Version classiqueVersion mobile

B C, Before Computers

Stephen Robertson

1. In the beginning..

Texte intégral

1Ever since the dawn of recorded history, and before, we have been trying to learn how to do things with information.

2This is not at all the grand claim it may seem to be. Rather, it is a tautology. We could not begin recorded history until we had ways and means of recording—and recording information is one of the things we have been learning how to do. This is perhaps one of the few necessities of recorded history. We didn’t have to come down from the trees, or even out of the ocean, before beginning our recorded history—though in fact we did both of those things. We didn’t have to learn how to plant crops instead of relying on hunting and gathering; we didn’t have to build towns, invent trade, organise markets and establish trade routes—though probably we did all of those things, and probably they all helped to stimulate the invention of writing. We certainly didn’t have to invent the wheel, and indeed it’s not clear whether we invented the wheel before or after we learnt how to write. But we did have to learn how to write.

3The written message is a specific human invention, just as much as the means to make that message. Once this was invented, information technology had begun to emerge.


4The components of the phrase information technology need a little discussion. First, what do I mean by ‘technology’?

5In today’s usage, technology is frequently bracketed with science, and has come to mean almost exclusively the gadgets and devices that we have invented to allow us to do things—as summed up in the advertising slogan ‘the appliance of science’. But this is a very limited view of technology. My (1944 edition) Shorter Oxford Dictionary defines technology as ‘a discourse or treatise on an art or arts; the scientific study of the practical or industrial arts; practical arts collectively’. It is no accident that this definition contains the word art (s) four times and the word science/scientific only once. Technology is the art of doing things, of changing the world. We might also think of the word technique, concerning ways of doing things, whether in the arts or the sciences.

6In this respect, technology is in some sense the opposite of science. Science is about understanding why the universe is as it is; it is about the rules and regulations, and the structures and regularities. Cyril Northcote Parkinson, in Parkinson’s Law, says

It is not the business of the botanist to eradicate the weeds.
Enough for him if he can tell us just how fast they grow.

7The ultimate achievement to which any scientist aspires is the discovery of a law; and a law of nature, just as much as a human law, is about what cannot be done—what possibly imaginable states of the universe are in fact forbidden. There is an old paradox, ‘What happens when an irresistible force meets an immovable object?’ But Newton has given us some laws, which tell us (among other things) that no force is resistible except by another force, but that any force is resistible by another equal and opposite force; and that no object is immovable.

8Technology’s view of the world is quite different. For technology, the existence of the universe in its present state is a constant challenge: how do we modify it? How do we mould it to our own ends? How do we avoid these famous scientific laws, or make them work on our behalf, enlist them to our service? Of course, we cannot actually break the laws of science (though sometimes technology discovers that the scientists had it wrong, and that what they thought was a law could in fact be evaded). But the ways in which we can make use of them are many and wonderful. The laws of mechanics, including those that govern leverage, are one thing that we learnt about the universe. But when Archimedes said (as a comment on those laws) ‘Give me a place to stand, and I shall move the world’, he was talking (metaphorically at least) technology, not science.

9In short, technology is about changing the universe, about knowing how to do things. Not necessarily on a large scale, of course—in fact some technology is about very small changes. But the knowing how is a necessary part. A tool or device is not technology per se; it is only technology insofar as it enables us to do things.

10Furthermore, technological change requires choice. Often, technological advances are proclaimed as liberating, as simply expanding our horizons and our opportunities. But as we adopt new ways of doing things through technology, we not only leave behind older ones, we render the older worlds impossible, unattainable (as discussed, for example, in David Rothenberg’s Hand’s End). The huge social change brought about by the availability of the personal automobile, for example, has now spread to almost all corners of the world. Even if the exigencies of climate change fail to force changes in this mode of operating, such changes will perforce come about when the oil runs out. But a return to the pre-car world of 1890 is simply out of the question—we have lost those ways forever.


11The word information is also a little tricky, and has been used in many ways at different times and by different people. Theses and books have been devoted to the question ‘what is information?’ and to discussing the consequences of the possible answers (two examples: Luciano Floridi’s Information—A Very Short Introduction, and Antonio Badia’s The Information Manifold). However, to go down that route would take me away from my main purpose. In this book I will take a rather naïve view of information. When a speaker speaks and a listener hears and understands; when the speaker’s voice is transformed by the telephone handset into electrical signals, and possibly again into radio waves, and something at the other end does the reverse process; when a writer writes and a reader reads and understands; when someone puts data into a database, and later someone else enquires of the database, and gets out these same data in a different form but still understandable—all these are processes involving information.

12The only general assumption I shall make is that there are indeed human agents involved at some point (even if I am temporarily concerned only with mechanisms and devices). That is, I shall assume that for some thing to be or carry information, there has to be at least the possibility that the result will at some time reach a human being and be understood. We may think of information as residing somehow in records, and in some sense this book is entirely about records and recorded information—but the human recipient is implicit in everything.

13There are certainly notions of information that do not depend on this assumption. However, identifying or understanding some general notion of information that would encompass all these is hard.

14There is a well-known theory of information, due in part to Claude Shannon (see his paper of 1948), which uses the idea that ‘information is that which reduces uncertainty’. It is possible to read such a definition as requiring no human; however, if we ask “Who or what is experiencing the uncertainty?”, it becomes clear that assuming a human (or at least a sentient being) helps with this conception as well.

15Shannon himself thought of his theory as having nothing to do with human beings or with meaning. His theory had huge influence in the years that followed, in many different fields, as described by James Gleick in his fascinating book, The Information. Gleick’s book takes off from many of the same historical starting points as will appear in the pages that follow. To Gleick, Shannon’s theory is central to the IT revolution, and encompasses all notions of information. To me, (and also to Badia, cited above) for all its undoubted power and usefulness, this theory fails to address some of the central features of information in the context of human-to-human communication. I shall make no further use of it.


16As we have studied animals over the last half-century or so, we have come to realise that many animals have some form of communicative behaviour. We know that bees dance to tell each other about good sources of food, that whales communicate over large stretches of ocean, that chimps learn from each other. Nevertheless, the human behaviours that come under the general heading of language are extraordinary in their range and scope.

17The evolution and invention of language is one of the great events in human pre-history. I say ‘evolution and invention’ deliberately. Steven Pinker argues strongly in his book of the same name that ‘the language instinct’ is exactly that, a basic instinct that has evolved and is part of what makes us human, common to all humanity. In this sense, language in the abstract does not really belong in this history of human invention. However, all those parts of language that are learnt and constructed, all the specifics of actual languages, must be regarded as invention.

18The invention of a language is a continuous process. Like most inventions, but even more so than most, languages are not the product of a lone inventor in a garret, but of a social process. Every writer or speaker who uses language inventively or creatively is contributing to the invention of the language, and every writer or speaker who copies or borrows from or imitates a previous speaker or writer is also contributing to the establishment of that invention. Language, in addition to being a basic instinct, is a technology that we use to change the world, by communicating with other people, and the social process by which particular languages develop is one long invention.

19We often refer to such a process of social invention as evolution. This is in homage to Darwin’s biological theory, but is not a part of it—it is an analogy rather than an application. Probably it is also the case that the biological evolution of humanity in matters relating to language continues to this day—perhaps in obviously physiological features such as voice production, but perhaps also in our language-processing capabilities. That is harder to see happening, however.

20It is also hard to know much about the origins of either the social or the biological process, and I will not attempt to go into either. The true starting point of this book, the point from which a genuine technology of information takes off, is the invention of writing.


21The question “What makes humans different from animals?” has been asked many times, and answered in many different ways—including, of course, the answer “They aren’t!”. Other answers have included intelligence, language, abstract thought and many other things. However, many of these possible answers have been undermined by discoveries in relation to other species. But it may be argued that one characteristic that really does distinguish us from other species is writing. Whether or not you feel the necessity for such a distinguishing feature, writing is certainly a strong candidate.

22Writing began, we believe, some time in the fourth millennium BCE—say five-and-a-half millennia ago—in Mesopotamia. A great account of the development of writing is given by Andrew Robinson in The Story of Writing. The first purposes of writing were relatively mundane; certainly not the recording of human history. They had to do with commerce and administration—with accounting, recording transactions, listing stock, identifying ownership, and so on. Later writing came to be used to glorify leaders, and to tell stories. These stories (and the people who wrote them) did not distinguish between myth and history. Later still came chronicles and real history, and philosophy and science and religious tracts and laws and administrative rulebooks and poetry and advertising and all the rest.

23But the receipts and the tallies and the laundry lists that started it off, however mundane, are central to the history. The kinds of information they represent are the mothers of invention. The people who invented writing felt the need to do several things. They wanted to supplement their own memory, they wanted to organise and impose order on the world, and they wanted to be able inform others of the validity of their own memory and their organisation of the world.

24Later I will expand these reasons for writing.

Systems of writing

25In order to make this invention of writing work, for any of the purposes mentioned, we need to have a notion of a system of writing. In principle, any mark (on paper or skin or stone or cloth or in clay or whatever) could mean anything we choose it to mean, as in Humpty Dumpty’s way with words (‘ “When I use a word... it means just what I choose it to mean”’—Through the Looking-Glass, Lewis Carroll). But that is not a lot of use unless there is some reasonable chance that someone, either the writer at a later date, or someone else to whom the message is directed, will be able to recognise the meaning. So we have to be systematic, at least to some degree, about assigning meaning to marks.

26The same problem has occurred long before, in language generally. We have some notion already of the relationship between words and meaning—that is, we can say in some way what a word means, and then construct new sentences out of existing words, whose meaning can be inferred from a knowledge of the words. Of course, this is a gross oversimplification of the notion of meaning, and in any case our present notion of words is rather highly dependent on written language. Consider for example German, where the written form allows some words to be run together to build longer words, or Chinese, which has no word boundaries in its written form. Nevertheless, it is a useful starting place for writing.

27At the very least, we might expect our system of writing to tie in with the words of our language, in the sense that the same word is represented in the same way when repeated. This assumes that written language does indeed represent spoken language, and that spoken language is made up of words. It provides us with one of the major ways in which early written languages were constructed—with symbols that may start as stylised pictures representing words.

28The system of writing that most clearly illustrates this method is ancient Egyptian—see Figure 1. Small pictures can be seen in many examples of Egyptian writing, and these can sometimes be translated via the words that the pictures represent (though actually ‘ancient Egyptian’ covers several different writing systems). But even the Cuneiform system, the wedge-shaped marks in wet clay that came out of Mesopotamia, has similar roots.

Figure 1: Ancient Egyptian writing—stela of Senusret III, Altes Museum Berlin

Figure 1: Ancient Egyptian writing—stela of Senusret III, Altes Museum Berlin​wiki/​File:Ancient_egyptian_stela_Senusert_III.JPG
CC BY-SA 3.0.

29In the earliest writing systems, a method based on puns was commonly used. That is, you may want to represent a word about which it is hard to draw a picture (an abstract concept, say). Then one of the methods open to you is to draw a picture of another, more concrete word, which sounds similar, and allow the punning picture to represent the abstract word. This picture now comes to represent the sound rather than the concept that it originally pictured. For example (using modern English words) if I want to represent the word son (‘my son’), I would find it difficult to represent that meaning directly with a picture. But the language has another word which is pronounced in the same way and represents a very concrete physical object, sun. A picture of the sun will do quite well to represent the word ‘son’, almost certainly unambiguously in the context of a sentence.

30This process can be taken one interesting stage further. If the word you want to represent has multiple syllables, it may be hard to find another word that sounds like it. However, you may be able to divide it into smaller parts (shorter words or single syllables or even alphabet-like sounds) and draw a picture to stand for each of the parts. Then you have a representation of the abstract word as a combination of pictures. You have taken a vital step towards a modern system! When each symbol represents a syllable of sound, we have a syllabary. Several such systems were invented, and indeed they still exist in languages such as Chinese and Japanese.

The alphabet

31For two thousand years or so, writing systems developed slowly; new systems were invented, and borrowed ideas from each other or introduced new ones, but the changes were not huge. Then, around the late second millennium BCE and the beginning of the first, a huge change took place. The alphabet was invented. In the account that follows, I have oversimplified many things—and borrowed a great deal from John Man’s book, Alphabeta.

32When children in cultures with alphabetic written languages are taught to write, they learn about alphabetic characters and sounds. The idea is that letters represent sounds, and that you can at least to some degree work out what (spoken) word is intended by putting together the sounds of the individual letters of the written word. It seems to be suggested that the natural units of sound are those encapsulated in the letters.

33Part of the theory of spoken language is based on the view that the smallest unit of sound that can be distinguished is something like a letter—a phoneme. But this is somewhat misleading, because it is hard to pronounce individual letters. Vowels may be pronounced on their own, but consonants usually need the addition of a vowel before we can actually speak them (hence the way, in English, we vocalise the alphabet as bee, cee, dee, eff etc.). In some sense, the natural units of sound in terms of which spoken language may be understood are not really letter-like; they are much more akin to syllables. Which makes the first stage of developing an alphabet—a syllabary—readily understandable, but lends an air of mystery to the next stage.

34So let’s step through the process. We have already seen how puns may be used, and how (as a result) a word may be broken into smaller words before it is written down. If we follow that process to its conclusion, we would try to think of an elementary set of single-syllable words, represent them as best we could (by pictograms or whatever), and then construct all multi-syllable words as combinations of these elementary words. Modern Chinese illustrates this approach very well. (At this point I am skating over some rather complex notions of the relation between spoken and written language, which certainly come into play with Chinese.)

35But a syllabary—a set of symbols to represent every possible spoken syllable—is a clumsy thing. From our vantage point of an alphabetic system, we may think of generating syllables from every possible consonantal sound, followed by every possible vowel sound, followed by every possible consonantal sound. There may easily be thousands of such combinations, therefore thousands of different symbols required, all of which have to be learnt (as any Chinese schoolchild will tell you!).

36How could we simplify it? Well, we need a couple of historical accidents. First, we need a language in which consonant sounds are always followed by vowel sounds—so that we can associate each consonant with its following vowel and not the preceding one. Among modern languages, both Japanese and Italian have some of this character. This means that we can get away with an open syllabary or ‘abugida’ (equivalent to consonant-vowel instead of consonant-vowel-consonant). An open syllabary can be very much smaller than a closed one. Modern Japanese makes use of three different scripts, two of which are essentially open syllabaries. For example, the hiragana script has 46 base characters.

37As a second historical accident, we need a language in which the vowel sounds do not vary too much. If most consonants, most of the time, are followed by ‘ah’ sounds, then we may be able to do without the vowels altogether. Modern Arabic is like this—it can be written without the vowels, and still be understood by the reader, because the vowel sounds are sufficiently predictable, and the ambiguities that sometimes arise can generally be resolved easily enough by context. Now all we need is a consonant alphabet or ‘abjad’—say 20-30 symbols.

38This sequence of events probably took place in the second half of the second millennium BCE, around the eastern end of the Mediterranean and the Horn of Africa. One of the cultures to adopt a consonantal alphabet was that of the Phoenicians, a people who traded throughout the Mediterranean region around the turn of the millennium. The consonantal alphabet was broadcast widely, and its survival was ensured. Note once again that it was the necessities of trade, rather than of literature or philosophy or history or science, that drove this spread.

39The final step towards the modern alphabet was an explicit invention, made by the ancient Greeks in the very early first millennium BCE. They observed the Phoenician system and realised just how useful and powerful an alphabetic system of writing could be. Unfortunately their own language was rich in vowel sounds, and would have resisted a purely consonantal solution. So they invented vowels to represent the vowel components of the sounds of language. Well, actually, they borrowed some of the symbols previously used as consonants, but which were not required for their language, and re-assigned them as vowels. And the modern alphabet was born.

40Later, of course, the Greeks would invent history and philosophy, and bring science and mathematics and many of the arts to new heights (I exaggerate only slightly!). Beside these, the final step in the invention of the alphabet might seem like small beer. Nevertheless, it is hard to overstress its influence.

41As this account suggests, the alphabet was (we believe) invented only once, albeit in several stages. It seems that the syllabary, which is perhaps the more obvious development of the idea of trying to represent the sounds of words, was reinvented more than once. But the alphabet is something altogether more peculiar. Its economy, the fact that we can get away with some 25 symbols to represent the entirety of our language, including words that have not yet been coined, is nothing short of astonishing. And its implications are going to reach far into the following 3000 years.


42Having achieved the astonishing knowledge that we only need a small number of symbols to represent the whole of past and future language, let us put general language aside for a while and think about numbers. Numbers figured strongly in early writing systems, being a very important component of the kinds of information we wanted to represent. And given that we have words for them, we can (in principle) write them down using the same system. However, they do have some peculiar characteristics, which we should perhaps worry about.

43For one thing, beyond a certain point we need to be systematic about how we name numbers—we can’t simply coin a new name for every new number we come across: there are far too many of them. This applies as much in spoken language as in writing, though whether the development of systematic ways of constructing names for numbers preceded the development of writing is not clear. Secondly, it seems obvious (though again, when it became obvious is not clear) that it makes sense to use the characteristics of numbers to guide us in devising a systematic representation. Thus if we can see a new number as the sum of two numbers for which we already have names, then it might make sense to use the names of the two known numbers to construct a name for the new one. Of course this requires that the idea of addition is already understood, but the early uses of writing suggest that this was the case.

44More generally, we would like the representations of numbers (verbal and/or written) to help us with the kinds of operation that we want to do with them. This general principle will take a long time to reach its final fruition—the Arabic number system with which we are familiar today. But in the meantime, early literate civilisations such as the Babylonians and Egyptians developed number systems of some sophistication, and the great mathematicians of classical Greece explored some of the ramifications.

Systems of numbering

45Before the alphabet took hold, numbering systems tended to use special symbols. The basic principle, that you have a symbol for each number of a special set, and indicate intermediate numbers (which don’t have their own symbols) as sums (additions) of these basic numbers, was established by the Sumerians in Mesopotamia and remained in place until the Arabic system took over. In the Sumerian system, the special numbers were one, six, ten, sixty, six hundred, and so on. Our 60-minute hour is supposed to be a relic of that system. But we are now much more familiar with the Roman system, where the special numbers are one, five, ten, fifty, etc.

46By the time of the Romans, the alphabet was established, and they did not need to devise special symbols for their special numbers—they followed the general principle of re-using their existing small set of alphabetic symbols for a new purpose. However, as we shall see later, this is not a true alphabetic solution to the number-representation problem.

47The Greeks actually had a variety of number systems. One of their methods had separate symbols for each of the numbers from one to ten, then twenty, thirty, forty, etc. This required 28 symbols to reach what we would now call 900, allowing numbers up to 999. This was a rather profligate use of symbols—they used the letters of their alphabet, but had to borrow a couple of extra ones from someone else’s alphabet. Also, 999 was rather early to stop, so they then repeated the alphabet but with a special extra mark on each letter, to get them up to 999,999. But it gave a rather more compact representation of numbers than the Roman one.

48A characteristic of all these number systems is that they run out. A point is reached where you have exhausted all the allowable combinations of all the defined symbols, and simply cannot represent the next number (without, that is, defining a new symbol for the purpose). Probably this did not, in general, bother the pragmatic Romans—as engineers and administrators they had the range of numbers that they required, and abstract notions of numbers which they could not represent, but would not need in any case, were of no concern.

49But the limitation has some interesting Roman consequences. Consider for example Julius Caesar’s books about his military campaigns. His armies or smaller forces are always measured in cohorts and legions, rather than in men. Those were, no doubt, convenient units to use; but it is also the case that Caesar would have had difficulty in expressing the size of his army as a number of men. The Roman system contained both names and symbols up to M (1000), and could therefore represent numbers up to 3999: if we wanted to represent 4000 in the usual Roman system, we would need a symbol for 5000, just as 400 (CD) makes use of the symbol for 500 (D). But a legion was between 3000 and 6000 men, and Caesar normally had several legions under his command. When he referred to the armies against him, he tended to use a mixture of numbers and words, such as ‘LX mille’—60 thousand, or 60,000. This is not unlike the modern habit of mixing numbers with the words ‘million’ or ‘billion’, but is forced on Caesar. In effect, he has to treat ‘one thousand men’, in words, as a single unit and then apply the usual numerical system.

50However, these limitations on all such number systems most certainly did bother the great classical Greek mathematicians. Archimedes, in the third century BCE, was particularly exercised, and invented his own number system which allowed him to express seriously large numbers (as an illustration, he calculated the number of grains of sand in the universe). But it was still essentially limited by an upper bound on the numbers that could be represented, albeit a very large upper bound.

51A true alphabetic solution to the number representation problem eluded even the Greeks. It had to wait another millennium or so.

The great Hindu invention

52The positional notation that we use today, whereby the same symbol can stand for many different numbers (for example, a "1" can mean one or ten or one hundred, depending on its position) was the revolution we needed. This in turn depended on the zero, as a position marker for an otherwise empty position.

53This invention was made by Hindu mathematicians in about the seventh century CE. Though once again, I am summarising with gay abandon a complex and not fully understood process, which may have occurred independently in different places at different times, and/or may have been influenced by earlier ideas—there is a great account in the book by Robert Kaplan, The Nothing That Is. The idea that reached the House of Wisdom in Baghdad, then the centre of the civilised world, was formulated as a general system of both numbering and arithmetic, and re-exported to the rest of the world as the Arabic system—as described in Jim Al-Khalili’s Pathfinders. One of the masters of this formulation was the Persian mathematician Al-Khuwārizmi, whose name has given us the English word algorithm.

54The Arabic system provided for numbers what the alphabet had provided for words—a way of representing any number, including those for which no-one has yet found a need. The ten decimal digits and the positional notation allow for the representation of any positive whole number. The Arabs also had a precursor to our (relatively modern) decimal point or comma for representing the decimal part of a number—allowing the same rules of arithmetic to apply to non-whole numbers too. It also provided a simple set of rules for arithmetic operations, again applicable to all numbers, which, stood us in good stead a millennium later when we began to try to mechanise arithmetic.

55Following the original revolutionary invention of writing, these two revolutionary and not at all obvious inventions, the alphabet and the Arabic numbering system, are two of the cornerstones of the developing technology of information. In ways that are quite unimaginable to their progenitors, they will reverberate down the centuries: they will give us ideas, and allow us to think things, that would otherwise have been, quite simply, beyond our ken.

Table des illustrations

Titre Figure 1: Ancient Egyptian writing—stela of Senusret III, Altes Museum Berlin
Légende​wiki/​File:Ancient_egyptian_stela_Senusert_III.JPGCC BY-SA 3.0.
Fichier image/jpeg, 59k


Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search