American Scientist

Many answers have been proposed to the riddle that the Mad Hatter posed to Alice at their famous tea party (for example, because Poe wrote on both). Let me argue for yet another one: the power of convention.

As I compose this essay, the writing desk I am using is a spanking new Macintosh laptop with many gigabytes of storage and enormous computing power. Its screen is a marvel of full-color clarity; it has a built-in video camera and microphone, and hence allows multiple methods of input and output. Yet it has a QWERTY keyboard. Why QWERTY?

The QWERTY keyboard, named for the order of the keys on the left side of the first row of letter keys, was invented in 1873 by Christopher Sholes, a Milwaukee newspaper editor. Its purpose was to split up keys that were commonly hit serially so that a too-fast typist would not jam the associated type bars. In addition to its primary goal of slowing things down, it also aided left-handed English language typists, for far more words can be typed with only the keys under the left hand than under the right.

Now, since its purpose has been long anachronistic, why do we still persist in using it? The reason is, of course, the power of convention. After it became the conventional keyboard layout and touch typists learned it, they were loath to give it up and learn a different system, even if the newer system was demonstrably superior. And so now, almost 150 years later, QWERTY has survived; and, because virtually all subsequent generations learn to type using it, the likelihood of its being improved remains small.

Pie in the Trash

In his 1801 Statistical Breviary, William Playfair, the Scot who invented many forms of statistical graphics, proposed the pie chart. Playfair's pie had but three segments and showed what proportions of the Turkish Empire were in Europe, Asia and Africa. It worked very well. Indeed, to test its efficacy, one can easily demonstrate that a small child can tell that 1/3 is larger than 1/4 from a pie chart far more easily than from the fractions themselves. But since Playfair's time, pie charts have become conventional and thus are often used to display much more complex information, this despite strong evidence that a pie chart's efficacy in such a situation is suspect.

In 1990, the New York Times used a pie very much like the one shown in the first figure to communicate the content of what New Yorkers typically discard. The content is a remarkably apt metaphor for the quality of the plot.

We can tell how much of each component is discarded, but only by reading the amount from the label. Because we must read the graph rather than see it, what value is added by the pictorial representation over the numerical? Concerns such as these led the statistician William Cleveland, in his 1985 book The Elements of Graphing Data, to propose the dot chart as an alternative. He found, through a series of experiments, that humans could visually judge lengths far more accurately than areas or angles. So by transforming the pie segments to line segments punctuated with a large dot, he was able to produce a plot that has, in this instance, substantially better perceptual characteristics than Playfair's pie (second figure). Nevertheless, in the intervening 23 years the popularity of the overmatched pie has not decreased, and, sadly, the use of Cleveland's excellent proposal has seen no substantial increase.

Once again, the reason is convention.

It is natural to ask how long it takes for a genuinely superior product to supplant one that is well established. For the pie chart 20 years is not enough; for the QWERTY keyboard it is 150 years and still counting.

Unhooked on Phonetics

One would think that the speed with which convention is overturned depends strongly on how much of an improvement the unconventional technique provides—and how much evidence is available to support the replacement. Would that it were so.

A powerful final example further demonstrates the obstinacy of convention and provides the connection with the raven in the title.

pronounced Ka-ma-gui, is the Korean word for raven, written in the remarkable Hangul alphabet. To understand the relevance of this example, we must go back almost six centuries. China's culture dominated Asia for centuries, so it isn't surprising that the written Korean language of the 15th century used Chinese characters. Yet because the Korean language uses inflections and suffixes to add or modify meaning, whereas Chinese sentences are qualified with particles, the use of Chinese characters was far from an ideal match. In addition, Chinese characters, known as hanja, were complex, numerous and so difficult to write that literacy was reserved for aristocrats.

Sejong the Great (1397–1450), the fourth king of the Joseon Dynasty (1393-1910) apparently deplored the fact that common people, ignorant of Chinese characters, were practically forced into illiteracy. He felt that this deficiency had important practical consequences, for they had no way of submitting grievances to the authorities for possible redress. Nor could they record their thoughts or experiences for posterity, placing obvious severe limits on the breadth of Korean science and art.

To ameliorate this situation, King Sejong set about creating an alphabet especially suited for the Korean language, and his success provided a model for others. It is far beyond the goals of this essay to explain the details of the Hangul alphabet. Instead let me choose a few of its distinguishing features.

First, it is completely phonetic. Thus, if you can already speak and understand the Korean language, all you need to do to be able to read and write is to memorize the symbols representing the 10 basic vowel sounds and the 14 basic consonant sounds. Apart from a few minor exceptions, the phonetic value of each symbol is invariant. Thus, any letter string, even if unfamiliar or nonsensical, can be sounded out instantly and accurately. There is never a need to consult a dictionary for sounds or spellings. Contrast this with how the sound 'f' is represented in English—fat, photo, laugh; how the same vowel, say 'a,' can have many different sounds—fat, farm, face, fall, hurrah; and some letters in English have no sound at all—psalm, indict.

Each Korean character usually consists of two or three components: (1) a consonant plus (2) a vowel, or (1) a consonant plus (2) a vowel plus (3) a consonant. Thus, each Hangul character is of one syllable structured as:

The accompanying table lists the Hangul letters and their approximate pronunciation in English. To see how easy it is to spell and read in Hangul, consider the two-syllable word "Hangul" which is written

Use the table to sound out the written word.

Second, great effort was expended to aid memory. Each consonant sound is constructed to look like your mouth when you say it! In English, only one letter has this characteristic ('o'). In Hangul all consonants have this characteristic. Sometimes the letter looks like the shape of the mouth viewed from the front (for example, the letter 'mium', which is an "m" sound pronounced with the lips spaced a little apart). Often the letter is a stylized drawing of the mouth when looked at from the side after a vertical cut down the center of the head, where the location of the lips and tongue indicate the letter. For example, the letter 'nium,' an 'n' sound, depicts the outline of the tongue touching the upper palate.

And pronouncing the letter 'shiot' requires that the tip of the tongue and the upper teeth be brought close together; it is created by blowing through that narrowed passage.

In addition, the very names of the letters provide information on their pronunciation. So, for example, the letter 'rial' is pronounced as an 'r' at the beginning of a syllable and as an 'l' when it appears at the end of a syllable. Compare with mium, which has an 'm' sound at both ends of a word.

It was generally agreed that a clever person could learn the entire alphabet in a morning and so go from illiterate to literate by lunchtime. Contrast this task with the 20th century Chinese requirement that children must learn at least 2,000 hanja characters by the end of high school to be considered sufficiently literate.

It would seem that this wonderful innovation would be adopted rapidly, but alas the power of convention prevailed. Confucian scholars of the 15th century believed that hanja was the only legitimate writing system. Choi Manli, a senior scholar of the time, presented a petition to the king criticizing the new alphabet:

Since the new alphabet is so easily understood, I fear that the people will fall into laziness and never make efforts to learn. Those who do not use Chinese characters but other letters and alphabets, such as Mongols, Sohans, Jurchens, Japanese and Tibetans, are all barbarians without exception. To use new letters would make us barbarians ourselves.

He continued,

Why does Your Highness seek to alter a language that has been used since early antiquity and has no ill effects, and place alongside it a set of coarse and vulgar characters of no worth at all? Is not this script, moreover, a mere transcription of the words spoken by the peasants, without the slightest resemblance to the original Chinese Characters?

Despite opposition of this sort, Hangul's use as an official Korean alphabet grew in fits and starts, although it took fully 500 years before it became widely accepted as the Korean alphabet.

Attending Convention

I hope that I have demonstrated that it is not unreasonable to believe that the inertia of convention acts as a drag on innovation. Yet simply recognizing this obstacle will not make the power of convention disappear, nor should it. When faced with a single communication task (for example, reporting the results of an experiment in a research paper), we should first consider the conventional way to do it. If such an approach allows it to be done adequately, then that is a sensible path to follow; fighting convention can age you quickly. If, on the other hand, the task is but the first of many (for example, writing the first of many national statistics reports, with many similar data sets—see Pickle et al.), the cost of fighting convention may be amortized over the long run.

What have we learned from these examples? It seems to me that QWERTY has little to recommend it, and various alternatives have already been proposed. August Dvorak, a professor of education at the University of Washington, patented one in 1936. The 1985 edition of the Guinness Book of World Records tells us that Barbara Blackburn, the fastest typist on record (150 words per minute for 50 minutes, with peaks of 212 words per minute), set her record on a Dvorak keyboard. Obviously improvements are possible and some experiments in design should allow us to build on Dvorak's design. Implementing such an innovation is easy, since programmable keyboards can allow those whose touch-typing skills require QWERTY to keep it, but as we start teaching the new keyboard in schools, graduates who can perform better will transform the old keyboard to the new. One generation should just about do it.

Pie charts have little to recommend them over Cleveland's dot charts; as soon as those who write graphic software replace pies with dots as the default option, we will see the end of this over-used invention.

The 15th century complaints from scholars about the ease of learning Hangul are surely relevant today. Yet that should not dissuade the Chinese and Japanese governments from considering the costs of keeping their ancient writing systems. Perhaps a Hangul-like Chinese keyboard might have more than 34 keys (24 letters plus 10 numbers), but it would surely be far less cumbersome than current methods. If such an alphabet were invented and implemented, the two systems would probably operate in parallel for a while, for old ideas never really die, just the people who believe in them. But eventually offspring of King Sejong's marvelous Hangul alphabet would take over. Other languages have alternative phonetic alphabets (for example, the Unifon alphabet is a way of writing English based on the principle of one letter per phoneme; it was created by John Malone in the 1930s), but they have yet to overcome the power of convention. Still, the cost of maintaining the convention seems much smaller for English than Chinese.

And so, why is a raven like a writing desk? My answer is because both of them can be improved substantially, but the major part of the task is overcoming the inertia of convention, not inventing a superior product.

Acknowledgments

My gratitude to the National Board of Medical Examiners for support of this research, despite its only very marginal relation to NBME's mission. I appreciate the breadth of their view. I also wish to thank Kyung T. Han for his invaluable help in providing expertise in Hangul so that I have some faith that my description of it is reasonably accurate. My thanks also to Paul Velleman, who showed me how to find both Dvorak and Hangul on my Mac, and to David Hoaglin, whose sharp eyes helped me to say more nearly what I meant.