Thursday, September 25, 2008

The Keyboard---Expert Interface: An Historical Perspective Essay

Few computer components are more recognizable than the keyboard. Though special-purpose setups such as living room media centers are beginning to chip away at was was once nearly perfect ubiquity among computer peripherals, the keyboard remains an indispensable device for nearly every computer user. Considering the vast range of consumer expertise and use durations, this input device is a strikingly odd champion of instruments. The layout originates as an intentional impediment to user efficacy and anyone learning how to use it can attest to the hundreds of hours needed to reach moderate proficiency. This essay traces the origins of this curious state of affairs and examines the side-effects created along the way.

The roots of the modern keyboard lie in development of the typewriter. Though it was a product of numerous inventors' refinement and experimentation, one model managed to be the first commercial success---the 1867 Type-Writer from Christopher Sholes, Carlos Glidden and Samuel W. Soule[1]. The arrangement of letters from the subsequent commercial product closely resembles the modern "QWERTY" key order [1]; this arises from the well-known detail that when the keys were placed in alphabetical order, the letter bars would collide and jam up the machine. Less well-publicized, though, is that a contributing reason to the new order was that early typists used two-finger hunt-and-peck typing style and the new order delayed location of each subsequent key, which allowed more time for the bars to clear each other[2].

The success of the original "typewriter" eventually compelled competitors to adopt the same idiosyncratic key arrangement which deprived consumers of a choice about their key layout[1]. Once the layout gained enough inertia, no other layout could afford the loss of sales to the established base, even if it were demonstrably superior, as was the case with the Dvorak layout.

Eventually, typists discovered the technique touch-typing, which was publicized by early speed-typing contests among touch-typists [2]. This did provide a large speed boost from the twenty words or less per minute that might be a reasonable expectation of a hunt-and-peck typist [2]. However, learning to touch-type on a QWERTY keyboard is a process that commonly runs hundreds of hours (any new typist will readily attest to the protracted education period). The advance provided by touch-typing is merely to allow an alternative to the poor usability base-case; the alternative, however, demands development of a very specific skill over a long period---it does nothing to improve the situation for users who do not wish to invest the considerable time and effort demanded by learning to touch-type. This marks the beginning of a schism in users of QWERTY key arrangements. Those who develop an expertise in the form of touch-typing are able to utilize the device vastly more effectively than those who cannot or do not follow that path. The non-expert (i.e. hunt-and-peck) typists, also reveal their comparatively low skill level immediately when typing, since the movement pattern used is so distinct from that employed by touch typists. This very clear skill hierarchy produces social consequences that are undoubtedly experienced by all present.

As long as QWERTY keyboards were not common, the effects above were easily dismissable---typing might be a useful skill (like double-entry bookkeeping), but advanced skill therein was only demanded in specific, largely administrative, professions. The advent of computers would change all this. Computers often need text input, so designers looked to the fastest easily adapted text entry tool to fill this need.

Typewriters fit the bill [3].

Considering the experts-only domain that largely characterized early computer work both in academia and with hobbyists, an expertise-necessitating input device like a typewriter made sense.

As computers became gradually more widely-used, perhaps there was never a large enough influx of new, unskilled users to warrant the development of a less expert-oriented text input device. Perhaps the difference in learning effort never overcame the increased cost of designing a new test input system. Perhaps adopters mostly switched immediately into use levels that justified a traditional keyboard. Whatever the reason, there were no serious efforts to obsolete the now century-old design.

Then the mouse appeared. First at the Stafford Research Center, but soon shipping with Apple Computer's hot new line of PCs, the mouse was an enormous break with interface tradition [4]. Suddenly, instead of slow-to-master keyboard interactions, the mouse offered a less efficient---but far easier to learn and understand---means for performing computing tasks. As evidenced by the considerable success of the MacIntosh line that sported the novel interface[5], there had indeed been a market for a new input device. A keyboard still tagged along with every mouse, regardless of the popularity of the latter. It might have been supplanted as the favorite interaction channel, but the keyboard still ruled text input, not to mention its utility for expert interactions.

This dichotomy of interactions again cleaved the user groups, one part regarding the graphical interface as a toy [6], the other believing that the command-line amounted to an archaic concession to bygone computers' limitations[6]. Neal Stephenson holds that the precision and complexity afforded by the command-line (a fundamentally keyboard-oriented interface) is both its greatest strength and its greatest weakness [6]. By limiting the user with simple graphical metaphors and a very simple control (the mouse), they need muster less intellectual effort to perform a task; conversely, complex or very precise desires may not be possible to express in such a framework [6]. An illustration of these differences would be a Unix server, which rarely would have a mouse since no task thereon would not be fastest, safest, and easiest expressed by strings of text. On the other end of the spectrum lies software such as MacOS, which, were it not for text-entry needs, would quite comfortably disregard the keyboard.

This, therefore, is the most significant effect of the design of the keyboard: keyed input (including typed commands) is largely targeted at experts and power users, mouse/pointer input is target at everyone else and often garners primary attention during the design process. This tradition seems to have indoctrinated users with the idea that hot-keys and shortcuts are inherently harder even if the user is comfortable entering text on a keyboard and that mouse-driven interactions are easier to learn and remember.

Of course, this author vehemently disagrees with the idea the keyboard control is any less easy to learn than mouse control once the user has learned touch-typing. One example of this idea is an experimental interface design discussed at the New Interface Advocate which leverages the unique advantages offered by keyboards.

  1. "Typewriter", retrieved September 21, 2008
  2. "Early Typewriter History", retrieved September 21, 2008
  3. "Teleprinter", retrieved September 21, 2008
  4. "Mouse (computing)", retrieved September 22, 2008
  5. "Macintosh", retrieved September 22, 2008
  6. "In the Beginning Was the Command Line" Neal Stephenson, 1999