyboard itself requires no waiting for operations to complete, though the software or hardware on the other end may, but that is beyond the control of the keyboard. Finally, is the keyboard a sort of lowest common denominator, maximally efficient for the maximal number of people? With changeable keymaps, even foreigners can quickly get work done on a non-native keyboard, and the standard shape and layout assists with instaneous skill transfer from keyboard to keyboard.

Fitts' law is a model of human movement, predicting the time required to rapidly move from a starting position to a final target area. How work-efficient is a keyboard? This is one area in which opinions are split, for several reasons. The already-mentioned keymap issue is one factor to consider, but there are others: the space bar, the enter key, and the backspace keys are the largest on a keyboard--they are also probably the most used. Some of the lesser used keys, DVORAK and QWERTY can agree on: ESCAPE stays tucked away in the upper corner and punctuation is off to the side. An obvious downside to the keyboard as efficient, ergonomic input is the high incidence of repetitive stress injury among keyboard users. The interface is not perfectly suited for every human body. But rarely has an interface seen such widespread protracted use among all segments of a society, so perhaps a similar incidence would prevail as well if we were all tailors or bakers.

Latency is minimal on a keyboard, but software can get in the way of this... the keyboard's work is done in milliseconds, but the software sometimes takes seconds or minutes to pass the visual feedback back to the user, or worse yet, drops input it is unable to handle. Nielsen also addresses feedback; namely that .1 second is invisible and 1 second is tolerable [Nielsen]. The keyboard comes in far under .1 second response time mechanically and electronically. Learnability, or speed of skill acquisition is also good: anyone can start typing without education, and learned speed is actually quite good, especially in comparison to other character input methods. Tactile sensation is a mixed bag. it's highly implementation dependent..but keyboards are cheap as tools amortized over time, so it's worth it to achieve perfect success.

Tog also suggests avoiding invisible navigation, which the keyboard itself embraces, but again software fails on follow through. The interface is totally transparent as far as letters, but function keys can be a bit opaque However popular "keyboard shortcuts" such as control-c to copy and control-v to paste are never repesented on the keyboard--so the interface as used by an experienced user would be opaque to the novice user. Overall a grade of neutral on visible navigation.

Developments in Keyboard Usability: Past and Future

Here we present a timeline of devices from past to future--realization of the evolution of interface.

While clearly the printing press does not resemble any notion of a keyboard as we think of it today, it descends directly from this machine. The ability to put your thoughts and ideas into perfect print and pass them on to someone else was something sought after hundreds of years ago. Setting up and using the press was much slower than actually handwriting something, but it was the finished product that exceeded the quality of handwritten works. This particular press is from 1811, though the press had been around for many years prior.

The machine we could begin to classify as a keyboard was this index type writer. Produced in 1829 and called the typographer, it required the user to twist the dial to the desired letter and press the button onto the paper, then inch the cursor down a character and repeat. You could not see what you were writing as you wrote it--it was obscured by the mechanism until you fed the paper out. Furthermore, this process was took much longer than writing your thoughts by hand. However, the device was portable and allowed for more spontaneous 'one-off' creations than the printing press.

In 1870, Ramus Hansen developed his brass typing instrument, dubbed the Hansen Writing Ball. His is one the first innovations that we still see in keyboard interfaces today. The ball features multiple buttons, one for each letter or symbol in a given alphabet, all marked according to their function. This concept forms a basis for keyboards as we know them today.

In 1874, the Remington firearm company created their Remington typewriter, and set a standard for keyboard functions and layouts that has triumphed as the standard we use today. This keyboard features the familiar QWERTY key arrangement. However, it only typed in capital letters; no shift key was present. Also missing were the number 1 and 0 keys. For these early machines, the only thing that mattered was the output. The output was what you read on a sheet of paper. So in place of 1, a lowercase L could be used, l. In place of a 0, the uppercase O could be used, O. It made no difference that the key hit was not actually the symbol requested, it only appeared to be on the printed output. However, you still could not see what you were typing as you typed it.

After sub-par sales of their first endeavor, in 1878 Remington followed up the Remington 1 with the Remington 2. This keyboard introduced upper and lower case letters, and provided a toggle between the two with the SHIFT key, a button that we still use today. What's interesting to note about these machines is that there was no visual indication to anything that you typed as you typed it. Only after completing a word or phrase could you inspect your work. To use this keyboard, you would enter your commands through the key interface, then open the machine to view what the results of your work.

This typewriter, the Underwood No. 5, produced in 1900, was one of the first to allow the operator to actually see interactively the results of a key press. On this keyboard, you could read what you wrote as you typed it. This is a clear indicator of things to come. It still followed the now conventional QWERTY standards, though there were still no 1 or 0 keys.

From 1900 until the 1950s, much work had been done on a system to augment the network of telegraph lines that blanketed the country. The teletype was a typewriter that converted the key presses of letters and symbols in to electrical impulses that could be sent over a telegraph wire, received by an identical machine, and converted to type on a paper. This was a giant step towards keyboards becoming interfaces to computers as for the first time key presses were being interpreted as commands to a machine, not just a mechanical indication to perform an action. With the output of a teletype machine no longer being the paper sitting directly in front of the user, 1 and 0 keys made their appearance. Making the distinction between a 1, L, 0, and O over a telegraph signal was important.

The IBM Selectric (1961) was one of the first typewriters to use an electrical system to help augment the intended function. However, the Selectric still saw key presses as analog indications, not as discrete commands as teletypes did. Contrary to what was expected, pressing a key on the Selectric did not activate a set of switches and relays that told the typewriter to put a particular letter on the paper. Pressing a key simply routed the mechanical energy of the internal motor to the printing element so the desired letter could be put on the paper.

What made the IBM Selectric so popular was its ruggedness, speed, and inability to jam from key presses made too quickly. This left it well-suited for the rising need for mainframe terminals, machines needed to interface with the large, sophisticated computers of the 1960's. While this wasn't the first, this Selectric based terminal from 1965 was one of the first to directly tie the key presses on the keyboard to digital impulses of ones and zeros, something the computer could understand natively. No longer was the F key directly representative of producing an F on an output, the F key now represented a command that the computer could interpret.

As computers became more sophisticated, graphical terminals were introduced and the need for having a printer directly tied in with your input device was no longer needed. IBM then created the keyboard as we know it today, a panel of buttons each labeled with a specific function. Unlike others before it, this keyboard was useless with out a computer attached, it's only output was digital commands. This particular model from 1981 was featured on a computer we would recognize as a PC and begins to show some characteristics directly related to the keyboard of today. There is a number pad, as well as a set of function buttons, however, the position is on the side, instead of the expected location above the characters.

Finally, we arrive in the 1990s with a keyboard that looks familiar to most computer users today. Still present are many lingering vestiges of the typewriting era: Shift, Return, Caps Lock. Also present are buttons that no longer serve much purpose, but are remnants of a text based computing era: Pause/Break, Control, Escape, System Request.

In 1995, people facing pain, discomfort, and even injury warranted a new type of keyboard. As computers became more central to the life and livelihood of more and more people, a keyboard that was easier on the user's body became important. Interesting to note is that with the introduction of this keyboard comes the automatic assumption that the user already knows how to touch-type in a specific, standardized manner. If the user is unfamiliar with the intended hand position that this keyboard assumes, the interface will do nothing to alleviate pain and suffering, but will be likely to add frustration.

Future directions for keyboarding

Announced in 2005 and slated for a 2007 release, the Optiums keyboard is the ultimate in abstraction from key to command. With Optimus, our keyboard is no longer a keyboard, but more like a panel of carefully positioned buttons that can issue commands to a computer. This is completely contextual, as pressing the same key in a number of programs could offer very different results. The Optimus will be the first to allow a visual representation of what action will be taken when a particular button is pushed, regardless of the computational context.

While there are some blue-sky systems under consideration, none have achieved significant traction as marketable or desireable to the average user. So, current "keyboards of the future" are mostly limited to ergonomic improvement rather than interface improvement: most still involve typing on a recognizably keyboard-like device on a desktop. Some examples:

The virtual laser keyboard is actually a device the size of a cigarette lighter but the keyboard is full-size and projected on any flat surface.

 
Next comes a bevy of ergonomic "split" keyboards--we've all seen the Microsoft version but here are some novel, differently ergonomic takes. The Kinesis keyboard allows independent section positioning, while the vertical safetype isolates wrist torsion. the datahand is more futuristic and features more natural hand positioning and minimized finger movement, but still uses the hands in front of a person, with fingers exerting force on keys.

One-handed keyboards require only... one hand!

 
Some have novel interfaces, but still involve use of all fingers and occupy entire hands. What makes these "keyboards of the future" instead of Keyboards of the Future! is that they have been available for years and adoption is slow and mostly technophiles--very little consumption or even awareness by the general public (pyramid and twiddler2 pictured above).
Keyboards are a fundamentally textual media and in that sense are lacking as interfaces to a digital world. While they are functional interfaces to today's relatively slow processors, they are constraining our workflow to serial and low-level interfaces. One need only imagine an artificially intelligent program sitting idle for 99% of its "run" waiting on manual instead of mental input to see future directions offering orders of magnitude improvements in efficiency and throughput. Thought input could dictate paragraphs at once, or better yet, do away with verbal thinking altogether and accept *idea* input. Finally, they are an inefficient use of the hands; speech recognition will one day be functional enough to leave hands free for other tasks or input devices (but watch out for eyeball-following or straight-up brain-controlled mouse pointers and/or spatial body sensors for 3d cursor movement! Keanu Reeves as Johnny Mnemonic, eat your heart out!)

References

"IBM Personal Computer XT." Wikipedia, The Free Encyclopedia. 24 Aug 2006, 06:23 UTC. Wikimedia Foundation, Inc. 18 Oct 2006
http://en.wikipedia.org/w/index.php?title=IBM_Personal_Computer_XT&oldid=71545252

"IBM PC keyboard." Wikipedia, The Free Encyclopedia. 17 Oct 2006, 19:12 UTC. Wikimedia Foundation, Inc. 18 Oct 2006
http://en.wikipedia.org/w/index.php?title=IBM_PC_keyboard&oldid=82049029

"Alphanumeric keyboard." Wikipedia, The Free Encyclopedia. 12 Oct 2006, 15:37 UTC. Wikimedia Foundation, Inc. 18 Oct 2006
http://en.wikipedia.org/w/index.php?title=Alphanumeric_keyboard&oldid=81025083

"Typewriter." Wikipedia, The Free Encyclopedia. 16 Oct 2006, 20:12 UTC. Wikimedia Foundation, Inc. 18 Oct 2006
http://en.wikipedia.org/w/index.php?title=Typewriter&oldid=81849239

"Hansen Writing Ball." Wikipedia, The Free Encyclopedia. 10 Oct 2006, 13:47 UTC. Wikimedia Foundation, Inc. 18 Oct 2006
http://en.wikipedia.org/w/index.php?title=Hansen_Writing_Ball&oldid=80613756

"Optimus keyboard." Wikipedia, The Free Encyclopedia. 17 Oct 2006, 02:09 UTC. Wikimedia Foundation, Inc. 18 Oct 2006
http://en.wikipedia.org/w/index.php?title=Optimus_keyboard&oldid=81916064

"Microsoft Natural keyboard." Wikipedia, The Free Encyclopedia. 9 Oct 2006, 19:47 UTC. Wikimedia Foundation, Inc. 18 Oct 2006
http://en.wikipedia.org/w/index.php?title=Microsoft_Natural_keyboard&oldid=80469360

"The IBM 2741 Terminal." 25 Feb. 2005. Columbia University. 18 Oct. 2006
http://www.columbia.edu/acis/history/2741.html

"IBM 2741." Wikipedia, The Free Encyclopedia. 12 Oct 2006, 13:46 UTC. Wikimedia Foundation, Inc. 18 Oct 2006
http://en.wikipedia.org/w/index.php?title=IBM_2741&oldid=81008018

"The Timeline." The Typewriter Museum Collection. 18 Oct. 2006
http://www.typewritermuseum.org/collection/timeline/index.html.

"Printing Press." Wikipedia, The Free Encyclopedia. 4 Nov 2005, 20:15 UTC. Wikimedia Foundation, Inc. 18 Oct 2006
http://en.wikipedia.org/w/index.php?title=Printing_Press&oldid=27380975

Hersch, Jay. "The Tyranny of the Keyboard." Typing Injury FAQ. (8 Jan. 1998) 18 Oct. 2006
http://www.tifaq.com/articles/keyboard_tyranny-feb98-jay_hersh.html

"History of the IBM Typewriter." Etypewriters.Com. 25 Mar. 2003. 18 Oct. 2006
http://www.etypewriters.com/history.htm

"Optimus Keyboard." Art Lebedev Studio. 14 July 2005. 18 Oct. 2006
http://www.artlebedev.com/everything/optimus/

"IBM Model F Keyboard." Clicky Keyboards. 18 Oct. 2006
http://www.clickykeyboards.com/index.cfm/fa/items.main/parentcat/11066/subcatid/0/id/171727

"Teleprinter." Wikipedia, The Free Encyclopedia. 13 Oct 2006, 14:04 UTC. Wikimedia Foundation, Inc. 18 Oct 2006
http://en.wikipedia.org/w/index.php?title=Teleprinter&oldid=81210519

Barber, Christopher. Beyond the Desktop
Academic Press, 24-28 Oval Road, London NW1 7DX, UK, 1997.

David, Paul A. Clio and the Economics of QWERTY
The American Economic Review, Vol. 75, No. 2, Papers and Proceedings of the Ninety-Seventh Annual Meeting of the American Economic Association. (May, 1985), pp. 332-337.

Isokoski, Poika. A Minimal Device-Independent Text Input Method
http://www.cs.uta.fi/~poika/g/

Nelson, R. A. "History of Teletypewriter Development." RTTY. Oct. 1963. Teletype Corporation. 18 Oct. 2006
http://www.rtty.com/history/nelson.htm.

Nielsen, Jakob. Usability Engineering
San Diego: Academic Press, 1993

Norman, Donald A. The Design of EveryDay Things
New York: Doubleday: 1988