Philosophy of Hypertext Theodor Holm Nelson 10 January, 2002 Submitted for the Ph.D. degree at Keio University "Everything should be made as simple as possible. But no simpler." -- Albert Einstein Philosophy of Hypertext Theodor Holm Nelson I 0 January, 2002 Submitted for the Ph.D. degree at Keio University PHILOSOPHY OF HYPERTEXT page I ABSTRACT Philosophy is everywhere and permeates everything, like electromagnetic fields. This the same as saying that ideas are projected everywhere by human beings, and their consequences fan out in all directions across the universe. Any point of view can permeate everything and become a universe-filling paradigm. The collisions of ideas, too, happen everywhere across the universe. When strong points of view conflict, each filling the universe in its own way, they may conflict with each other throughout all subjects, on every detail and principle. Disagreements can reach into every area, especially when deep and important choices are to be made, as the interpenetrating philosophical fields of the different parties become complex and perturbed in their far-flung disagreement. The great question, "What should be the form of electronic documents?" is exactly such a deep and important choice, resulting in just such far-flung disagreements. The design of electronic documents is an issue with far-reaching consequences for work, society and the human heritage. It is also a question on which everyone seems to have an opinion now. The technical community, accustomed to a world of hierarchical containment and paper documents, has expanded this to a universe-filling, simple-minded philosophy. Looking only at the form of previous documents, especially typography and layout, they have given the world electronic documents based on hierarchical containment and the imitation of paper. The World Wide Web expands this paradigm across the universe, at the expense of content connection PHILOSOPHY OF HYPERTEXT page 2 and the needs of authors and intellectual property. The very different philosophy of General Schematics, first propounded in 1958, is about intercomparison and connection among free-form structures. From General Schematics comes an emphasis on parallelism, cross-connection, interpenetration, overlap, multiplicity, and the indeterminacy of boundaries. Based on General Schematics, the Xanadu* project has for decades striven to create a free-form parallel electronic literature consisting of deeply-connected parallel free-form electronic documents. Its technical mechanisms- through many been generations of design- have intended to support deep intercomparison, two-way links, the flow of content (with frictionless re-use of material under a unified and fair copyright solution), and facilities to help authors and editors in depth. Everyone understands the needs but not the great possibility- a clean alternative literary world offered by the Xanadu family of hypertext designs. *Note: "Xanadu" is a registered trademark; "xanalogical structure" is intended as a related generic term (non-trademark) for general use. "ZigZag" is also a registered trademark. "Parallel Textface", CosmicBook", "HyperCoin", "FloatingWorld" and "Possiplex" are claimed as trademarks. PHILOSOPHY OF HYPERTEXT page 3 CONTENTS ABSTRACT Contents [this page] 1 3 INTRODUCTION AND THESIS QUESTION 5 Part 1. THERE IS TOO MUCH TO SAY AND IT GOES IN ALL DIRECTIONS 9 The Rowboat Story 10 Writing 12 The Indeterminate Boundaries of Writing 13 Free-Form Writing Methods 23 Part 2. INFINITIES OF CONNECTION: The 1958 Schematics Paper 25 Discrete Representational Systems 28 Where Schematics May Have Been Ahead of its Time 39 What Led to the Schematics Paper 42 Part 3. WHY HYPERTEXT AND WHY ME 50 The 1960 Epiphany 51 My Hypertext Agenda 53 Why I Thought I Could Do It 64 Where the Ideas Came From 65 How I Was Right and Wrong 71 Part 4. TIIE PHILOSOPHY PART 72 Part 5. THE STRUCTURE WARS OF ELECTRONIC LITERATURE 87 Private Struggles: Perfecting the Xanadu Model 90 Public Struggles: What Happened Outside the Xanadu Project l 05 Part 6. WHAT DO YOU DO AFTER WORLD-WIDE HYPERTEXT? 109 Part 7. CONCLUSION 122 BIBLIOGRAPHY 128 ACKNOWLEDGMENTS - - - - - 153 PHILOSOPHY OF HYPERTEXT page 4 APPENDICES . Appendix A. 1958 Schematics paper: "Schematics, Systematics, Nonnatics" Appendix B. Walkthrough of the 1958 Schematics Paper Appendix C. Hypertext Notes, 1967 Appendix D: Transpointing Windows "Documents Are Parallel by Nature", 1998 "Examples of Parallel Documents", 1998 "Parallel Visualization: Transpointing Windows", 1998 "Parallel Data Structures", 1998 Appendix E. Xanadu User's Manual, 1976 Appendix F. Walkthrough of the Xanalogical Model Appendix G. The Transcopyrigbt Concept Appendix H. 1999 HTS design, "Transpublishing Structures for Today's Web" Appendix I. PUTTING XANADU ON THE WEB "Deep Hypertext: The Xanadu Model", 2001I "The New Xanadu Structure for the Web," 2001 "Transcopyright for the Web", 2001 "The Virtual Literary Format", 2001 "Problems of the Virtual Literary Format", 2001I Appendix J. Xanadu Design Ideas Appendix K. Principal Xanadu Internal Designs Appendix L. 1999 Disclosure of Technical Secrets, "Xanadu Technologies- an Introduction" Appendix M. Xanadu Designers PHILOSOPHY OF HYPERTEXT page 5 Introduction and Thesis Question PHILOSOPHY OF HYPERTEXT page 6 INTRODUCTION AND THESIS QUESTION This is about alternative designs for a media universe. In particular, it is a history and a theoretical discussion of design issues, and an explanation of the structure of the Xanadu software design. It contains philosophical elements and a personal story. A STORY OF PHILOSOPHICAL INFLUENCE One traditional form of a Ph.D. thesis delineates the influence of some obscure figure- say, a little-known philosopher- on some other, better-known individual or movement. That is the form of this thesis. The obscure philosophical figure was a college student who thought about a lot of abstract ideas from what was then a new angle, and attempted to codify these ideas sweepingly in 1958. It was a philosophy begun perhaps in a rowboat at the age of 5, which reached a certain plateau when I was twenty, in my third year of college. I will have to discuss this at some length. These ideas were much a part of the hypertext work began two and a half years later, in 1960, when I thought of the basic ideas ofhypertext. A perhaps-annoying amount of personal background will be required to explain why I quickly saw the computer as the cornerstone of a new world of literature and thought; why I believed I could design and build such a world myself; and what philosophy lay behind the idea. PHILOSOPHY OF HYPERTEXT page 7 The thinking I will discuss here may seem egotistical, even solipsistic, taking place in a closed universe divorced from other developments in the world or in the computer field. That is because all the ideas came from sources which preceded the work. In fact this work was influenced by many people, but these influences came early in the game, before I encountered computers. The work since then was intended to carry out the ideals and ideas that came from such people. The parallel developments of the rest of the world, and the computer field, have always continued to seem to me irrelevant and incorrect, even more so at this moment. In a famous anecdote, a different Nelson, an admiral, was warned about approaching ships. "I see no ships," he said, holding the telescope to the eye everyone knew was blind. At the end of the day the other men knew he had been right in pretending not to see the ships. We all choose what we see. I choose a better world of hypermedia than we have now. Please share my telescope. PHILOSOPHY OF HYPERTEXT page 8 THE RESEARCH QUESTION A Ph.D thesis traditionally has a "research question", although in many cases that research question has been surreptitiously framed after the research is actually done. Not in this case. The research question is one that I asked in different forms throughout my life, and in approximately its final form forty-one years ago. First formulation (circa 1942): What is the right way to describe things, when there are so many possibilities, and so many descriptions seem wrong? Second formulation (circa 1951): What is the right way to write things, given the incredible limitations of paper? Third formulation (late 1960): What is the correct design of electronic literature, to help us escape from the prison of paper? To explain this question in its variety, and answer it properly, will take considerable exposition. PHILOSOPHY OF HYPERTEXT page 9 Part 1. There Is Too Much To Say and It Goes in All Directions PHILOSOPHY OF HYPERTEXT page 10 Part 1. There Is Too Much To Say and It Goes in All Directions I have always been concerned with the immensity of description and the indeterminacy of writing. This is the fundamental problem of expression, in small and in large. It is not possible to say everything, and worse, to say anything leaves something out and cuts connections. All boundaries and divisions are artificial, and break vital connections. These problems are not merely abstract. They menace every attempt to say anything well, and they make the problem of writing immensely difficult- if you care about representing the real complexity. From a very early age, and a mystical experience I had at the age of five, I have been overwhelmed by the immensity of the world in its possibilities and its complexities, which I believe I envisioned in that early moment, imagining and understanding the vast complexity of the world's relationships, as abstract lines whirling and flexing through space, spinning out in all directions between all things. THE ROWBOAT STORY \Vhen I was five years old, or less, I was in a rowboat with my grandparents, trailing my hand in the water. And I felt the water opening around my fingers and closing on the other side, and I thought about the complexity of the relationships in the water. Those particles (though I did not know the term) which were together, as my hand approached, separated around my fingers, and then rejoined on the other side. But now the water particles were next to different particles. PHILOSOPHY OF HYPERTEXT page: II Actually it was not ·particles I envisioned, but relationships- distances that changed, lines and triangles and nets between the particles, all moving and changing at great speed in their tiny space. As two points moved apart; for instance, I could imagine them moving apart in some constant relation to one another-- perhaps a straight line, or a curve; but then that relation might change in some way, but still persist in some other respect, at a greater remove-- as a straight line became a gradual curve, for example. Thus persistences and constancies become more and more attenuated and far-fetched as the context expands. I believe that is how I thought about it in the flash of that very moment. And trying to imagine this glove-shaped slowly-swirling body of water, I was awestruck at its complexity, and the difficulty of correctly describing even so simple a scene. The movement of the water, so seemingly simple, was incredibly complex. The multiplicity of those relations in the water, the lightning-fast changes in their miniature universe, the inexpressibility of their persistences as they became more attenuated, confounded that visualization in their complexity. The many aspects and interpenetrating facts of this tiny scene as I imagined it filled me with awe. But if such a small part of the world were so complex, I thought, how much more so the whole universe! I remember all this, however vaguely, as a mystical glimpse into a vast and yet intimate relational manifold, in me and around me and everywhere- beyond PHILOSOPHY OF HYPERTEXT page 12 customary description, its particulars all beyond specific expression; not threatening, but a beautiful and awesome presence. I loved good descriptions even then, and strove to describe things well; but this was beyond expressionexcept as I describe it now, schematically abstract. Somehow I still sense warmth and awe and amazement from that moment, probably like that of people who believe they are experiencing God; but neutral, secular, wonderful and surprising, for it put an end to my easy belief in the expressiveness of words. Writing Description and then writing have always been centers of my life. As a child I was greatly respected for my ability to see and describe things. Then came writing. Members of my family were always writing, and in high school I worked with and studied serious techniques of writing, especially the deep reorganization then called "cut and paste". I have always seen the process and nature of writing as enormously complicated, embodying and forever re-enacting that experience in the rowboat. That rowboat experience was for me an epiphany of awe and wonder, and in a way all my thinking has come from that moment; and all the problems of expressing that thinking, for I have always had too much to say, even when describing small topics, because everything has so many aspects and relations. We can call this multi-describability. Everything has many aspects, facts, relationships and connections; therefore everything can be described in many ways. PHILOSOPHY OF HYPERTEXT page 13 This is where the complexity of writing (and explaining) begins: choosing and expressing complex ideas so they can be read and understood. This means trying to say about a thing what will convey the most understanding. And whatever you say will leave something else out. (The philosophical ideas of General Schematics, which will be discussed later, are an attempt to express on a technical level these aspects of complexity and description.) The Indeterminate Boundaries of Writing The central problem of writing- of description, really- can be formulated in several different ways• There are so many relations to be presented, and so many connections. • Every assertion extracts some relation from the conglomerate of possible connections to describe, the possible things to say. • There is no knowing what to say first. • There are no natural boundaries to what may be said. • Any description cuts connections by omission. A SIMPLE BUT INTERESTING EXAMPLE To illustrate this last point, I will list below some interesting items (or points, or notes) which cluster around a single subject. But the subject has no boundary. I have chosen the story of the Alamo because it is famous in American history, and it interconnects in fascinating ways with many other topics, including the PHILOSOPHY OF HYPERTEXT . page 14 history of chewing gum. Let me enumerate some written notes, without regard to sequence or boundary. PHILOSOPHY OF HYPERTEXT page 15 AlamoPlex of Items I Notes • The Battle of the Alamo was fought on 6 March 1836, by about 189 Americans defending a compound in San Antonio, Texas. All the defenders were massacred by a force of Mexican soldiers. • The defenders of the Alamo called themselves "Texians." •At the time of the Battle of the Alamo, the Mexicans believed that Texas was still a part of Mexico. Those fighting for independence believed they had become an independent state. • Prior to the war, many Americans from southern states of the USA had moved to Texas, many believing that eventually Texas would somehow become part of the USA. • The defenders of the Alamo are great heroes in American history, and included three famous men- William Barret Travis (who was in command), Davy Crockett (woodsman and former Congressman), and Jim Bowie (for whom the big Bowie knife was named). • The Battle of the Alamo was possibly the pivotal event in the War of Texas Independence. By slowing down the movement of the Mexican army of cadets under Santa Anna, it gave the Texians time to prepare. • General Houston did not send to the Alamo the reinforcements he had promised. Instead he prepared to meet the Mexican troops after they left San Antonio. • Soon after the Battle of the Alamo, General Sam Houston won the decisive battle of San Jacinto. • San Jacinto was near the city which is now called Houston, in honor of the general who won the battle. •Sam Houston went on to become president of the Republic of Texas. • In a famous moment, Travis drew a line in the sand, and said that all the men who were prepared to die should cross the line. All crossed but one. Jim Bowie, who was sick, was carried across in his bed. • The Alamo battle essentially won independence for Texas, although it - - - --- PHILOSOPHY OF HYPERTEXT page 16 was years before the issue was settled politically and before Texas joined the USA. • Many regard the war of Texas Independence as a plot to steal the best part of Mexico for the United States. • Lt. Col. William Barret Travis was a young man of 26. He disobeyed the order to abandon the Alamo and determined to hold it, which was militarily impossible. • Travis sent out a call for others to join him in the defense, which went across the USA like wildfire. It ended, "VICTORY OR DEATH". • Travis' letter was possibly the most successful piece of advertising in history. Those who came included Bowie and Crockett. • The number of defenders of the Alamo killed is generally reckoned at about 189. A few non-combatants were spared by the Mexicans. • The story of the line in the sand was told by Moses Rose, the man who declined to cross it and left. He was disgraced for the rest of his life. Had he told the story in a different way, he might have arranged to be considered a hero. • The song most people remember as "There was something in the air that night, the stars were bright, Fernando"- a Mexican folk song made popular most recently by the group ABBA- was from the Mexican side of that war. • Davy Crockett, a folk hero in his own lifetime, became a folk hero again circa 1953, when the Disney movie "Davy Crockett" became a big hit, and millions of children wanted raccoon-skin hats. • Santa Anna, the vainglorious general who stopped to kill the defenders of the Alamo and because of it lost the Battle of San Jacinto, continued to dominate Mexican politics for many years and was twice in exile. • During a period of exile, Santa Anna stayed on Staten Island in New York with a photographer named Adams. Adams saw Santa Anna chewing chicle rubber, became interested in its commercial possibilities, and after several unsuccessful experiments created chewing gum, which PHILOSOPHY OF HYPERTEXT page 17 was wildly successful. • "Black Jack" chewing gum, created by Adams himself shortly after the Santa Anna stay, remained on sale for a hundred years. (I often chewed it as a boy.) • Emily Morgan is revered as a treasured heroine of the state of Texas for her legendary role in defeating Santa Anna. According to the story, on the eve of the battle she seduced Santa Anna and next morning delayed his readiness- so that the battle for Texas independence was won with hardly a shot fired. • The song "Yellow Rose of Texas" is said to celebrate the achievement of Emily Morgan, who was "mulatto" (and thus, in the terminology of earlier days, "yellow"). • Some recent scholars claim that the Emily Morgan story is a myth. However, a hotel in San Antonio is named after her, existent or not. • Sam Houston was a cantankerous general and later governor of Texas. He won the war ofTexian independence. • Houston promised to send reinforcements to the men at the Alamo but didn't. However, their sacrifice won the war. • Some say the real motive of most Texians was to join the United States, but they first declared "independence" in order to make it look less like the theft of territory by the USA. • Thomas Adams tried various uses for chicle, including substitute tires for automobiles. • Thomas Adams packaged sticks of chicle, without flavor, for chewing. The first production run sold very well to pharmacists. Flavoring was only added later. • The Disney movie "Davy Crockett" was an enormous hit m the nineteen-fifties. • Fess Parker, the leading actor in the Disney film, had no further hits and left the movies for a successful career in real estate. PHILOSOPHY OF HYPERTEXT page 18 [ Figure I. The Alamoplex of individual factoids. ] ong, "There Was Something in the Air that Night" Battle of Goliad William Barret Travis Travis' letter calling for help Response to Travis' letterthey come from everywhere BATTLE OF SAN JACINTO The Line in the Sand Travis' death at the Alamo Jim Bowie's death at the Alamo Santa Anna as general Santa Anna's political career LIFE OF Raccoon hat craze DISNEY MOVIE, "Davy Crockett" Thomas Adams CHEWING GUM Adams attempts to make tires out of chicle PHILOSOPHY OF HYPERTEXT page 19 This particular subject area or zone, which I will call the "Alamoplex", is by no means unique in these regards; I simply chose it because it is particularly interesting and because the chewing-gum connection is so outlandish. WHAT THESE ITEMS ILLUSTRATE These items (especially as shown in Fig. I) illustrate many things • The tangle of topics in the world • the indeterminacy of subject boundaries • the indeterminacy of "natural" contents of a document • the difficulty of categorizing and bounding subjects • The category problem in general • Overlapping sets of ideas • The filing problem All these problems are inseparable. SUBJECTS, WRITING AND FILING It is difficult to write about things for the same reason it is difficult to file things. Subject headings are a complicated compromise: in every book there are many topics, but the book is generally put "under" one subject for the general convenience of librarians. Subjects are fractal. The "subject" given to a book is often a forced description; read individual chapters and they may no longer fit the so-called subject of a book. The subject of a chapter, a paragraph or even a sentence may not fit well into any given category. PHILOSOPHY OF HYPERTEXT page 20 CATEGORIES AND BOUNDARIES OF WRITINGS AND DESCRIPTIONS Now let us consider: what could be done with these items? What categories do they fall into? What could be written from them? What is the "natural fonn" these notes should take when collected into some article or essay? They include many different people, history, geography, several industries, and three hit songs. It is my belief that there is no natural fonn. There are many ways to arrange these notes, many ways to categorize them, many possible sets and boundaries for writings that could be made of them. If one chooses to make an article of these notes, there is no determinate natural boundary which would determine its subject or boundary. Figure I, above, illustrated this variety; containing and overlapping ellipses show the overlapping subjects that are included here. Figure 2, below, shows how a subset of these notes may be selected for some particular use, but such selection ordinarily loses connections. The new group of these particular items becomes separated and lost from the context of the original full set. As shown in Fig. 2, you may select a subset of these items for some special use or story-telling. But normally the subset loses its connection to the larger context (shown by the green lines). This will be mentioned again. PHILOSOPHY OF HYPERTEXT page 21 [Figure 2. A subset of notes selected for use in a piece of writing. ] PHILOSOPHY OF HYPERTEXT page 22 CONVENTIONS OF WRITING According to the usual conventions of writing, the author chooses a subject, perhaps a central point to make, and organizes the notes to make that point. We have various expectations about this. But at the deeper level, these conventions of writing make no sense. If all the points are of possible interest to some reader, and the boundaries of a piece of writing are completely arbitrary, why must there be boundaries? Why cannot the structure of a written piece continue in many directions, following the connections away from the supposed central point or theme? This is the sort of point that was made in the Twentieth Century by such "stream-of-consciousness" authors as James Joyce and Samuel Beckett. Just as there are no boundaries to experience or to thought, there need not be any boundaries to stories either. FOOTNOTES Footnotes are small initiatives to reach out of some arrangement of content, but they are extremely limited, like hands reaching out of jailhouse windows, constrained by a different plan. CONVENTIONAL ORGANIZING Conventional documents are often hierarchically organized, based on numerous conventions as to what are acceptable subjects and their boundaries. For example, there are conventional formats for written reports in the army and many other organizations, and this can simplify many writing tasks. But in the general case- say, writing an article for the New Yorker, if you are lucky and talented enough- any sentence could in principle be at the beginning or the end. There PHILOSOPHY OF HYPERTEXT page 23 is no given structure. OUTLINING VERSUS FREE-FORM WRITING Some people are able to make an outline of a subject and follow it in their writing. This is certainly legitimate. But I believe that if an author is trying to find the true shape of an idea, and represent it correctly, the problem is far more difficult. In free-form writing, there are many issues. One is optimizing the boundaries and connections in deciding what connections to break. And another is in finding the mot juste, or finding some surprise connection that determines the organization of a piece. Free-Form Writing Methods Most writing is rewriting, to improve the emphasis and shape and flow of a document. And most rewriting, especially of large documents, is rearrangement. Unfortunately skills and understanding in this area seem to be going down. The term "cut and paste", prior to the Macintosh*, meant cutting up one draft of a work and rearranging it, with changes, to make the next draft. In old-fashioned cut-and-paste, the author repeatedly tries to find more appropriate boundaries for the topic and sequences for the content. * Two Xerox PARC veterans, Larry Tesler and Jeff Rulifson, have both claimed to me that they were responsible for this change of meaning (117, 113). - - -- - - - -- -- PHILOSOPHY OF HYPERTEXT page 24 However, the Macintosh in 1984 radically redefined these terms: no longer a parallel mental process, but a sequential process of putting things in a hidden place (the "clipboard") and blurting them back again. This justifies a bad writing tool by assimilating it to the old terminology. FREE-FORM WRITING METHODS: THE KJ METHOD As I understand it, the Japanese KJ method (112, 111, 106) is a combination of free-form and hierarchical organization. Spreading the items out (in a way similar to old-fashioned cut-and-paste), the author considers their best groupings and creates heads for these groupings. (I look forward to software for th.is method being generally available.) THE EVILS OF PAPER AND PUBLICATION The problem with creating sequential documents is this: any sequence cuts connections, just as any grouping omits items. The cutting of connections is the loss of information. The problem is worsened with publication. Usually, a document submitted for publication has content edited out. Information is lost, content is lost, connections are lost, often forever. Before hypertext, these problems were intrinsic to writing and publication. PHILOSOPHY OF HYPERTEXT page 25 Part 2. Infinities of Connection PHILOSOPHY OF HYPERTEXT page 26 Part 2. INFINITIES OF CONNECTION: The 1958 Schematics Paper In 1958, at the age of twenty, in my third year at Swarthmore College, I wrote an unusual term paper. I worked on it very hard. The paper was late, being handed in well past the deadline. It was brash, playful, hard to understand, sweeping in scope, vastly ambitious, and sloppily expressed. Looked at from today's perspective, it may have been rather original and perhaps ahead of its time. (To be discussed on page 39 ff.) The paper, entitled "Schematics, Systematics, Normatics," is included here as Appendix A, as scanned from the original. (Unfortunately some pages had to be scanned from old-fashioned hectographic masters, and two of the pages are not completely readable at present.) I believe the paper got a B+. The paper was a somewhat inspired and sweeping, but not fully baked, attempt to put together a new philosophical system of thought, logic and expression. It was confusingly stated and hastily finished. It should be thought of as a daring leap of thought by a brash third-year college student. It is phrased irritatingly in a semi-private language, with strange and utterly unclear allusions to various fields. I apologize. Much of this was due to impatience coupled with a great sense of inspiration. PHILOSOPHY OF HYPERTEXT page 27 Without detailed explanation, the Schematics paper is incomprehensible. Indeed, unless it is read with great sympathy, the paper borders on gibberish. However, an overview is provided below, and a walkthrough for page-by-page clarification in Appendix B). The reader is invited to try to understand the sweeping point of view. It is difficult to put in perspective as a reader, let alone as the author of long ago. But the paper expressed in peculiar fashion a number of ideas which constitute a specific philosophical position, a system of ideas which have been in my thoughts ever since; and which have had a strong bearing on my computer work. I believe the paper is interesting both as a very ambitious student project with few direct antecedents and some prescience of what was to come after; and as a rich way of thinking about structures, relationships, abstraction and design. Working on it sensitized me to many different issues across many fields. I had no technical background for these ideas, but rather a keen interest in abstraction and thought. I felt that the basic ideas were obvious (still do). I had not taken higher mathematics (which required calculus as the next course, very far from these ideas.) I had not heard of graph theory. I had not heard of constraint structures. I had not heard of finite mathematics. I had certainly not heard of data structure. But in this fierce little piece I kind of took a stand on all these topics. I think that anyone else would also have had great difficulty in trying to say these things, especially back then. After the Schematics paper was handed in, I realized that it was not fully baked and that if I wanted to continue thinking along these lines, I would have to do so for at least several months more. That was in 1958. It has taken much longer. PHILOSOPHY OF HYPERTEXT page 28 Because the scope was sweeping, alluding to and trying to fit in so many ideas, it has taken a long while to sort out the thoughts. I can now say these things much better (below and in Appendix B). I did not abandon this work, but put it aside for slow improvement. Rather than work on it in a rush, I have been reworking the ideas at a leisurely pace ever since. I have continued with these ideas, which I now call General Schematics, for the forty-three years since, refining the terminology, attempting to build them into a sort of catalogue raisonnee of interesting relational structures and how they figure in our lives. It is not, unfortunately, ready for publication in any substantive fashion. There have been many working titles for it, should such a writing occur. At the present moment the working title is: Principia Schematica: Structures, Ideas, and Strategy. Discrete Representational Systems I will here outline the main ideas of the piece, which the interested reader may just be able to recognize in the original; should the reader be deeply interested, there is a more detailed explanation in Appendix B. I began by imagining all ideas as discrete relational models ("schematics"). Here were some principal assertions implicit in the paper: • Everything can in principle be expressed as discrete relationships (since any symbol-system is discrete, and discrete symbols manage to represent continuous structures). • The really big idea, to me, was that anything could be represented as a set of relations, in fact a huge set of relations- even the immensity of properties and nuances in a human face (and its resemblances to family). This was distinct from thinking merely about "things" and "objects" and PHILOSOPHY OF HYPERTEXT page 29 "properties", which I now saw as clumsy. To think in terms of overlapping, interpenetrating relational structures was a much broader canvas. • Our minds deal in large part with discrete models; these can be represented as wholes of discrete relations (which may sometimes be represented diagrammatically). We think in large part with such discrete structures; we argue with such discrete structures; we plan with such discrete structures. • Such models need to be appreciated as wholes, and we needed to create a new form of visualization and deduction within such wholes which does not break them into parts. • Resemblance- the cornerstone of learning and of language and of science- is the discovery of shared relational structures between instances. (Such shared relational structures these might be enumerable in some cases, leading to a machine that could find resemblances from the enumeration of all relationships in different instances.) •Rich intercomparison is thus the key. By intercomparing examples, and compressing the examples into relational structures representing those rich relations that the examples have in common, we create models that generalize the examples. • This was how linguists found out the relations of Indo-European languages (to be discussed on pages 34ff.; unfortunately this key example was somehow omitted from the actual paper). It is also what happens when you see that two people look alike. From the extremely large number of relationships in the two faces, you find second- and third-order properties that are the likeness. Similarly, we may seek to find commonalities among the way particular people throw flying disks (":frisby thrusts"). PHILOSOPHY OF HYPERTEXT page 30 A MODEL FOR UNIFIED STRUCTURES Another key aspect was seeking to understand structure and unity in general. General Schematics, as of 1958 (though I had not yet chosen that name), sought any way to delineate unified structures- structures of relationships which held together as a whole. The 1958 Schematics paper was, among other things, a proposal for a new system of human logic- meaning a representation of ideas, assertions and their consequences- that will help people think and discuss. (As distinct from formal logics not useful in everyday matters.) Applied to thinking and reasoning, this would treat thought as a unified whole, without dividing it into propositions, on the classical Greek or Euclidean model, or deductive models that proceed on a proposition-by-proposition basis. Here are some of the thoughts behind that, stated far more clearly than in the 1958 paper: • Dividing a system of ideas or assertions into statements is misleading because the unity is lost. • There is no beginning or end for a system of assertions. To say something at the beginning gives it false emphasis. (This is a fundamental problem of writing; it is an unfortunate aspect of education; it is also a problem in conversation.) • Conventional logic has a number of shortcomings. The division of ideas into separate statements loses the unity, and division into axioms for deduction is misleading. (Euclidean geometry is an empty formalization in its traditional form; the Euclidean system requires a series of specific deductions in fixed lattices of derivation precedence.) • Some sort of deduction ought to be possible that works from a unified representation of the whole set of ideas. Such a holistic representation of ideas ought to begin with diagrams. Such a holistic representation of ideas PHILOSOPHY OF HYPERTEXT page 31 ought to have transformations that are the equivalent of deduction and work graphically. • It would be better to have some holistic system of representationideally some visualizable holistic representation, though I did not necessarily mean diagrams literally. • We should be able to deduce facts directly from such a diagrammatic representation as a structural unity. Thus the individual "facts" and premises are given less emphasis, and their intermingling as a whole becomes the emphasis. •Such a system could enormously clarify discussion (and even, one might hope, political argument). I was proposing, implicitly, a new way of expressing ideas: a new kind of symbolic logic based on the individual discrete relation (and its holistic representation by a structuring built from arcs of undirected graphs), rather than beginning with Boolean relations and working up to predicates. Indeed, I believed you could begin somehow with the unified structure, perhaps represented as an overall diagram. Such a new unified diagrammatic representation would hold the set of ideas and relations as a coherent unity. I further thought that we may be able to find various graphical transformations on such diagrams which allows them to be combined (combining diagrams would combine assertions); and manipulated, "rotated" in some sense, to bring forward other aspects of the relational structure. I was calling for both a unifying diagrammatic way to represent structures of ideas, and some way of transforming the diagram or structure to make other assertions in the complex more visible. (Computer animation as a possibility was of course not then on the horizon.) TRANSPOSABILITY AND COMBINABILITY PHILOSOPHY OF HYPERTEXT page 32 The models discussed in the 1958 Schematics paper are also particularly concerned with transposable and combinable structures. It happens that linguistic structures are very transposable and combinable. A RADICAL VIEW OF CONSISTENCY There are remarks in the 1958 Schematics paper that take a tolerant view of inconsistency. This is because inconsistency of ideas is not necessarily wrong. Inconsistency of ideas can generally be well defended by the holder of those ideas. It is very hard to catch someone in a contradiction, much though we may think his point of view is inconsistent. More recently, I have come up with the term supercompleteness * for structures which might be considered contradictory but which have some rationale- for example, screenwriters who are brainstorming about possible plot twists are working with a supercomplete set of elements. So is a detective considering a variety of contradictory hypotheses. Supercompleteness may go with a state of creative and useful uncertainty based on a lot of information that has not yet been narrowed down to a consistent set. *It seems that this term also has a mathematical definition, which may or may not have a similar meaning. RELATIONAL DEFINITIONS OF :MEANING I was proposing a relational definition of meaning in many areas: such terms as vicious circle, tight squeeze, victim, keystone, "melting pot" may clearly be relationally defined. PHILOSOPHY OF HYPERTEXT page 33 Further, that everything that matches a structure is an instance of it. (This is very like Russell's definition of the number 2 as "the class of all pairs".) THE MODEL OF ABSTRACTION General Schematics, as of 1958, sought to delineate a model of abstraction (whose origin in linguistics will be discussed below). The model is simple. • An object can be thought of as a virtual infinity of relational structures. Any two objects have in common a huge number of relationships, although those relationships may be at a considerable remove. • We study a domain by selecting examples, aligning them, finding the corresponding relations out of these virtual infinities of relations. • From the intercomparison of these corresponding examples, each being considered as a multiplicity of relational structures, we seek to derive a hard-edged model which expresses the commonalities and eliminates the differences. PHILOSOPHY OF HYPERTEXT page 34 REMOVES OF GENERALITY Generalizations can be broadened to include more and more instances, but the predicates on them are at further and further removes, likely to become more attenuated and lose definiteness with fewer certainties. KEY EXAMPLE: HOW STRUCTURAL LINGUISTICS FOUND RESEMBLANCES AT SUCCESSIVE REMOVES (Unfortunately, the principal example that I had in mind was omitted from the Schematics paper; I present it here from memory. However, the closely-related example of flying disks is on pp. 20-21 of the original paper, Appendix A.) In propounding this I was trying to generalize the way laws had recently been found in structural linguistics. First, consider the plight of 19th-century linguistics. Studying only Indo- European languages, linguists had created a false model of the structure of language, with a grammar based on Greek, and a list of 'fundamental sounds' based on Italian (as in Italian operas, which were thought to use the prettiest vowels). However, in the nineteen-twenties, the linguists Bloomfield and Sapir, studying Native American languages, found that these languages had absolutely no resemblance to Greek in their structure, and indeed that the different language families of American Indian languages were radically different from each other. Contrary to earlier linguistic theory, the sounds of Latin and the grammar of Greek are not reasonable models for other languages. Bloomfield and Sapir found the fundamental similarities among all languages to be at a greater remove than a resemblance to Greek or Latin. PHILOSOPHY OF HYPERTEXT page 35 • Phonemics. One of the principal things all languages have is a set of sounds and their alternative versions (phonemic system). It was discovered in the 20th century that that each language has its own distinctive integrated phonemic system, an abstracted layer of speech sounds that is quite unrelated related to Latin or Greek (except for the European languages). The phonemic system of a lanaguage consists of the particular recognized sounds of that language and their mutual separation. The structures in a phonemic system will vary from language to language. Within a given language family, the sounds will vary from dialect to dialect (perhaps unintelligibly), but in the mutually maintained relationships of the sounds it remains the samephonemic system. • Morphemics. In the same way, modern linguists discovered morphemics as the abstracted layer of speech units, separate from previous notions of Greek grammatical categories as perfect. HOW TIIESE COMPARISONS WORKED: SCHEMATIC ABSTRACTION AT SUCCESSIVE REMOVES We may look at these discoveries schematically, in terms of relational abstraction. The 19th-century linguistic model considered the sounds of Italian as a model or template against which other languages were compared. However, this misfit between the model and the actual sounds of other languages was finally recognized as overwhelming. Instead, by comparing the different languages of a language family (say, English and German, or two Algonquin languages), it was possible to find out the phonemic system the two languages have in common- the specific number and relative spectral position of the sounds that are kept distinct from one another. At a higher level of abstraction, if we consider English and German and Persian, there is still a common phonemic system, but the numbers and separations are PHILOSOPHY OF HYPERTEXT page 36 more complex to state. In general: the further apart the things we are comparing, the fewer identities between the nets of relational structures will be found, and the more attenuated the commonalities. At a still higher level of abstraction, looking at many different languages makes it possible to understand phonemic systems themselves. In the 1958 Schematics paper, it was this linguistic example I was endeavoring to make into a general model of abstraction, as follows: intercomparing instances will show you the similarities, giving you relational models of what two or more instances have in common. As the instances grow further and further apart, however, the commonalities (and the predicates to describe them) become more stretched and their expression more removed. In a similar fashion, it was possible to infer from Indo-European languages the former existence of a Proto-Indo-European language. It was inferred as a structural abstraction from the commonalities among a number of languages. Continuing the same abstractive process, there are scholars today who want to infer the structure of a hypothesized earlier language, the parent language of all the Eurasian language, called ''Nostratic". In principle this should be possible, but at greater removes of certainty and precision. To generalize this model of abstraction: • Take a number of examples- languages, creatures, cultures, whatever. • Seek to find what they have in common. With luck and good research, discrete hard-edged models may be found. • The further apart the instances are, the more attenuated will be the defining properties of these models- in other words, the more "abstract" PHILOSOPHY OF HYPERTEXT page 37 the findings will be, with exceptions and irregular predicates. OTHER GENERALIZATIONS IN THE SOCIAL SCIENCES In the 1958 paper, I hoped to apply these models to some key domains from the social sciences. (This was not made very clear in the paper.) One was kinship systems- the names of human relatives in a given culture, which are definitely schematic relational structures: they are transposable and combinable systems of relations. What made this more interesting were the claims of the anthropologist Levi-Strauss; he said the kinship structure transposed deeply into a culture's psychological realms. Here was a very interesting example of a relational structure transposed and generalized. Would this be an attenuated predicate, made vaguer by the breadth of examples it was made to fit? There are many other models in the social sciences, some of which are extremely peculiar. The hypotheses of Sigmund Freud, for example, who postulates a set of secret internal personalities that control behavior. There are references in the 1958 Schematics paper to such models. STRUCTIJRES FOR DOING AND BUILDING Wherever people do things, they create units to put together. From the great universe of possibilities, we select elements and relational structures- specific parts and mechanisms- that can be fitted together repeatedly and with variation. One such example of transposable components was advertised every day in the newspaper when I was a boy. The "Arthur Murray Box Step" was a structural component- a square diagram of foot positions- marketed as the cornerstone of a system of dance teaching in the 1940s. Supposedly, if you could count and move your feet according to this very - - - - - PHILOSOPHY OF HYPERTEXT page 38 simple structure, you could learn to dance. As a student progressed, the Arthur Murray teaching system would add new structural models to be used between box steps. This worked, up to a point; but it completely missed the characteristic body swings of each dance. It was a marketing gimmick designed to make students confident, rather than a good teaching method in itself. This particular example made a considerable impression on me for a number of reasons. I already liked structural models; this one was theoretically quite clear; but though I took a number of lessons at Arthur Murray, I did not learn to dance well. This indicated that particular transposable discrete structures may not be the right ones. This experience tied in with my thinking about discrete structural models, abstraction and mismatch as a unified issue. Another interesting example of trained movements as structural components is the martial arts. As I understand the martial arts, they all have the same basic approach: they divide battle into a few particular practiced moves. The martial arts initiate is trained in a number of specific moves until he makes these moves as fast reflexes. (The particular patterns of moves vary between the different martial arts.) The moves become a vocabulary of transposable actions from which the practitioner selects at lightning speed. Rather than thinking to decide which move to make and how to make it, the practitioner instantly selects a move from the long-practiced set and adapts it to the need of the instant. This skill in particular moves more than makes up for the loss of thoughtful optimization of a particular choice. (I will speak on pages 75 to 80 about this design and selection of components, which I call "substruction".) RULES: NORMATICS AS SUBSUMING Ennes AND DESIGN PHILOSOPHY OF HYPERTEXT page 39 The inappropriateness of many rules, and their application, also made a considerable impression on me- as I am sure it does on every kid. However, I came to see ethical and moral problems as being like the Arthur Murray box step- badly-fitting templates of action and judgment, usually in some larger system of templates of action and judgment. I called this branch of schematics Normatics, and was surprised to find the word in an old dictionary, meaning exactly what I wanted it to mean. Normatics is thus the study of templates of action and judgment, their complications, applications and consequences. We will not deal further with it here. Where Schematics May Have Been Ahead of Its Time I like to think that this paper is historically interesting in several respects, however unclear and rough it may be. It seems that it may have foreseen a number of later scholarly developments. The paper implicitly put together a number of ideas that were floating around in psychology, linguistics and the philosophy of abstraction. These ideas have taken shape in subsequent history along a number of different lines. These include: • Cognitive Science- discrete relational representations of structures of thought and cognitive operations on these structures. Cognitive science, especially since Bruner, Goodnow and Austin's A Study of Thinking, published in 1956 (83), has treated the process of abstraction and concept formation as an important empirical study, but one with strong philosophical overtones. The mind learns distinctions and commonalities, and forms mental maps, as internal representations of the world with PHILOSOPHY OF HYPERTEXT page 40 certain benefits and payoffs. Thus in cognitive science there is a dual concern with both the abstractable characteristics of the world and the abstracting characteristics of real minds. The 1958 Schematics paper endeavored to represent the process of abstraction, commonality, representation and modification of ideas as a unified whole based on discrete models. While theorizing about mental models had been part of psychology since William James ( 1890) and Tolman's cognitive maps (1918), the notion of mental models as precise structures is generally attributed to Johnson-Laird's Mental Models (118). Work on such systems of representation continues to the present day. • Knowledge Representation systems have been a vital and widespread source of study, both in artificial intelligence and beyond. Knowledge Representation systems attempt to hold, and analyze, structures of assertion (miscalled "knowledge")- represented as discrete relations among concepts. The idea of representing holistic, overall structures of understanding has developed as a key aspect of both Artificial Intelligence and Knowledge Representation over the decades since the 1958 Schematics paper was written. It is interesting to note that in that same year, 1958, John McCarthy (who coined the term "artificial intelligence") "proposed that all human knowledge be given a formal, homogeneous representation, the first-order predicate calculus."* *Paul N. Edwards, The Closed World: Computers and the Politics of Discourse in Cold War America, chapter 8. Available at http://www.stanford.edu/group/mrndd/SiliconValley/Edwards/ClosedWorld 1995. b ook/Chapter8.html Continuing structures of Knowledge Representation have gone in many directions. Said to be the most ambitious is the "Cyc" system of Douglas Lenat, which has seriously endeavored to represent all the structures of everyday thought. PHILOSOPHY OF HYPERTEXT page 41 •• Holistic approaches and declarative programming. In another direction, such holistic approaches as the Prolog language, and the holistic approach which came from it (now called declarative programming), have attempted to find unifying mechanisms which can treat bodies of information as manipulable wholes. This holistic representation was implicit throughout the Schematics paper. I am told anecdotally that such systems are appearing in mathematics now. For instance, I am told that "Microlocal Analysis" is a completely new formalization that compresses what used to be complex expressions of the calculus into small elegant unities that are better grasped (114). ••Semantic Nets. The term "semantic net" was introduced by M.Ross Quillian, whose Ph.D thesis was written in 1960, two years after the Schematics paper, (summarized in his later book (98)). These represent specific strategies to knowledge representation. It should be noted that the tree-structured asterisk notation introduced two years earlier, in the Schematics paper, was intended as a semantic net of a kind. This was intended as a higher abstraction than specific predicate calculi. However, because it used an asterisk for any relationship, it represented unassigned morphisms rather than specified relationships, and was limited as to what structures it could represent. This limited notation could represent graphs of connectivity, dependency or constraints. (However, it did seem to me that theoretical automata based on this approach might somehow be refined to somehow find all predicates as special cases.) •• Ontologies. The world of Knowledge Representation has most recently become organized around the concept of "ontologies", formalized semantic nets of relationships in which particular objects can be assigned places. The ontology differs from previous taxonomies and vocabularies in endeavoring to represent subclasses in a machine-implementable form permitting deduction. Tim Bemers-Lee's concept of "Semantic Web" (99) has gotten PHILOSOPHY OF HYPERTEXT page 42 attention as a proposal for a very large semantic net based on agreedupon ontologies. It is hard to know what will result from this ambitious concept, and how agreement can be maintained among all parties. What Led to the Schematics Paper Many preliminary thoughts had led to this work. I was deeply philosophical as a boy: at five, in a rowboat, I had an epiphany of the awesome and magnificent complexity of the universe and the difficulty of expressing that complexity. (Discussed in Part 1, "There is Too Much to Say and it Goes in All Directions".) I have been concerned ever since with extremely abstract ideas- meta-ideas and meta-abstractions that other people could not understand. I was searching for deep abstractions underlying life and thought. DESCRIPTION The main issue for me was always the uncertainty and appropriateness of description, not truth or falsity as such. People did not so often lie as shade the truth; news was not so often false as misstated or misleading; different points of view stated things in different ways that were not exactly wrong, but tilted in different ways. Descriptions tended to be inappropriate by degree; deductions involving "truth" and "falsity" sharpened this inappropriately. To describe is to cast a big picture. You may describe a small thing, but you can describe it in a way that has deep implications for your view of the universe- as anyone with opposite beliefs will notice right away. MULTIPLE DESCRIBABILITY PHILOSOPHY OF HYPERTEXT page 43 The center of my thinking was multiple describability. It was clear that anything could be described any number of ways; each description in effect filled the universe, presenting a point of view; each description was in a sense false, because of what it left out; any description was in a sense true, since it could be somehow forced to apply. FORCED MAPPINGS AS INAPPROPRIATE MODELS I noticed early that this was true of the way that words were used to control people, mapping the world with terms which had some applicability, and yet had many strong connotations. For instance, consider the term "good citizenship" as a term applied to children. It means compliance and cooperativeness, but the term chosen has propagandistic and persuasive thrust as well. Another such term (which was pointed out by a commentator on the BBC as I was writing this) is the term "pacification", which is often quite misleading: it has a peaceful sound, but it generally means killing insurgents and destroying their means of support, creating orderliness by violence. Such terms are used in a kind of forced mode, with truth and falsity not so much the issue as "misleading picture". This issue concerned me from an early age. In high school I became aware of a popular philosopher- a philosopher who was not in the philosophical mainstream, and not respected among professional philosophers, and yet popular in the nineteen-fifties- named Alfred Korzybski. He was popularized by three different figures: Stuart Chase, a popular author; John W. Campbell, whom many regard as the dean and founder of modem science fiction, and S.I. Hayakawa,who later had a career in politics. Korzybski is famous for his phrase "the map is not the terrain"- meaning that a description never precisely matches the irregularity and complexity of the world. This is similar to what I have been calling multiple describability. PHILOSOPHY OF HYPERTEXT page 44 Korzybski called his philosophy General Semantics. The main tenet of General Semantics, as I understand it, is that everything would be better if we could specify more exactly what we are referring to, and to be exact in our vagueness as well. (The word "Etc." he saw as a clarifying term, to avoid a precise-seeming generalization when what was meant was some approximate but undetermined continuation of a concept.) Korzybski's writing more or less runs out of steam after this fundamental point is made, but "the map is not the terrain" is an excellent slogan, and partly in honor of Korzybski's influence I later came to call my ideas General Schematics. The vast number of possible descriptions of anything was based on the vast number of properties of anything: virtual infinities of properties meant virtual infinities of descriptions. One of my slogans in school became "Everything has an infinite number of properties," and another was "The only true representation of a thing is the thing itself." I am not sure whether I used these slogans in high school, but I know I did in college. EXCITING FRINGE ACADEMICS At Swarthmore College, I managed to read the writings of a number of interesting fringe academics who all contributed to these ideas. I was fascinated by the proposed extensions oflinguistics by such people as• Edward Bloomfield himself.- a great linguist who, like Korzybski, spoke of infinite numbers of relationships and properties extending in all directions. • Benjamin Lee Whorf, who popularized what is now called "psycholinguistics"- the idea that language determines thought; • Edward T. Hall, who extended the linguistic model to how people act with one another PHILOSOPHY OF HYPERTEXT page 45 • Ray L. Birdwhistell, who extended the linguistic model to gestures (who extended structural linguistics to human gesture and movement ("kinemics")) • Kenneth Pike (who expanded structural linguistics with a grand theory of "tagmemics"). In psychology, a number of exciting approaches caught my eye. I was interested in the holistic approaches of Gestalt psychologists such as Wolfgang Kohler and Kurt Koflka (indeed, I was actually a subject for an experiment by Kohler himself, in his last days). I purchased the pioneering cognitive-strategy book, A Study of Thinking (83) by Bruner, Goodnow and Austin, when it came out in 1956. All of these were antecedents for the 1958 Schematics paper, though I did not manage to include them in a bibliography, which was omitted from the document. THE TRENHOLME CONVERSATION At college, in about 1956, I had a conversation with a fellow student named Russell Trenholme which has stuck in my mind for over forty years, for several reasons: first, what I said was so decisively clear to me at that time; second, what I said still seems correct in the light of current scientific advances; third, Trenholme's reaction was so extreme; fourth, because we were both right; and fifth, because the consequences of what we talked about are very much in the headlines today. I believe I was talking about facial resemblance, and how difficult it was to imagine what genetic mechanisms underneath- what we referred to at that time as only "the chromosomes"- what genetic mechanisms could be so sensitive as to generate subtle facial resemblances within families, resemblances which combined in the most amazing ways- so that a child can resemble both parents, for example. PHILOSOPHY OF HYPERTEXT page 46 I believe Trenholme said something about how science would eventually find the mechanisms by which this operated. (This was an excellent prediction on his part, as it anticipated the cracking of the genetic code and now the human genome- a term which could not have even been understood at that time.) I believe I replied to Trenholme approximately as follows: "Even when the mechanisms are known, the vast complexity of their interacting ramifications will probably go far beyond what can be charted or understood." To which Trenholme replied, "Hey, you're out of your depth!" I found this reply most amusing, because I was certain that I was right. The reason is simple: the complexity of ramifications goes on and on and on. Whatever we find out usually enlarges the boundaries of the unknown. Put two complicated things together and the resulting situation escalates in complexity, potentially without end. For instance• The rules of chess are simple, but chess books are being written every day with new strategies. • Hundreds of millions of people speak English, but not every English poem has been written yet. The relational structures of ramifications spin out in their virtual infinities in all directions. It is now some forty-five years since that conversation, and I believe we were both right. Fundamental mechanisms have been found that elucidate genetics, and the general locations and precise chemical constituents of the human gene structure are now known. However, the ramifications of their interactions is stupefyingly complex and may be unknowable in that complexity. In an interesting echo of that conversation, I PHILOSOPHY OF HYPERTEXT page 47 had dinner in the year 2000 with a well-known molecular biologist friend. I asserted to him (willing to be corrected) that in "solving the genome" we now knew only where the individual program segments start and stop. "Absolutely," he agreed. (I 00). I am not insisting that these things are unknowable, but I think we should be deeply prepared for the possibility that all cannot be known, as the ramifications fan out indefinitely. MODELS AND APPROPRIATENESS: FORCED MAPPINGS My schematic philosophy can best be summed up as: "Shoes are not shaped like feet." Shoes are designed backward from appearance, and feet are forced to fit them, so that human beings generally have misshapen feet that have been pressed into unnatural shape by shoes. This is a simple example of how forced mappings affect and twist our lives. The key cycle is this: we develop models, but then we impose them as false mappings where they do not belong- whether as facts or as rules. (Schematically, facts and rules are identical: they are structures held ID correspondence to the world. I refer to these as schematics and normatics.) This model of multiple describability and mismatch became clearer and clearer to me, first in high school and then in college. In college, majoring in philosophy, I continued thinking about multiple describability and wrongful imposition of models. Abruptly in college, under the influence of structural linguistics, it became clear to me that almost anything could be expressed as discrete relational models. (This was in the mid-fifties, before such structures- now called Knowledge PHILOSOPHY OF HYPERTEXT page 48 Representation Systems, or KRS-became popular.) My course in structural linguistics was an exciting revelation: how the real structure of language was rediscovered in the nineteen-twenties (having been known by scholars in India long ago). I began to wonder intently how this particular kind of abstraction could be extended to the discovery of similarities in other areas. (The chief exponent of this model, and one of the discoverers of the structure of language, was Leonard Bloomfield, who also spoke of infinities of models.) These ideas crystallized in a seminar I took my third year in college with Michael Scriven, a remarkably brilliant and forthright philosophy professor. He provided more information faster, and more swiftly-flowing arguments, than I had ever heard before, and gave me a great sense of inspiration for the project. 1958 SCHEMATICS The 1958 Schematics paper was thus written as my term paper for Michael Scriven's course in Social Philosophy. I meant it to be the foundation of a new field of abstraction and representation. The experience of writing it was one of the most intense I have ever experienced, in an exalted state of excitement and inspiration. The same epiphany I had experienced at the age of five, of the immensity and indescribability of the world, came to me again, but this time with regard to realizing how models and language and thought worked, a way of approaching the great complexity I had envisioned long before. These ideas have never left me, and I have made many notes about them over the years, notes whose interconnectedness defied all conventional organization. PHILOSOPHY OF HYPERTEXT page 49 The interconnectedness of these ideas and their corresponding notes, and these problems of conventional organization, perplexed me deeply; and prepared my mind for yet my third epiphany of interconnectedness two years later. PHILOSOPHY OF HYPERTEXT page 50 Part 3. Why Hypertext and Why Me - - ---- - - -- - - ----- PHILOSOPHY OF HYPERTEXT page 51 Part 3. Why Hypertext and Why Me In the fall of 1960 I had a vast, unified vision of a hypertext future. Hypertext hit me as an idea simple and explosive, with ramifications in every direction, and it led me to a revolutionary intellectual agenda. In its sweep and emotional power this vision resembled the rowboat experience of 1943 and the Schematics paper of 1958. It was similarly huge in scope, similarly sweeping in apparent unity, and similarly impossible to express. I told this vision to a few people, but most could not understand what I was talking about. Call it brashness, call it extreme egotism, but I have always been willing and determined to proceed with my own designs based on my own understandings, which have always been very different from those of other people. The 1960 Epiphany In 1960, in my second year of graduate school (studying sociology), I had a chance to take a course called "Computers for the Social Sciences." It was a good course, which I found thrilling, and as soon as I found out what computers really were- All-Purpose Machines, as von Neumann had called them (but the press did not catch onto that term), I desperately wanted one (although no individual in the world owned a computer at that time). The explosive moment came when I saw that you could hook graphical displays to computers. At once- over a few weeks- I saw that this would be the future of humanity: working at screens, able to read and write and publish from everexpanding new electronic repositories. PHILOSOPHY OF HYPERTEXT page 52 This meant whole new forms of writing, extending documents far beyond the walls of paper. WHY HYPERTEXT I did not choose the word "hypertext" for three years, but hypertext as an idea came to me very quickly, I believe within two months of knowing what computers were about. I already believed that the existing world of writing, publishing, subject division and education were wrong. •Ideas are complex and interconnected (from General Schematics); • Therefore, "subjects" are illusory- slippery, unbounded, undefinable, misleading, fractally inappropriate; • Therefore, Publication and Education are falsely divided into arbitrary subjects; • Publication and Education are governed by shallow people, who oversimplify everything and deny access to unusual points of view; and many other issues. Furthermore, it was easy to imagine interactive text. Branching and interactive media already existed•light-up museum displays that responded to different buttons • records with more than one groove, where you could not predict what would play (later used by Monty Python for the record "Matching Tie and Handkerchief') Thus it became immediately obvious that new forms of writing would have to ___ PHILOSOPHY OF HYPERTEXT page 53 come about. But the question of exactly what these forms of writing would be was- and indeed remains today- an ongoing question that only grows. (I didn't know what to call these ideas; "hypertext" came to me in 1962, and I published it in 1965 (23 ). I first heard strangers use it around 1986.) My Hypertext Agenda Within weeks, I had a personal agenda that was vastly ambitious and totally obvious, based on everything I had thought about before in my lifeGENERALISM AND ACCESS This was the way to help make the world safe for generalists, and make everyone more of a generalist• Increase knowledge and access • Make clear the interconnectedness of everything • Make all ideas clearer and more accessible • Progressively make all the written works of mankind available. Obviously no one can read or know everything, but it can all be made much more accessible. •• New kinds of quotations and reviews would provide better access to the contents of all documents. Every quotation could open to its original context! Every quotation would be a kind of punch-through gateway to the original. And anyone could create such digests, quoting parts of any documents, because we'd be able to handle the copyright problem (Appendix G). •• We would harden the archive, in digital form, against such future dangers as nuclear war; possibly storing it in orbit or further out in deep space. PHILOSOPHY OF HYPERTEXT page 54 EASE OF CLARIFICATION Non-linear writing would make it easy to emphasize the the main points of any document, allowing the reader instantly to draw up details. • Any new organization of the material, in a new version, could at once be made clear through rapid new editing techniques and vivid new visualizations. (Not like what we have so far.) DEPTH OF SCHOLARSIIlP AND ARGUMENT We would be able to see the origin of every quotation. Furthermore, an author could easily make available the citation of every fact without being pedantic or cluttered. (This idea was later called "drill-down".) MAKING TIIE WORLD SAFE FOR SELF-PUBLISHERS With printing unnecessary, it would be possible for anyone, at low cost, to publish to the world. • Alternative and unpopular viewpoints would have an equal voice. • Publishing would be open to everyone- no longer just the anointed few who controlled the current market of paper publishing, filtering it through their often-shallow minds. • Authors could be freed from the tyranny of editors who dumbed down their work, removing its connections and richness. OVERTHROWING THE EDUCATIONAL SYSTEM The educational system, it seemed to me, was a way of destroying minds and rendering them incapable of deep thought. Now we could change this, by overthrowing, in particular- • boundarized education- we could eliminate the false, artificial PHILOSOPHY OF HYPERTEXT page 55 boundaries of "subjects", which make education less interesting and more . . . 1mpnsomng • the Curriculum itself.- the requirement that everything be learned in a certain sequence in a certain amount of time •• we could de-schedule education- making it administratively possible for people to choose the timing and sequence of their own learning. • the lack of chances for initiative in learning •• we could let students choose how to meet the required goals- what approaches, what texts (This is not to be mistaken for the "let kids do their thing" view, which leads to illiteracy; rather, we could let students do their thing toward fulfilling established criteria.) • the domination of secondary education by personalities and attitudesyour "good subjects" (if any) were determined by which of your teachers you did and didn't like •• we could unseat this strange dictatorship of each field by the personality of the instructor, allowing each student access to each subject in accordance with the student's own personality, interests and style. CLEANING UP THE ON-LINE COPYRIGHT ISSUE I knew in 1960 that copyright had always been an issue, and would always be an issue; but with the proper design, I saw that it could be cleared up in the on-line world. • tekkies I'd talked to thought that copyright would be 'no problem'obviously they were dead wrong • I envisioned a special document method that would clean up the issue in the coming on-line world. (Appendix G.) PHILOSOPHY OF HYPERTEXT page 56 DESIGNING A STANDARD It was clear in all of this that I would have to design a new kind of electronic document standard. The structure of electronic documents was going to matter very much. • In any area of design, the different possibilities have different consequences, different generalizations and extensions; and this determines what you can do with the resulting design. This would be especially true for electronic documents. • Unless electronic documents were designed right, great chances could be missed (This turned out to be all too true.) • The referential model should be the center, but the tekkies wouldn't understand it. (This turned out to be all too true.) THE REFERENTIAL MODEL AND ITS ORIGINS My designs started from one particular insight: for two different reasons, it seemed to me that the data structure should point at a pool of content. This is now generally referred to as the Xanadu model (Fig. 3). Whatever is freshly typed should go into a pool, and the document should be a list pointing backward to that original data as the work evolves. I still believe in this model- the referential or indirect model- as the fundamentally right model for electronic media.* * This same idea evolved independently in Hollywood as the EDL, or Edit Decision List. A finished scene in a video production is brought together from a list, of particular shots to be taken from specific rolls of videotape, which the director has chosen carefully. This is the Edit Decision List, and it refers to contents stored elsewhere; this is the Xanadu model. More recently, the same method has been used for film. PHILOSOPHY OF HYPERTEXT page 57 AN AUTHOR-BASED, LITERARY AND CULTURAL DESIGN The Xanadu Document Model A document is delivered as a LIST of contents-- a virtual file. This is the fundamental form. VIRTUAL FILE Any new content goes into the pool, not into the virtua file. D AVAILABLE CONTENTS - an ever-growing addressable pool, or indexable carpet [ Fig. 3. The Xanadu Model. A document is maintained and delivered as a list of content. ] PHILOSOPHY OF HYPERTEXT page 58 I reasoned that the freshly-typed pool of content should be continually backed up, and different documents should all be represented by different series of pointers to this content. As the docwnent evolved, revised versions of this listing would point at the separate parts that had not changed, in whatever sequence they now formed. I think I originally had two reasons for this referential model. (This is from memory; I hope to verify it some day from my notes of that time.) Saftety Safety of contents- so that the user's input would not be lost by some breakdown or crash. Con Controlling the re-use of notes in many places (and, by extension, all other source content). I personally needed referential structure for my note-taking system. As mentioned earlier, I was taking notes on many subjects- by now I had thousands of file cards- and they were in many overlapping categories [as in Fig. I]. This was an enormous problem- if I used a note in in one project I might want to retire it from other projects. In principle I could make a copy of a card for each project, and retire it from the other projects if one were completed. However, there were two problems with this idea: first, there existed no copiers at that time (as we now know them); and second, the nuisance and overhead of removing the note from the other topics would be unthinkable using hand methods. However, ifthe notes were kept in a common pool [as in Fig.I] and then virtually included in some specific document [as in Fig.2], the problem would be solved: the system could immediately show me where else a note was virtually included, and I could decide whether to delete it from those other locations. PHILOSOPHY OF HYPERTEXT page 59 REFERENTIAL STRUCTURE RESTATED Referential, or indirect, structure means simply this: A document (or version) is a list pointing to contents (as in Fig 3). Through the referential method, it becomes simple to keep track of the use of material between two documents-- by finding overlapping pointers in two lists. (This is crucially different from just having the computer compare text of the two versions, since it matters which is the original.) Shared portions may be automatically shown side by side. This structure makes all contents specially sharable- so that different people may see exactly what parts are shared, even as they continually change their documents. This can be important in many work situations. The referential model allows what we call transclusion, the inclusion of content across a document boundary; more broadly, transclusion is knowing, principled re-use of content. The word "transclusion" will be used frequently hereafter. Such structure, and what can be done with it, has been the heart of many Xanadu designs. (One full version is explained in detail in Appendix F.) LINKS In addition, links could connect all these new writings. I have always thought of links as an extension of literature. Their structure would extend the footnote, the parenthetical note, the citation, the marginal note, the parallel element (marginal gloss or caption)- the ways in which text has always tried to break free. PHILOSOPHY OF HYPERTEXT page 60 DEEP DESIGN QUESTION ONE: HOW HANDLE UNITS? Documents typically have various units, parts and pieces, and these interpenetrate the content of a document and vice versa. How do such units fit with the referential model? DEEP DESIGN QUESTION TWO: HOW COMBINE LINKS Willi TRANSCLUSION? Given this model of indirect referential structure, which potentially makes all content sharable- how would links fit in? This was another design issue that was to remain a problem for decades. (The xu88 structure would eventually decide this; see Walktbrough of the Xanalogical Model, Appendix F.) A DEEP HIGH-PERFORMANCE EDITOR By the versioning methods mentioned above, I resolved to build an editing system based on this referential method, keeping track of all versions- in principle of every change- and most important, keeping track of the re-use of content so the author could optimize the placement of each item. This meant never having to lose information! The naturally destructive processes of writing, with all its breaking of coIUlection, could be tamed, and all coIUlections could be maintained. This meant an editing system that would maintain all coIUlections- links and transclusions- at once, even as a user or author changed the contents. (Even today, such an editing system does not exist.) As these ideas crystallized in the fall of 1960, I at once began designing the highperformance editing system I wanted. I envisioned fast-moving text and PHILOSOPHY OF HYPERTEXT page 61 the ability to do a dozen rearranges per minute- impossible even today. A NEW FORM OF WORK This could also mean a new form of work- without boundaries or deadlines! • Every new item would be put directly into the whole, placed in all the places it belonged (with each use traceable among all its different contexts-the green lines in Fig.2); • no changes would be dangerous, since you could backtrack to previous versions easily; • every operation could in principle modify the whole structure with bold changes, while keeping all the previous minute connections; •multiple uses of the same content could be easily managed and kept track of; • no connections would be lost due to false emphases or boundaries (as in having to cut down writings to fit a particular plan, or space limitations, or editorial preference); • all notes and text would remain knitted together in one single loaf; • publishable "articles" could be sliced off from time to time, locally severing the connections among individual items, but not disturbing their relation to the true work- the whole structure. (Again, refer to Figure 2.) This meant a total change in the nature and texture of work. I foresaw an entire form of work for my own career. Like Marcel Proust, one of my heroes, I would be able to write continuously on a single unified whole structure for my entire working life. This meant not having to narrow down, not needing to let go of the whole structure of the work, not needing to reduce an overall concept in order to complete some subset. PHILOSOPHY OF HYPERTEXT page 62 It was what every author needed, but what I needed especially, with my broad interests and ambitions- to keep my work together as a single undivided unit. I call this idea of a totally unified, indivisible work "Irv", Americanizing the French word oeuvre. My Irv was a good plan. Unfortunately, because I have been unable to implement it, and because the tekkies have given us broken tools for writing, all my notes remain scattered and still trapped on paper. (I have spent a great deal of my working life trying to retain the unity of my work on the Irv model, for such time as such a decent system should be available; but it is hard for people to understand my working methods because this aspect is invisible.) EXPANDING IRV TO A LITERATURE The Irv concept can be enlarged to a whole literature. Whatever could work for an individual could of course be generalized to an entire literature. This meant a fundamentally new and comprehensive genre which would subsume all writing. These same referential structures, extrapolated to everyone's personal computers (that we surely would have), meant a new hypertext literature that would subsume and perfect all the writings of the past, as well as leading the way into the future. And, of course, text was just the simplest case, in its fluidity and rearrangeability the model for audio, video and whatever was to come. If experience-recording is ever invented (as in Douglas Trumbull's 1983 film "Brainstorms"), it would have the same properties. The Irv model of piecewise change could be extended to this whole electronic PHILOSOPHY OF HYPERTEXT page 63 literature: small changes could be sent by each author one at a time, to his or her published on-line document, without having to do large-scale version changes. (This has later been called the microversioning concept.) The global concept could thus extend to deeply-connected hypertext with deep content connections (as in Appendix F) and microversion publishing. Some method of royalty payment for authors was also called for; but this could be proportional to use- not just buying a whole document, but, in effect, proportional to using portions of documents. (See Appendix G.) HYPERTEXT AND FILING This global concept of hypertext extended to filing as well, smce electronic documents would not be filed in cabinets or separated out. No longer would we be crippled by the filing concepts of paper literature, with their absurd categorical divisions! (Note that the opposite is still the case in the computer field, with hierarchical directories still the order of the day- pun intended.) Those in the conventional computer paradigm still have no way of understanding these ideas, because deep overlap (fig. 1) and interpenetration (fig. 2 being a simple example) do not figure in their thinking. MY PERSONAL-COMPUTER AGENDA Hypertext was the less obvious part of the vision. What was totally obvious was that personal computing would take over the world. The personal computer would bring us- • A media machine. In addition to the fun of computers, the computer would be a media machine with everyone an auteur. PHILOSOPHY OF HYPERTEXT page 64 • The Simplification of Life. The computer would be an easy tool for everything. Tomorrow's software, which I intended to design, would make it•• simple to keep track of everything •• simple to maintain notes and smoothly organize them into writings (The reader may suppose that these things have been achieved. No way, in my view. Instead we have a nightmare.) Why I Thought I Could Do It Every line in my life had led to this point: every line of thought and experience and innovation had led to this one new idea. From the same temerity with which I had created the 1958 Schematics paper, this vision was now generalized to a solution to all the things I disliked about our society. In about three months' frenetic thinking, I saw that this meant a new world of information, a new world of possibility, a new renaissance- and a particular design for it. I thought I could do it and had to, because I thought nobody else could. (I had already had negative experiences with engineers and their attitudes in college, and knew there would be a conflict between their attitudes and mine.) All of these ideas were clear to me by the end of 1960. (But, of course, not all the details- particularly how links and commonalities were to work together. And indeed those details have varied over the years, as many designs were floated and discussed (see Principal Xanadu Designs, Appendix K).) PHILOSOPHY OF HYPERTEXT page 65 I still saw myself as a Combat Intellectual, designer and innovator. My experience with writing and magazines and photography and theater and film had given me a sure sense that I could do new things with any media. And here was the all-embracing new medium! I saw clearly the problems of writing at computers and what authors would need; the issues of copyright that would emerge; and the all-embracing solution I saw in terms of a particular structure which would not be obvious to the tekk.ies. This is why I had such a clear sense of certainty that I knew what to do and could do it better than anyone else. Where the Ideas Came From These ideas did not occur magically in my head from nowhere. They came from all the things I had read and thought about in my youth. I.FUTURISM From early boyhood I had a deeply Futuristic point of view: I believed in a great new future, where everything would be clean and chromium-plated. When I was ten, I tried to persuade my grandfather to buy the new Tucker Torpedo automobile, which was the modernist dream- it had three headlights and would supposedly go a hundred miles an hour. When I was thirteen, I expected that I would soon live in my own personal earth satellite. (This was eight years before Sputnik.) And when I got out of college, I thought I would make a contribution to some new field that had not been invented yet. 2. DESIGN HEROES I have always believed (and still believe) that almost everything in our lives is designed wrong. (More dangerously, I have always believed that everything PHILOSOPHY OF HYPERTEXT page 66 would be better if only it were completely different, and this has led to many disappointing experiments.) Designers were always my special heroes, starting when l was a small boy with Leonardo daVinci and Frank Lloyd Wright. At the age of nine or ten I was very taken with Wright's "Fallingwater" building, which has a stream running through it, a radical and beautiful design. Inspired by this, I designed in my mind, and wrote up for the school paper, a house that would be completely under water. I read about Buckminster Fuller when I was eleven, in 1949, long before he became well-known. He was a designer who said that we must create radically new designs to live on a planet with dwindling resources. Following Bucky Fuller's point of view, I believed that everything would be far better if only it could be redesigned completely. Many other designers were my heroes: Raymond Loewy, who designed the 1947 Studebaker and an earlier, muscular-looking locomotive engine that practically defined the term "streamlined train"; Bernard Rudofsky, who mocked our systems of clothing; Charles Eames and his wife Ray, who for many years were principal designers for IBM. In high school I was bemused- and amused- by the designs of Peter Schlumbohm, who designed the Chemex coffeemaker, the Chemex cigaretteholder, and the Chemex electric fan (which cleaned the air considerably). Each of these used a disk of filter paper every day; Schlumbohm loved filter paper. It was a wonderful example of an idea filling the universe. I thought this man's obsession was quite charming. This to me has always been the model of an idea expanding to fill the universe. PHILOSOPHY OF HYPERTEXT page 67 There existed two well-known electronic musical instruments when I was a boy in the nineteen-forties. (Young people today might be surprised to learn that.) One was the Hammond Organ in high school I learned how that worked; the other was the Thereminvox, and in college I learned how that worked. Both were extremely elegant, clean designs inside, brilliant and minimalistic. (Discussed further on p.78.) In college I too became an actual designer. I published my own magazine and posters, and experimented with printing, spending most of my money at the print shop in town. A very special moment for me was one day at the print shop my second year of college. I had designed a strangely-shaped magazine, and printed it together with my friend the printer. Excitedly I assembled the first copy, and the printer was astonished. I thought he had understood and approved of what I was doing; in fact I had done it entirely by myself. 3. SOPHISTICATION My very unusual background centered, I realize now, around sophistication. I lived with my grandparents in Greenwich Village, a very sophisticated neighborhood in New York City. I was unusually familiar (for a young boy) with history, literature, the Broadway stage, the worlds of alternative publishing and little intellectual magazines, and English movies. When I graduated from high school, the three pillars of my interest were the New Yorker magazine (the cornerstone of sophistication in America); the Museum of Modem Art, where I would go to see classical movies; and the Rand Corporation, the center of thinking about issues of nuclear war- where I thought I might someday work. 4. ALIENATION I also grew up increasingly alienated. I think it began with difficulties of PHILOSOPHY OF HYPERTEXT page 68 communication. I always felt that I understood things other people did not, and wanted to explain it to them as a unified big picture. Thus I always had too much to say. And in the meantime, other people seemed to me more and more simpleminded. 5. .MEDIA I had a very unusual media background. My parents, who divorced when I was born, both became successful in the theater. I saw them rarely but participated vicariously, cheering them on. My mother was a stage star by the time I was seven, and later a famous movie actress. Meanwhile, my young father, while he was training other pilots during World War II, somehow managed to write a play that was a hit on Broadway. After the war he then got in on the ground floor of television and became one of the top directors in what is now called the "golden age" of television- black and white, live, no videotape. After that he went to Hollywood and directed some twenty films. What was most significant for me from all this was the privilege of sitting behind my father in control rooms at NBC and CBS- at the age of eleven and twelvewatching a new medium come into existence, and seeing men working in a new art form sitting at screens in the control rooms. From boyhood I followed all media avidly: in grade school I loved the interactive media of that time (exhibits that lit up when you pushed various buttons), and in high school coming to understand the layering of media- how movie special effects were put together out of multiple layers of images and moving blackout masks; and how the new medium of tape recording was becoming as artists like Les Paul and Mary Ford, Louis and Bebe Barron created - - - - --- - - - - - ----- - - - PHILOSOPHY OF HYPERTEXT page 69 recordings of recordings that multiplied the sound magically. The effect was not like the cause. Knowing the desired effect, an artist had to design backward through the strange layering it would take to get the desired effect. By the time I got out of college, my media experience was quite extensive. I had been an actor on TV and the professional stage; I had created my own magazine and newsletter, written a prize-winning play, written and directed what I believe was the first rock musical,* produced a long-playing record and a 30-minute film, and published a small book. * "Anything & Everything", performed at Swarthmore College, November 22-23, 1957. I also had copyright certificates for some of this work, registered with the government, and I firmly believed in copyright as one of the few ways that creative individuals could advance in the world. However, I had also experienced the frustration of negotiating to republish copyrighted material, so I also knew the difficulties that copyright created. In 1959, finishing college, I looked forward to a career across numerous media, and I especially expected to direct films. MY 1959 CREDO, IDEALISTIC AND CYNICAL By the time I graduated from college in 1959, besides having an unusual outlook, I was very bitter and alienated. I thought most conventional people were extremely shallow and had their heads in the sand. I was concerned about the threat of nuclear war; the growing pollution of the planet; the loss of native cultures and the growing sameness of modem society. I was disgusted at what I saw as the ideals of the middle class- fancy houses, PHILOSOPHY OF HYPERTEXT page 70 showoff clothing, and working frantically to save up in order to prolong their final illness. Above all I was infuriated at the difficulty of reaching people with new ideas. Conventional publishing was managed by shallow people with conventional ideas; I believed that self-publishing was the only solution, but this was very difficult because of the high cost. I was also very concerned about the preservation of history and knowledgeespecially the parts that librarians ignored, which I felt might tum out to be more important. Education in its existing form, and inflicted by the present means, seemed to me hopeless, for it trapped and bored its victims, made everything uninteresting, and perpetuated conventional thinking in every area. And universities, which gave the appearance of idealism, were really all about money. Something entirely different had to be done for education. There had to be new forms of media, a complete change for education- smartening up the public and freeing them from cliched ways of thinking and absurd shallow categories. I believed that only very radical change could improve the world; and conventional political solutions (including socialism and Communism as "conventional") went noplace. I wanted a completely different world- a sexual revolution, new forms of marriage, new forms of society. I wanted a complete cultural revolution, somehow to spread sophistication and wider understanding to a much broader base. The designers and futurists I admired, especially Bucky Fuller and John W. Campbell, told me that a completely different world could be created- along lines of possibility that most people simply could not imagine. · -- - -- - - - -- - - - --- - -- PHILOSOPHY OF HYPERTEXT page 71 Also following Bucky Fuller, I sought to be a serious generalist, and that meant being a serious intellectual- unconcerned to look the way people expected an intellectual to look, with tweed jacket and pipe. And to me this meant taking far more notes than anyone had before, in order to tie things together in new ways. The problem was how to keep track of all the notes, given the total overlap of all subjects (Fig. I). My great quantity of unsortable notes was already a big problem. Still, I was up for the fight. I saw myself as a combat intellectual, generalist, designer and innovator, and I was going to change the world as many ways as I possibly could. That was my thinking before I took the computer course a year later. That is why I was so primed. How I Was Right and Wrong The reason for presenting all this background- my concerns and discontents with writing, media, education, publication and conventional thinking- has that I was uniquely qualified to been to explain my certainty in December of 1960 understand all this new world of electronic media possibilities and their right designs; that no one in the world understood it as well as I did, and that I was uniquely qualified to design the future in line with the great traditions of human culture and the complexity of human thought and creativity. In this I believe I was remarkably correct, at least as of that moment. I also believed that because of this special understanding, it should not be too difficult to persuade people to back my designs. wrong. In this I was spectacularly PHILOSOPHY OF HYPERTEXT page 72 Part 4. The Philosophy Part ----- PHILOSOPHY OF HYPERTEXT page 73 Part 4. The Philosophy Part: GENERAL SCHEMATICS AND THE XANADU DESIGN Calling this section philosophy part" is a joke, since there can never really "the be a separate "philosophy part". Philosophy permeates everything. \Vhat I will discuss in this section is where philosophy meets software. meeting point. There is no one Ideas and principles- the force-fields of philosophy- cut across everything. General Schematics led at once to a number of concerns which were in my mind from the beginning of my computer work in 1960 CONSTRUCTS General Schematics was about constructs- relational structures created intentionally or not. And what computers deal with is artificial constructs. (Even when the computer imitates some "real" structure, like paper, it must be artificially made to do so.) It was clear that the computer was a tabula rasa, a construction set into which we could build any structure. What structures would best carry forth the world of writing and literature into this new world of possibility? Constructs were relational structures. The constructs of this new world would be mapping these relational structures to the content and relationships of documents. From this 1960 beginning I believed we should eliminate paper and books- but PHILOSOPHY OF HYPERTEXT page 74 not by imitating paper. We had to abstract the forms of information that paper and books had already contained. There are many imaginable forms of hypertext. But how design this new universe? There were far too many options and factors and pieces to the puzzle. There are many hypertext issues to reconcile, many relationships to be represented. The question is how. Are the issues technical or literary? If they are technical, these are issues best left to the technical community. But if they are really literary, as I have always believed, leaving such decisions to the technical community could be disastrous- as I believe it has been. If the issues are literary, we must deal with conceptual structure for the user, and create a sensible world for users' minds and work. Anything can be made the center of hypertext: literary structure for the user, delivery formats, data structures. What center is chosen determines what kind of a literature it will be. This is the issue we will consider now. PHILOSOPHY OF HYPERTEXT page 75 A STRACTION, SU STRUCTION AND DESIGN: THE SELECTION OF CENTRAL CONCEPTS AND THEIR DETAILS First let me introduce the notion of substruction as a particular kind of abstraction. In abstraction of the ordinary kind, we are confronted with a number of cases in which we seek to find common properties- for instance, what is it that birds have in common, or highway accidents? And what is it that languages have in common? This kind of abstraction seeks a commonality of/act, some properties of the real world which the specific instances have in common. Abstraction is crucial to both learning and science. To survive as living creatures we must learn continually, which means finding out which parts of our environment are significant, and what are the best responses to them. This is an abstraction of payoff. The scientist abstracts from the visible and known to generalities about them which are less visible; Newton's laws of motion and Maxwell's equations of light are both abstractions which summarize and codify many examples and observations. These are abstractions of factual generalization. But abstractions of design are somewhat different. In computer science, the term "abstraction" is often used for seeking software designs and constructs that will cover a number of cases. It is this kind of abstraction I am focusing on. What I am talking about is a different form of abstraction because it is not just about payoff, or about fact or expressing reality. I would like to introduce another term for this special case of design abstraction. In the kind of abstraction I want to call "substruction," fact is not an issue; we are deciding a design that will cover the significant cases. This is more creative PHILOSOPHY OF HYPERTEXT page 76 and much more arbitrary. By substruction I mean the selection and detailing of central concepts of a design- central concepts from which the desired results can be built. This means selecting and deciding the main ideas, concepts, relations and boundaries, and working out their details. Ideally this means working out details in a way that provides conceptual unification. Substructing in this sense is invention and decision, rather than delineating realms of fact, as in science. (Note that substruction may also occur in the process of writing, since writing is design.) Substruction, like most abstraction, is compressive, because it begins with a range of concerns- in this case, desired functionality- and compresses the necessary solution to what may be a small set of elements, just as abstraction does. It is also a kind of unification and concentration of what we select as the central ideas. Thus we seek the best unification and the best central concepts. This can be very difficult. Why not just say "design abstraction"? Because we are not merely discovering common features; we are creating something underneath the concepts we start with, and the details of this new level become conceptually significant. --- - ·- - - - PHILOSOPHY OF HYPERTEXT page 77 EXAMPLES: E-MAIL VERSUS PAC-MAN Consider the e-mail function. While the main function of e-mail was obviousdelivering text messages- there were many possible ways to do it, and a number of ways were tried. This was the process of substructing the e-mail function. Today's e-mail solution ended up with a lot of disparate components. In today's e-mail environment, the principal components are the Sendmail program (or equivalent), the Pop server, the SMTP server, the user's e-mail reader program, and the mailbox format used by the user's email reader program. The design questions of what to show in the email header were highly political-meaning that many disagreements were possible. Should cc's be allowed? Should blind cc's be allowed? "Priority" was allowed but has fallen into disuse. "In-reply-to" has become a key aspect of usage. (Many other data fields could have been considered, and may have been; this history has yet to be written.) These political issues were resolved by the standard as it evolved. Pac-Man is of theoretical On the other hand, consider the game of Pac-Man. interest because of the elegant substruction that was chosen by the designer.* The game of Pac-Man may be thought of as two games ingeniously combineda maze game (where the player must maneuver through a complete set of hallways to advance to the next level) and a chase game (where the player must escape monsters- but can sometimes chase them). * I have been trying to get in touch in Japan with the designer of Pac-Man to discuss this, so far without success. It is my understanding that he presently works at Namco. The final design of Pac-Man may be thought of as the answer to the design problem of combining these two games. We may phrase it as: "How can this game be designed so that the player seeks to go through an entire maze while PHILOSOPHY OF HYPERTEXT page 78 being chased by monsters, but also gets a chance to chase the monsters himself?" The Pac-Man design solution, which was completely non-obvious, was to combine both features- the maze and turning the monsters- under a single idea: the eating of dots. The player's on-screen presence, or avatar, rushes around the screen eating dots; and when the avatar eats a special dot, the monsters become vulnerable for a little while. This invention of dot-eating was not trivial. It simplified the game to only one user activity- fast steering around the maze- and yet provided a strange conceptual unification that people enjoyed. We see here an enormous difference between email and Pac-Man. In the e-mail case, the substructive solution was a large number of disparate components spread across the network. In the Pac-Man case, the substructive solution was one extremely simple mechanism that simplified the user's experience. I am not arguing that e-mail could have been as simple as Pac-Man, though it is conceivable that if it had evolved under other circumstances in the hands of other designers it might have become so. Let me give three other examples of substruction from the world of electronics. Substructing radio. The invention of the superhetereodyne circuit made it possible to tune into radio circuits with only one tuning circuit instead of the previous two. This made radio practical as a mass medium. Substructing electronic music: Leon Theremin. The inventor Leon Theremin in 1920 invented a musical instrument of great simplicity. It had two high- frequency oscillators which together produced an audible beat frequency. The PHILOSOPHY OF HYPERTEXT page: 79 player changed the pitch by moving a hand near it, effectively altering one of the inaudible frequencies by the inductive presence of the hand in the circuit. This was an extremely elegant and simple method of producing music- and it had a most interesting sound. Substructing electronic music: Lauren Hammond. A few years later, the inventor Lauren Hammond substructed electronic music in a different way: he rotated polygonal disks next to inductive coils and ran the result through a loudspeaker. This produced a remarkably sweet tone, which became the heart of the Hammond organ. (He had originally been a maker of clocks, but after losing a court case, needed to find another use for the wheels (101... In three of these cases- Pac-Man, the superheterodyne radio circuit, and Theremin's instrument- a simplified design was found which covered a selected domain of desired operation while minimizing components and structure. In Hammond's case, he transposed an existing elegant substructure to another domain. This kind of elegant design speaks for itself. Besides the fact that such elegant and minimalist design is aesthetically desirable, some argue that it is also more efficient and more reliable. Mark Miller you go for efficiency and speed, you end up not getting puts it this way: "If them. If you go for elegance, you get both." PHILOSOPHY OF HYPERTEXT page 80 SU STRUCTING THE UNIVERSE OF DOCUMENTS With this background, we may finally approach the fundamental question: " hat is the right design for electronic documents?" But we must broaden that. For we have at last arrived at the promised thesis question, "What is the correct design of electronic literature, to help us escape from the prison of paper?" In other words, how shall we substruct literature? The question of "what substruction?" is the entire great question of hypertext design and implementation. The answer is: the right substruction is what best represents and extends the relations already present in literature - previous documents and their contents. From the virtual infinity of previous literature and all its properties, what shall we abstract that best suits the human mind? (Fig.4) Basically there are two answers. The The conventional answer, HTML and the World Wide Web, we will deal with briefly near the end, as a trivial and degenerate case which has been given great status by its success. The other answer is: a much deeper kind of hypertext, with richer connections and content that knows its origins. For the rest of this thesis, or most of it, we will deal with these issues-- which I believe are the important relations. - - - - - -- -- PHILOSOPHY OF HYPERTEXT page 81 [Fig. 4. The Substruction of Literature.] SU STRUCTING LITERATURE D CUMENT AND IDEA-REPESENTATIONSOF THE PAST Intercomparison, selection of commonalities Substruction: Crystallization of chosen SUCH AS: Links by anyone Rearrangement Version and item management for authors Quotation Anthologies Payment to Rightsholders Key Xanadu Substructions Transpointing windows Links to content Knowable re-use (transclusion) Referential document model Version management by referential editing Rights management by referential distribution PHILOSOPHY OF HYPERTEXT page 82 (The reader who is interested in tracing my thoughts during the early period may want to look at Appendix C, "Hypertext Notes", which represented my thoughts as of 1967.) As I said earlier, substruction is the selection and detailing of central concepts. These details affect all aspects of the result. The chosen concepts, with their exact details, become paradigm kernels which expand to filigree the universe. lntercomparison and commonality were at the center of General Schematics. The General Schematics work stressed intercomparison as a way of understanding: helping us to see differences and similarities, and helping us to create new generalizations. Hypertext should help with this mental process. According to this view, a principal function of hypertext should be tracking complex parallelismsmaking visible the various parallelisms that exist everywhere. The hypertext design issue is an issue of meta-abstraction- how do we create an environment for supporting abstraction? To my thinking, commonality and intercomparison must be at the center of a hypertext system: the presentation of corresponding parts and the tracking of commonality. This was my thinking from the beginning of my computer work in 1960 INTERCOMP ARISON OF DOCUMENTS A particular example of what we need for abstraction is the intercomparison of documents. Authors and readers must be able explicitly to intercompare documents, and to annotate their corresponding parts. Intercomparison is fundamental to all thought: it is how we come to understand in every field what things are the same and what things are different. -- -- PHILOSOPHY OF HYPERTEXT page 83 So, above all, we needed ways to intercompare documents side by side. REPRESENT A TING THE RELATIONSHIPS OF THOUGHT Authors and readers must be able easily to annotate documents as one would write in a book- highlighting, connecting, selecting portions which may be tracked by bookmarking. Authors should not have to represent everything in text or sentences. There is a continuum of relationships of thought that we have tried to represent on paper, from simple text to diagrams; all of these should be comprehended in the new literature. What is necessary is some way to combine diagrams with the representation of text and other schematic relations. A good slogan might be: "If you can think it, there must be a way to represent it." VERSION MANAGEMENT ORLD- IDE Referential methods make it simple in principle to handle detailed version management, as demonstrated recently in the OSMIC project (48). But the real issue has always been how to unify the ever-changing versions throughout the world. This is a cosmic coordination problem. (Gregory and Miller's solution in is extremely the xu88 model, discussed in the walkthrough of Appendix F, ingenious.) STANDARDI ATION Above all was the standardization issue. This hardly seemed a problem in 1960, since it seemed certain we would get there first. Now let us consider it from the user side. These political issues were resolved by the standard as it evolved. PHILOSOPHY OF HYPERTEXT page 84 WHAT WILL MAKE SENSE TO THE USER A crucial issue is the question of what forms of hypertext will make sense to users. The issue is to find substructions which are simple but powerful- as in the previous examples of substruction. We have agreed there will need to be intercomparison and parallelism. But what structures will there be? What will be seen and intercompared? It is certain that some sort of "page" will be an option, but not the only one. And the question is what other options and structures there will be. If we are not mimicking paper we can consider all sorts of possibilities, such as stretchtext (Hypertext Note 9, Appendix C) or wormholes (more recently called tunneling text) (6, pp.46-7). WHAT ARE THE SIGNIFICANT PARTS AND UNITS? So many units are possible and imaginable. What should be stabilized? The question is what levels and what types. The point is to have the most mental leverage for the user from the concepts he has to learn. The document? the link? the item within a document? Which are addressable units? (For example, an HTML link cannot itself be referred to or addressed.) SUBSTRUCTION CHALLENGES To substruct chunk hypertext is trivial; to substruct Stretchtext is harder; parallel versions more so; to manage versions world-wide is much harder .. The achievement of Gregory and Miller in the xu88 design (discussed on pages 101 to 104) represents a unique design substruction for handling all of these. PHILOSOPHY OF HYPERTEXT page 85 ITEMS VERSUS TEXT, DISCRETE VERSUS CONTINUOUS We may think of items and notes- the kind writing starts with, as in Fig. 1- as being discrete units. But flowing prose is created when the notes are melded together and merged into actual sentences. The resulting passages still correspond to the original note. How is such correspondence to be managed? This distinction presents a fundamental design paradox. How are the two levels of text- the item and the text passage corresponding to it- to be represented, in a way that maintains that correspondence between the item and the flowing prose? PHILOSOPHY OF HYPERTEXT page 86 WHAT ARE THE SIGNIFICANT MECHANISMS? So far we've talked about the conceptual structure for the user, which is the most important aspect. But what about mechanisms? There is a vast number of possible mechanisms, any of which can in principle be used. Servers, protocols, data structures with and without embedment offer an enormous variety of possibilities. For small-scale systems with one-way links, like HyperCard, any method will do. A large-scale system with one-way links can be done, as the Web has shown, with trivial embedded structure. For the two-way links and transclusion management we have been talking about, however, there are basically two approaches. The principal distinction is between a table approach and a universal addressing method. Various Xanadu designs over the years have used both. The two most sophisticated Xanadu designs, xu88 and xu92, have both used a large-scale address space, or universal addressing method, as a common level of reference. And their link structures have been designed for management by global systems of distributed tables. PHILOSOPHY OF HYPERTEXT page 87 Part 5. The Structure Wars of Electronic Literature - - - -- - - - - - - - PHILOSOPHY OF HYPERTEXT page 88 Part 5. THE STRUCTURE WARS OF ELECTRONIC LITERATURE Designs have consequences. Early decisions determine what you can do later. The design of software determines what people can do with it in many deep ways. I did not anticipate- indeed, nobody could have known- how different would be the ideas for electronic literature, and how ferociously these ideas would be held by different parties. Personal ambitions, corporate maneuvers, and strange ideologies have created a battlefield of alternative structures and designs. When I started this hypertext crusade, probably no one else except Doug Engelhart had designs in the area of electronic documents, and I had not heard of him. Today, major players includeThe World Wide Web Consortium, with their XML, and "Semantic eb" initiative (hierarchy and linearization) Microsoft, with their seeming approach to world domination RealNetworks, with an independent plan for streaming services Macromedia, with its own agenda AOL Time Warner, with a vast network and a lot of content Sony, with a huge base of success in recording equipment and a large amount of content The FBI, which is concerned to read e-mail The NSA, whose mandate is to watch foreign message traffic Congress, especially concerned about pornography, taxation and free PHILOSOPHY OF HYPERTEXT page 89 speech issues Libraries, intent to keep their niche in the world Librarians, applying their notions of document categorization and structure The ACM, which owns the official document database of computer science Writers' associations and unions concerned with authors' rights, such as the National Writer's Union, Writer's Guild of America East, Writer's Writers' Guild, Guild of America West, Writer's Guild of Canada, Screen Authors' League, Television Writers of America, Internet Writers and Artists Guild, National Conference of Editorial Writers, Dramatists' Guild, to mention a few. Music publishers, now feeling the impact of free MP3 distribution. Academic publishers, who presently sell very expensive paper journals but see their market eroding under pressure from the Web Pro ect Xanadu, with a vision of simplicity, clarity, accessibility of content separate from structure, and a clear copyright solution with new formats and standards. Where this will all lead is completely unclear. But each of these parties has a different agenda, and the war is continuous on many fronts. Each party has technical proposals in line with its particular agenda. I will deal with the last, Pro ect Xanadu, before the others. --- . PHILOSOPHY OF HYPERTEXT page 90 Part Sa. THE PRIVATE STRUGGLE: PERFECTING THE XANADU MODEL In various implementations, Project Xanadu has endeavored to complete its original objectives, and the work continues to this day. .bave taken years to work out The fundamental design issues laid out in 1960 well. There have been several substructions in the different main designs. THE XANADU SU STRUCTIONS There have been three main Xanadu substructionsthe fundamental referential model, pointing to content rather than including it directly. parallel documents and visualization the symmetry of links and transclusions. We will discuss first parallel documents, then the symmetry of links and transclusions. PARALLEL DOCUMENTS, INCLUDING PARALLEL VERSIONS Let us consider, as our selection-set to be substructed, a fundamental document type: parallel documents (17). This concept compresses and substructsINTERCO ARISON VERSIONS MARGINAL NOTES PHILOSOPHY OF HYPERTEXT page 91 DIFFERENT DOCUMENTS PARALLEL DOCUMENTS Parallel documents are the selection-set we want to substruct. We consider parallel documents supremely important (where other hypertext researchers seem not to). Parallel documents are everywhere, but are not generally acknowledged. There are relatively few explicitly parallel documents (like Torn Stoppardplay s "Rosencrantz and Guildenstem Are Dead", which is explicitly parallel to "Hamlet"showing events that occur offstage in "Hamlet", and vice versa). But implicitly parallel documents are everywhere- the parallelism of commentaries, the parallelism of long and short versions of reports, the parallelism of translations, the parallelism of holy books ( 18). It is vital that we be able to see this parallelism of documents and to intercompare and work with their side-by-side connection. 1. VISUAL SU STRUCTION: TRANSPOINTING INDO S This leads to the basic interface of our model, parallel visualization. To clarify our model we will look at pictures. Side-by-side connected comparison of parallel documents on the computer screen has always been Xanadu's fundamental visualization, first published in our 1965 paper (23 ). PHILOSOPHY OF HYPERTEXT page 92 ... [Fig. 5. Side-by-side connection as schematically illustrated in 1965 (author's original diagram for (23), later redrawn for publication).) Figure 5 shows what we believe is the first published visualization (23) of parallel screen documents, in abstract form. This was meant to represent side-by-side items in columns. Each rectangle indicates a sequence of paragraph-like content items (like today's "mailbox" files, to be viewed as a column on the screen). Braided lines in the illustration indicate that some of the items to be viewed in one column are to be shown as identical to items in the other column (transcluded), while dotted lines indicate that they are to be shown as only linked. We published this proposal for side-by-side screen connection as a more detailed visualization in finer grain in 1972 (8) (figs 6, 7). PHILOSOPHY OF HYPERTEXT page 93 [Fig. 6. Mockup oftranspointing windows, 1972. ] [ Fig.7. Same mockup. closeup view.] In figs. 6 and 7, the actual contents in one window are shown cross-connected to actual contents in another window. Naturally, these connections must remain connected to the contents no matter how the windows scroll or move around. We now call this visualization "transpointing windows" (15). These early photographs show only a cardboard-and-celluloid mockup of transpointing windows. PHILOSOPHY OF HYPERTEXT page 94 In 1976, in the Xanadu User's Manual (Appendix E), we published transpointing windows as mapped to a character display (Fig. 8). Fig. 8: Transpointing windows mapped to a character display, 1976 (Appendix E) We believe that the first functioning interactive screen implementation of transpointing windows was created by John Walker at Autodesk in 1988, supporting parallel screen connection and scrolling, served from our xu88 software. Unfortunately no pictures of Walker's system are available. A different implementation of transpointing windows, a demo running under Windows 95, was produced by the author and Ian Heath in 1998 (53, Fig 9). (Source material courtesy of Caerdroia magazine (54).) PHILOSOPHY OF HYPERTEXT page 95 Fisher reports on New Labyrtnths in the Eastern USA: Sig Lonegren details the example s on his list The Great Maze, Wanaka. New Zealand: Stuart Lands borough tells the story of his maze . The Gibson Lane Maze: Martin Douglas on the construction of a simple school maze . A Swedish Schoolyard Labyrinth: improve a school courtyard. Anita Stjemstriim explains bow! [Fig. 9. Transpointing windows: screen shot of the 1998 working demo] People are amazed to see these windows move and scroll, with connections following the content as it moves on the screen. Unfortunately that is all that the demo does, since it has no support for editing or saving the connective structure. These methods are currently being put into a product under the tentative trademark CosmicBook(tm). A more recent interactive screen implementation of transpointing windows (screen shot fig. 10) was created by Ka-Ping Yee in 1999. This was served, like Walker's, from our xu88 server (rechristened as Udanax Green). Yee's PYXl is a skeleton front end that shows and edits text, links and transclusions, and may be PHILOSOPHY OF HYPERTEXT page 96 extended to reach more of the server's functionality. Both Udanax Green and PYXI were released under an open source license in 1999 (25), and are available at udanax.com. 1.1,0cl.0,3 A Dec:laratian of the Representatives of the United States of America, assembled · The Declaration ot the Thirteen United States of America [Fig. 10. Screen shot oftranspointing windows by Ka-Ping Yee, showing his PYXI viewer served from Udanax Green server (xu88 model). Only transclusions happen to be shown, though PYXI also handles content links. ] In fig. 10 we see early and final drafts of the Declaration of Independence (104), highlighting transclusions and differences with color. Jefferson's final draft PHILOSOPHY OF HYPERTEXT page 97 (right) is familiar to every school child in the USA. The earlier draft (left) is not. Being able to compare them side by side gives interesting insights into both the politics (for instance, the phrase "subject them to arbitrary power" being replaced by "reduce them under absolute despotism") and to Jefferson's own rationalist point of view (the phrase "sacred and undeniable") being replaced by the phrase "self-evident".) USES OF TIIIS VISUALIZATION We believe that everyone needs transpointing windows to support analysis and detailed understanding- by parallel commentary, precise annotation and the explication of contents; by the facilitation of pinpoint controversy. These are needed not just in scholarship, but legislation, diplomacy, and anywhere that interrelated documents need to be seen. We need this visualization for side-byside parallel documents (from holy books to legislation); for detailed explication, commentary or disagreement; for comparing successive versions of a document. We even need it for today's one-way hypertext: many users would be more comfortable seeing how they recently reached a document. TRANSPOINTING WINDOWS FOR EDITING Transpointing windows are also badly needed in the writing and editing processes. The real work of writing is rewriting; and especially in big projects, is principally the overview and control of large-scale rearrangement- a rearrangement process that used to be called "cut and paste" until those terms were redefined by the Macintosh in 1984. The writing process, and especially sophisticated rewriting, consists of repeatedly reorganizing evolving drafts. Thus what is most needed for writing PHILOSOPHY OF HYPERTEXT page 98 and editing are ways to track the continuities among drafts, and compare alternative plans of organization and parallelism among different versions. The Xanadu project has long intended to use transpointing windows for editing that shows origins, as illustrated in Fig. 11. SOFTWARE PHILOSOPHY Short Version SOFTWARE PHILOSOPH\' Long Version Some say, "Software is created by committees offools." e case that software is never exactly what rs want, and rarely what the designer-- if e-- wanted either. the process of n it becomes something else. [Fig. 11 . Pullacross editing is another use of transpointing windows (simulated graphic). ] Instead of deleting content from one place and plugging it into another (today's distorted meanings of the venerable terms "cut" and "paste"), the author should be able to pull screen contents from old versions into new, seeing all points of origin and also seeing what contents have not yet been used. In fig. 11, we see trails of origin, or origin beams, showing content transcluded from one version to another. Such beams, optionally visible by the author after content has been moved, should make clear where the content is from and what's been left out. (Shown at paragraph level for simplicity; in post-1965 models we have striven to take this to the level of individual characters.) PHILOSOPHY OF HYPERTEXT page 99 Such a mechanism of side-by-side companson and work- what we call pullacross editing- should allow authors far easier revision, able to track visually the original content from previous drafts- telling at a glance where things come from and what is missing. Transpointing windows, as shown for editing, can be both a tool for the writer to perform the operation, and a tool for the reader to see versions and their differences. Note that version management and CASE (Computer-Aided Software Engineering) tools, as they have independently developed in various areas, often do the same things in more formalized and difficult style, and have typically not been generally intended for users as well as authors. Making transpointing windows come true has always been a key milestone for the Xanadu Project. These self-explanatory pictures have been the simplest sharable tokens inside and outside the group. You do not have to understand computer technicalities (let alone ours, new or old) to understand this visualization, or see its importance. We are amazed that the goal of making it possible for people to work with transpointing windows appears not to have been embraced by anyone else in the computer field. (A search at Google.com for "transpointing windows" on 7 May 2000 found only ten references. The same search on 16 Feb 2001 found only thirteen; on 5 Jan 2002, fifty-six.) PHILOSOPHY OF HYPERTEXT page 100 A LITERARY STRUCTURE WITH TWO FUNDAMENTALLY DIFFERENT FORMS OF CONNECTION Transpointing windows have long been a clear Xanadu objective, both to show links (as in figs. 2, 3, 4) and shared content (as in figs. 5, 6}- or both (as in fig. I). Now we must discuss these two forms of connection. "Links and transclusions" may be simplified to: connections between things which are different, and connections between things which are the same. For this to be a workable distinction, they must be implemented differently and orthogonally, in order that linked materials may be transcluded and vice versa. This double structure of abstracted literary connection- content links and transclusion- constitute xanalogical structure. They are literary connections because they deal with literary boundaries ( 15). In the Xanadu designs we have always striven to represent both forms of connection. Transclusion is what quotation, copying and cross-referencing merely attempt: they are ways that people have had to imitate transclusion, which is the true abstract relationship that paper cannot show. A transclusion is not a copy and not an instance, but the same thing knowably and visibly in more than once place. (In The Future of Information (6), I argued that this may best be understood as cosmic identity.) This is a simple point which is remarkably difficult to get across to people. Throughout the world, people's methods of filing, copying and cross-indexingon paper or computers- are simply workarounds to imitate transclusion. Some computer filing and copying methods are actually forms of transclusion (aliases and caches, "frames" and "banners" on the Web). PHILOSOPHY OF HYPERTEXT page 101 Note also that the famous "trails" of Vannevar Bush's memex system (64) were to be built from transclusions, not links. Not distinguishing between links and transclusions is causing rrusery everywhere, for instance in lawsuits against having one's page brought into someone else's frame (a form of transclusion) which the lawyers refer to as "linking"- hopelessly confounding a key difference. Failure to distinguish between links and transclusions has also reduced the impact and clarity of the ACMs endorsement of the Xanadu transcopyright model (38). It appears also to be part of the misunderstanding of the Xanadu copyright model in the paper by Samuelson and Glushko (3 5). Any implementation of transclusion is necessarily a simulation or an enactment of knowable identity, whether accomplished by a live connection, a cache, or an alias on the desktop. But in these simulations and enactments we must make visible the knowable identity; which copies and instances do not. THE Xu88 MODEL (the Gregory-Miller Synthesis) The xu88 Xanadu structure makes an extremely clean distinction between links and transclusions. (A walkthrough of the details of linkage and transclusion in this model is in Appendix F.) The elegant simplification of document structure into these two symmetrical structures was the achievement of Roger Gregory and Mark Miller. They designed the xu88 system in 1981 and partially finished it in 1988 (when it was - -- ---- PHILOSOPHY OF HYPERTEXT page 102 set aside in the reorganization at Autodesk.). The xu88 system enacts two types of connection: profuse and unbreakable deep links to embody the arbitrary connections that may be made by many authors throughout the world (content links); and a system of transclusion- visible, principled re-use- showing the origins and context of quotations, excerpts and anthologized materials, and showing content transiting between versions (transclusions). This dualistic symmetry of links and transclusion is what I have since the nineteen-eighties called "xanalogical structure" (discussed in 13, 14, 15, and passim through various other pieces in the bibliography). The xu88 design was an elegant solution, in part because it eliminated most of the possible units that might be referred to in a document. No separate items were distinguished within the document; and the connections were simply between passages. This elegant design (now open source at udanax.com) created a non-centralized, efficient, grand address space for an ever-growing universe of documents, and provided a profound system of version management. This single vast and expandable address space is called tumbler space (or the tumbler line), and potentially addresses everything that will ever be published. This is because it uses segmented arithmetic (actually defined by transfinite mathematics), which very compactly can refer to extremely large and small spans of content and sets of servers. The units of the xu88 design are the byte, the link, and the version. (There was no "document" as such). As revealed when the secrets were disclosed in 1999 (Appendix L ), the xu88 design uses ingenious extensions of the enfilade mechanism, including one PHILOSOPHY OF HYPERTEXT page 103 enfilade mechanism which managed versions. At the address of a version- a single location on the tumbler line- is a structure of permutation matrices which perform the "I to V transform" and "V to I transform". (I stands for "invariant", the original input address where an element is stored permanently, its permanent I-stream address, and V stands for its current, "variant" address in any particular version.) If you know the current address of a character, this mechanism will tell you its permanent address; if you know the permanent address of the character and want to know where the character is located in a particular version, this mechanism reveals that.* * This is the "poomfilade," as yet insufficiently documented. A principal revelation in the 1999 disclosure (25, 26) was that the addressing scheme explained to the public, and presented in the interfaces of the system., was in fact the principal internal reference scheme as well. The permutation matrices inside the xu88 processor, managing the entire distributed docwnent and link space, are built on tumbler arithmetic in enfiladic structures. The following visible units of xu88 each have an address in the grand address space: • CHARACTER (a minimal unit, given a permanent address when content first arrives and later permuted into different versions) •VERSION (a list of contents in the grand address space- implemented as a hidden structure of permutation matrices) •LINK (two lists of contents in the grand address space) •LINK TYPE (a property associated with a link) The xu88 link supports a deeply different model of linking from that popular today. A link connects two collections of content; any part of the first PHILOSOPHY OF HYPERTEXT page 104 collection of content in one version remains linked to any part of that content in any other version. While transclusions are supported, they do not appear as units at all; they are simply spans discovered to be present in common between two versions. If the same content (as identified by its address) is in two documents, it is recognized by the system as transcluded. THE LEAP TO THE xu92 DESIGN Regrettably, the xu88 design was dropped in 1988 when the Xanadu project received major funding from Autodesk, Inc. of Sausalito, California, creator of the highly successful AutoCAD. Through complicated changes of management and design approach, a different design was begun, which we refer to as xu92. This was based on the "ent" data structure designed by K.Eric Drexler, a highly efficient version management method, but the work bogged down over a four-year period as the group attempted to optimize the structure. Development of the xu92 system halted when Autodesk shut down its support of the Xanadu Project in 1992. Work on both xu88 and xu92 has been put into open source, as will be discussed in Part 6. PHILOSOPHY OF HYPERTEXT page 105 Part 5b. PUBLIC STRUGGLES: WHAT HAPPENED OUTSIDE THE XANADU PROJECT While my particular concern is with Project Xanadu and its concepts, most of the electronic document struggle has happened outside Project Xanadu, especially in recent decades. The conventional computer world has substituted documents which merge two trivial concepts: hierarchy and the simulation ofpaper. 1. HIERARCHY No one knows when hierarchical directories first appeared. It is generally agreed by people I have talked to that they probably existed by 1952, perhaps considerably earlier. Before magnetic tape and magnetic disk they seem less likely. Hierarchy matched the engineering way of thinking. Engineers live in a world where categories don't overlap, all work is scheduled and predictable, and nothing is out of order. Engineers are used to well-defined, closed problems supplied by management or customer. These problems are departmentalized, compartmentalized and scheduled, and they match well to the notion of jobs' which get 'finished'. This is very different from the world of literature and art, which go on changing forever; of academics and artists, whose problems and tasks are never finished, and with the issues of overlap and interpenetration (Fig 1, earlier) and its PHILOSOPHY OF HYPERTEXT page 106 principled re-use (Fig 2, earlier). Hierarchical structure is deep in virtually every document file format today. There is almost no criticism among professional computer people of hierarchical files and directories, although Jakob Nielsen has a small piece on the matter (110). The World Wide Weh Consortium is building further on hierarchical structure with XML. 2. THE SIMULATION OF PAPER At Xerox PARC, they imitated paper. (See PARC history, 102.) They did this with great fanfare, impressing computer people everywhere with their great demos throughout the 1970s. The main reason they imitated paper on the computer screen was that Xerox was a paper-based company. Although the PARCies considered themselves a rebel group, they deeply accepted the paradigm of paper on which the Xerox Corporation based its fortunes. As one P ARCie put it: "Documents are pretty much black marks on white paper."* *Dick Shoup, quoted in 102, p. 233 (paperback edition). I believe the PARC attitudes represented a total misunderstanding of documents and literature, mistaking the appearance for the nature underneath. (This is like saying that a child is just an animal that happens to have been born to adult humans, or saying that love is just physical attraction.) Documents are interconnected units with connections between them, and the fact that they PHILOSOPHY OF HYPERTEXT page 107 have been on the confining medium of paper for so long was just a regrettable fact to which there was no known alternative. The real goal of the Xerox PARC effort was to create more paper documents, so Xerox could sell more copiers. The PARC work was not productized by Xerox, but it was successfully productized by Steve Jobs at Apple. His theatrics and marketing persuaded the world that this dumbdown of decorated directories ("folders" and the "desktop") and of WYSIWYG (paper simulation- "what you see is what you get WHEN YOU PRINT IT OUT) was actually a new kind of freedom. Insofar as it provided new computer tools for artists and typesetters, it was. But in locking the world into a paper model, I believe it was a great harm to the world. WEB HISTORY IN FOUR PARAGRAPHS The World Wide Web has taken the paper model and built it up into trivial hypertext. The World Wide Web has been basically five initiatives: • Tim Bemers-Lee created it as a simplifying common interface to various Internet services, with a simple page format for a delivered file- either statically stored or generated on the fly- and a relatively secure connectionless protocol for delivering these pages. Evidently he did not imagine two-way links. (69.) • Marc Andreessen at the University of Illinois used a number of simple initiatives to decorate these pages, adding GIFs, JPEGs, later frames, tables, cookies and so on and making the former paper-like page into a mosaic (hence the original name of his browser, Mosaic). The effect on the public was spectacular. (No one in the hypertext research community who saw it in the early days imagined it would have any importance.) This project became Netscape. I believe the real creator of the Web was - - - - --- - - - -- - PHILOSOPHY OF HYPERTEXT page 108 Marc Andreessen, because he created and standardized the "Web browser" which even Microsoft had to copy. I believe it should be called the Andreessen browser. • Microsoft tried to take over Web standardization, fighting a brief feature war with Netscape. • Bemers-Lee took over the "standardization" effort, getting a kind of control back for his W3C Consortium. But many protocols and presentations on "the Web" are outside the W3C purview: streaming audio and video, Flash, and a variety of other delivery formats simply do not concern the W3C planners. • Therefore the Web is best defined as what appears to users to fit in the browser, and that will keep changing. However, since the model of embedded markup and one-way links are so broken and trivial, the opportunity for true hypertext- with two-way links and transclusions- remains wide open, if it is given a Web browser-compatible structure. PHILOSOPHY OF HYPERTEXT page 109 Part 6. What Do You Do after World-Wide Hypertext? PHILOSOPHY OF HYPERTEXT page 110 Part 6. WHAT DO YOU DO AFTER WORLD-WIDE HYPERTEXT? My own work since the World Wide Web The World Wide Web, defined and driven by the Mosaic browser, arrived as a spectacular world presence in roughly the fourth quarter of 1994, and a large proportion of people on the planet have had to adjust considerably. Others were surprised at world-wide hypertext. We of the Xanadu Project were not, but were instead outraged at its triviality, one-way links, and emphasis on decoration rather than content facilities. This ended our advantage of having started early, and completely changed the strategic situation, placing us in the same catch-up position as the rest of the world. All corporations and governments are now attempting to control or guide the future developments of hypertext, building on various levels of understanding, and misunderstanding, of the ever-changing World Wide Web. In the meantime, the various Xanadu implementations failed over internal power struggles, leading to the cancellation of the project by Autodesk. The period since 1992 has completely redefined my work and life, since I ceased to be the leading advocate of, and authority on, hypertext, and someone else's project had precisely the impact that I had for over thirty years expected the Xanadu project to have. My personal chagrin and disappointment can hardly be overstated. PHILOSOPHY OF HYPERTEXT page 111 However, this has not deflected my ideals or my concerns. The terrible flaws of the World Wide Web have greatly occupied me since its success, and I have continued on a number of alternative paths toward the same objectives. My personal commitment to the Xanadu model is absolute, though the brilliant attempted optimizations of the xu88 and xu92 designs were very premature. I cannot believe people do not see its powers. However, the previous hi-tech Xanadu implementations (xu88 and xu92) are not working, and it has been vital to move the idea forward. In what follows, I will outline this work as it has continued since 1992. First, however, it is important to mention what happened to the Xanadu project since that time. TIIE FATE OF TIIE PROJECT AT AUTODESK At Autodesk between 1988 and 1992, the author and Roger Gregory were removed from all technical authority, and other management took over. The xu88 code was set aside, and new programmers brought in with different ideas and varying degrees of commitment. In 1992, the Xanadu project lost Autodesk's support. The principal reason was that the group had not delivered product. The principal error had been the coding team's decision to change designs in 1988, replacing a fast drive to product with protracted redesigns which continually repackaged the ent versioning mechanisms, deliveries, buffering, and helper concepts. However, in a larger sense the error was the traditional mistake of premature optimization. No one on the team imagined that such a trivial one-way structure as HTML would become the world hypertext standard, and most of our design - -··-.- - - - - - -- PHILOSOPHY OF HYPERTEXT page 112 work was intended to create the deeper structures discussed in this thesis. Had we imagined the need for a trivial hypertext model like HTML running on simpleminded protocols like HTfP, I am sure that we could have designed something just as simple but having the powers we believed were important, especially reuse with visible origins and permission. (The VLIT structure now underway, discussed pp.118-120, is intended to be such a simplified design.) In 1992, the Xanadu assets were given back to the members of the project, about twenty of us, though Autodesk continued to hold the "Xanadu" trademark for a time. Formally we were still a company (XOC, Inc.), but the group was unable to converge on a plan or a strategy that would hold us together as a team. The company shrank to Roger Gregory, Keith Henson and myself, with Henson serving as CEO. Several deals were attempted, and development continued by Gregory and Henson for several years. (It was during this period that the Web hit public consciousness, escape velocity and critical mass; and our deliberations were correspondingly affected.) In 1999 the proprietary knowledge of the Xanadu project, and the formerly proprietary code of the xu88 and xu92 versions, was formally put into Open Source at the O'Reilly Open Source Conference. Since that time Roger Gregory has continued as custodian of that open-source development. THE XOC DIASPORA The collapse of Autodesk support for XOC, Inc. may be called a diaspora- the expulsion and spread of a number of people and ideas. Aside from Gregory, Henson and myself, who have participated directly in trying to keep the code and company alive, a number of others have tried to promote these ideals in PHILOSOPHY OF HYPERTEXT page 113 their separate ways. In particular, an initiative for Web commentary has been sponsored by K. Eric Drexler- inventor of the "ent" data structure of xu92- at the Foresight Institute and crit.org. (This system is one of several commentary mechanisms on the Web.) A principal participant in this project was Mark Miller, traditional Xanadu designer of both xu88 and xu92; principal implementor was Ka-Ping Yee, then of Industrial Light and Magic (and creator of Fig. 10). RECLAIMNG THE XANADU TRADEMARK In various steps and degrees, I have recovered the "Xanadu" trademark from Autodesk, recaptured the xanadu.com domain name (as well as xanadu.net), and received governmental registration of the "Xanadu" trademark in both the USA and Japan. This completes the circle of the trademark's history. The Xanadu project began in 1960, around a referential or indirect model of electronic documents maintained as a list of contents. This project was given the name "Xanadu" in 1967; that "Xanadu" trademark was my personal trademark from 1967 to 1988, owned by Autodesk from 1988 to 1992, and became mine once again, following the Autodesk diaspora. In legal terms, "Project Xanadu" is today another name for Ted Nelson. In intention, however, it is the same idealistic approach to hypertext and software, around a referential or indirect model of content listing, that it has been since 1960. HOW PUT THE XANADU MODEL ON THE WEB? In the first two years after the Autodesk Diaspora, it was still not clear that another system would beat out the Xanadu plan. I quickly began investigating a variety of alternative approaches for the Xanadu model without XOC's PHILOSOPHY OF HYPERTEXT page 114 incomplete code. Investigations included• models for bidirectional links using relational database • micropayment systems • different schemes for using the Web's browser These are among the issues that nearly every project in every company in the world is trying to grapple with. They have both technical and political (standardization) aspects. Since 1994, in Japan, I have had the chance personally to demonstrate various aspects of these systems in new and simple versions. WORK IN JAPAN AT SAPPORO HYPERLAB At the invitation of Professor Yuzuru Tanaka of the University of Hokkaido, creator of the IntelligentPad system (115), I joined the Sapporo HyperLab in 1994. • I worked with a team building a Zip Editor (based on my 1965 Zipper Lists (23 ), and implemented using IntelligentPad). The lead programmer was K. Ookubo*, a former member of the Common Lisp standardization committee. This proved to be an interesting and valuable test-bed for IntelligentPad, exercising it in previously unexpected directions. * He has subsequently changed the English spelling of his name to K. Ohkubo. Also at the Sapporo HyperLab, • I worked out the legal doctrine of Transcopyright, a vital enabling method for the referential model on the Web. This embodies the previous Xanadu model of virtual inclusion. It has been pronounced legally sound by lawyers from the USA and Japan, among others. PHILOSOPHY OF HYPERTEXT page 115 Along with this, I continued to adapt the Xanadu designs to the Web, designing various components which were intended to fit together m a distributed Web model. This included• Theory of proportional micropayment • the Deep Cache model for acquired content WORK IN JAPAN AT KEIO UNIVERSITY Invited to Keio University from March of 1996, I have had a chance to work out these ideas step by step, with the encouragement of Deans Aiso and Saito, and the deep collaboration of professors Hajime Ohiwa a former physicist now running the very hospitable CReW Laboratory ("CReative Workspace"), and Kenji Naemura, formerly a manager with NIT. The first of these ideas to be worked on at Keio were • a fully-functional demonstration of a microversioning text editor (OSMIC, programmed under my direction by student Ken'ichi Unnai (48)). • The ideas of transpublishing and transquotation as parts of the Xanadu concepts which could be moved to the Web. Transpublishing means making contents available on a server from which portions may be quoted. Transquotation means arranging to have such quotations appear in a Web browser. • The "Keio formats" for transquotation- especially transquotation marks leading to permission statement and original context (44) To support transpublishing and transquotation, •I worked out (with Andrew Pam of Xanadu Australia), a proposal to the IETF for a "TXT SRC" tag, which he wrote. • Yousuke Igarashi created a system of Perl scripts for transquotation, PHILOSOPHY OF HYPERTEXT page 116 based on fonn submission by the user . .• I collaborated as a designer on a transpublishing system using Keio formats for serving text quotations under a grant from MITI, through JIPDEC. (This was done in collaboratoration with Ohiwa CReW Lab at Keio, under the programming supervision of Yoshihide Chubachi.) • This resulted in a transpublishing database server built under the supervision of Yoshihide Chubachi (as team leader) registering documents in a database for transpublication. • I worked out client mechanisms bringing in quotations in either Netscape or Internet Explorer (45) • This system was integrated with HyperCoin micropayment ( 56) • Yoshihide Chubachi supervised a new implementation of the OSMIC microversioning system, this one called INLUV. WORK IN JAPAN IN COLLABORATION WITH ASCII CORPORATION Kazuhiko Nishi, founder of ASCII Corporation in Tokyo, for a time sponsored my investigations of micropayment. • As part of this effort, I organized under Nishi's sponsorship a micropayment seminar, attended by Dave Theurer (creator of the video game "Tempest", as well as scriptable graphics systems); Phil Zimmerman (creator of the PGP encryption system), and Tsutomu Shimomura, a well-known chaser of rogue hackers. OTHER INTERNATIONAL COLLABORATIONS: COSMICBOOK™ • With Ian Heath of RunLevel3, Ltd. in England, I am working on a product entitled CosmicBook™(55) with transpointing windows operating under Microsoft Windows. Made publicly available in June 200 I, the free CosmicBook reader is the first product offering transpointing windows. PHILOSOPHY OF HYPERTEXT page 117 OTHER INTERNATIONAL COLLABORATIONS: ZIGZAG®, A NEW GENERALIZATION OF STRUCTURE • Working with Andrew Pam of Xanadu Australia, I finally implemented my 1986 design of ZigZag(tm), a new generalized computing model based on animated crossed lists (71 ). ZigZag is intrinsically nonhierarchlcal, but extensible in hierarchical ways if anyone wants that. • The ZigZag prototype was programmed in Perl by Andrew Pam. •A U.S. patent has been issued for ZigZag structures and concepts (82). • A more recent implementation is "Gzigzag", done by a team at the University of Jyvaskyla in Finland, under the directorship of Tuomas J. Lukka (80). It is of interest that the Gzigzag project incorporates the referential text model of Xanadu structure in general. • Another Zigzag-compatible system has been created by Les Carr at the University of Southampton in England. Carr's system holds the data structure in XML and manipulates and shows it through JavaScript (79). •Most recently, Adam Moore, a chemist at the University of Nottingham, has produced a ZigZag viewer which he calls "ZAM" (ZigZag As Molecules). It uses a chemistry visualization system, RASMOL, as an output projector for ZigZag structures. ZIGZAG SOLUTION OF THE ITEM QUESTION The issue of structural units and how they fit with the referential model- "the item question"- has always been a puzzle in the Xanadu designs. Throughout this thesis, I have mentioned the "item problem", referring to the question of how to resolve conceptually the distinction between discrete items at the user level and passages of text. As a new structural model, ZigZag solves the Item issue of all the Xanadu designs, but in a new and different way. In the Miller-Gregory solution, items were eliminated in favor of deeply trackable PHILOSOPHY OF HYPERTEXT page 118 fine-grain transclusion. ZigZag, adapted to a referential model of text, is a completely different approach to that paradoxical issue. SUPER-ZIGZAG: THE FLOATINGWORLD™ DESIGN The most far-reaching vision of ZigZag potential is in the FloatingWorld design (78), which postulates an n-dimensional world of interaction, visualization and new programming methods based on a structural ZigZag backbone. (Lukka and the Gzigzag group (80) have been essentially programming with reference to this design document.) WORK UNDER PRIVATE AUSPICES • Under private funding, I developed with Andrew Pam, and patented, a pro-rata miccropayment system (HyperCoinTM ( 56) ). TIIE NEW XANADU FOR THE WEB: A QUOTABLE UNIVERSE I am now turning my attention to putting the Xanadu model on the Web in an open-source version, interoperable with Web links and Web browsers. It is being publicized without the name "Xanadu", to avoid confusion. This design work centers on adapting the Xanadu model to a Web-compatible model, intended to be cross-linkable with HTML, meaning that HTML links can point at these documents and vice versa. However, between these documents there will be two-way links and transclusion on the classic Xanadu model. This is intended to support a principal Xanadu objective- deep quotability with visible origins- and other related forms of deep interconnection. (Fig. 12.) PHILOSOPHY OF HYPERTEXT page 119 THE HIDDEN, INDIRECT COPYRIGHT SOLUTION A QUOTABLE UNIVERSE Permitting indirect re-use changes the nature of the problem. HOR of .VLIT VIRTUAL FILE )--- READER/ DEEP AVAILABLE CONTENT permanently promised; with no embedded with permission to re-use indirectly W o D o D D D E A reader may easily see the original context of every part, and may in turn quote quote any of this material by the same indirect method. [Fig. 12. The Quotable Universe, a Web extension of the Xanadu Model (see Fig. 3)] PHILOSOPHY OF HYPERTEXT page 120 This most recent Xanadu-based design is now called the VLIT file format (62). Preliminary design of the VLIT format is underway with David Durand of the W3C Consortium, and Tim Brailsford and Craig Stewart, both of the University of Nottingham. Unfortunately, while we have intended to follow W3C specifications, the W3C specifications do not yet work in any present browsers, and it is unclear when they will, because of the political and standardization problems of the Web. I am investigating alternative plug-ins that can bypass this problem. The possibilities are still great. However, the computer world is about the politics of standardization, and where it will go is a mystery to everyone. POSSIPLEX™, THE EDITOR For over forty years I have planned a super-editing-system and user environment totally unlike anything that has yet been seen, helping authors to revise and reorganize huge bodies of content at far greater speed than is now possible. Those who think in terms of today's paper simulation cannot imagine it. It is intended to show the origins of content and interrelations of many versions of parallel document complexes. I have now chosen "Possiplex" as the probable name of my new design for this Xanadu super-editor and browser. It is now being mapped to ZigZag mechanisms. This editing system is designed around transpointing windows, intended to show link and transclusion beams among many documents and versions simultaneously (if the user wants), varying the color and transparency of these beams quickly. (While transpointing windows were always part of the earlier Xanadu designs, Possiplex is being designed from scratch with no previous PHILOSOPHY OF HYPERTEXT page 121 structural commitments.) In the Possiplex design, the user may do editing operations between versions (as in Fig. 11), rather than within a single changing version (as is now conventional). Since much writing is rewriting, this makes clearer the structure of the revisions taking place. This new visualization would be impossible without the success of ZigZag as a programming platform. Because of the simplicity and consistency of the ZigZag internal structure, the ZigZag backbone should make it easy to implement this design with no concessions to current traditions. The main data structure has been fundamentally designed and is now being polished. IRV ANDMYNOTES Alas, the vast body of written notes I have accumulated over the last forty years- in the hundreds of thousands- are broken and separate; and the "Irv" concept, a dream of writing without ever breaking connections, has been painfully violated in everything I have written. But the system that will make this possible, no longer entrusted to other designers, seems finally near at hand. In full circle, it is my hope that such a super-editor will smooth the completion-or at least the progress during my lifetime-- of my Schematics work on the unified Irv model. PHILOSOPHY OF HYPERTEXT page 122 Part 7. Conclusion PHILOSOPHY OF HYPERTEXT page 123 Part 7. Conclusion In our time, data structure has replaced philosophy. For thousands of years, philosophers have created universe-filling generative concepts, but these were obscure and far from the common man (unless they were associated with deities). Now, anyone can create universe-filling generative concepts at home or at their desk-- and these concepts have consequences, if they become nuclei of functioning software. Everyone argues these days about the proper forms of software. Ten years ago I was hearing software design arguments in the line at the grocery store in California. Now the arguments in the grocery store are about the structure of hypertext. The World Wide Web has created a strange standard of one-way hypertext built around hierarchical directories, and now linearizing these hierarchies into XML. It has no recognizable philosophical basis, except for the ritualistic glorification of perfect hierarchy (XML). On the other hand, Project Xanadu, the original hypertext project, was from the beginning based on a broad philosophical perspective. Project Xanadu is often misunderstood as an attempt to create the World Wide Web. It has always been much more ambitious. As a substruction of a broad range of document, filing and quotation concepts, Xanadu-based design takes in and compresses far more objectives than the Web, and does so more simply. PHILOSOPHY OF HYPERTEXT page 124 This thesis has been a brief statement of a deeply-held philosophical position, its philosophical origin, and a few of its corresponding methods. We have presented a referential system of data representation that allows overlaying applicative structure; and which implements in particular two forms of visible crossconnection between parallel documents and versions: re-use (trackable automatically) and user-created links. This view and its techniques have many other tenets and ramifications that we have no room to discuss here. I have stayed as far away as possible from particulars, except to be clear in the hypothetical examples, which suffice only to explain the basic feasibility of our referential structures and methods. Referential structure of media has great power. Besides clean support for unbreaking links and transclusions, separating structure from content has many other benefits. Data may be used in place. All content is additive. All structure is additive and applicative, rather than tangled inside (as in the SGML model). This permits many structural variations on the same particular documents and their contents-- variations whose cross-connections may in tum be viewed. If the list of content is made the fundamental unit, many things become possible and principled: nondestructive, additive editing; branching versions, all accessible and re-branchable; profuse unbreaking links; principled and visible re-use (transclusion); deep intercomparison along both links and transclusions; and transpublishing under transcopyright. This proposes a vision of a very different world of media: a network literature of a totally different form and nature from anything that can now be seen. It will allow completely new ways of organizing material-• where a fundamental building block is the excerpt directly connected to its origin; - - - - -- -- - - - PHILOSOPHY OF HYPERTEXT page 125 • which does not require closed packaging of content in segregated files or directories; • where you may recomposite anything, and may build layered and overlapping structures where needed; • which allows detailed annotation and detailed intercomparison-- showing and following commonalities, showing and following content links; and • which allows open recompositing and open republication, based on the right of open referential re-use, •with copyright handled automatically. I believe this all fits together cleanly. I think the simple technical examples given above will make clear that this is an entirely feasible approach-- and, once you understand it, obvious. And I think it is altogether possible that many people will want to visit and work in, and help build, such a subuniverse. There is no question that these visualizations and methods are a long way from the prevailing protocols, browsers, and "standards". But that would not be the important question anyway. (Standardization wars will continue forever.) The important question is whether these ideas can enable a principled and feasible alternative universe, beneficial even if implemented only in part, a potential zone of clarity in the chaos of the Web, desirable enough to be worth the uphill effort. To underpin this, we will need some alternative infrastructure-- protocols, servers, permanent publishing methods, compliant browser plug-ins and editors, transpublishing caches, and a very different payment method from those currently popular. I am still proposing an entire form of literature where links do not break as versions change; where documents may be closely compared side by side and PHILOSOPHY OF HYPERTEXT page 126 closely annotated; where it is possible to see the origins of every quotation; and in which there is a valid copyright system-- a literary, legal and business arrangement-- for frictionless, non-negotiated quotation at any time and in any amount. The Web trivialized this original Xanadu model, vastly but incorrectly simplifying these problems to a world of fragile ever-breaking one-way links, with no recognition of change or copyright, and no support for multiple versions or principled re-use. Fonts and glitz, rather than content connective structure, prevail. Serious electronic literature (for scholarship, detailed controversy and detailed collaboration) must support bidirectional and profuse links, which cannot be embedded; and must offer facilities for easily tracking re-use on a principled basis among versions and quotations. Xanalogical literary structure is a unique symmetrical connective system for text (and other separable media elements), with two complementary forms of connection that achieve these functions-- survivable deep linkage (content links) and recognizable, visible re-use (transclusion). Both of these are easily implemented by a document model using content lists which reference stabilized media. This system of literary structure offers uniquely integrated methods for version management, side-by-side comparison and visualizable re-use, which lead to a radically beneficial and principled copyright system (endorsed in principle by the ACM (38)). Though dauntingly far from the standards which have presently caught on, this design is still valid and may yet find a place in the evolving Internet universe. Is it too late? Would it mean just too much overhead? We believe that like the - ----- - - - - - - - - -- -- PHILOSOPHY OF HYPERTEXT page 127 hidden overhead of the Internet itself, such a structure will pay for itself manyfold. We can see no other important agenda for hypertext. PHILOSOPHY OF HYPERTEXT page 128 Bibliography - PHILOSOPHY OF HYPERTEXT page 129 Bibliography, Thematically Organized In order to make this bibliography separately understandable and useful in itself, rather than just a pool of citations, I have divided it into sections on different themes or categories. Items in brackets [ ] are cross-references to sources which belong in multiple categories of this bibliography. This thesis itself is even included in its own bibliography, in order to provide a more complete listing for future separate use by others. Best Writings on the Xanadu® Project by Others Some excellent summaries of the Xanadu project have been written. Here are three of them. 1. Anonymous, "Ted Nelson, Hypertext Pioneer", at http://www.techtv .corn/screensavers/showtelVstory/0,23008,2127396,00.html This brief Web page published by Ziff-Davis is by far the clearest and most accurate summary of Project Xanadu's work and views to be published anywhere. I do not believe it was written by anyone associated with the Xanadu project. The page also offers streaming video of an interview with the author. 2. Anonymous, FEED magazine piece on Xanadu, at http://www.feedmag.com/htmVdocument/98.02nelson/98.02nelson_master.html NOTE: this piece seems to be only intermittently available. "An interesting article in Feed reprints short excerpts from Literary Magazines and invites three leading writers to provide commentary. Part of a continuing exploration of the foundations of the docuverse." -- Mark Bernstein 3. Peter Schmideg, "Ted Nelson", at http://www.wigged.net/htmVnews/nelson.html This provides colorful background on the Xanadu project. PHILOSOPHY OF HYPERTEXT page 130 Canonical Writings of the Xanadu Project Here are some writings which are central to the Xanadu project. 4. The Xanadu Vision http ://xanadu.com/xuvision.html This is a statement of the overall Xanadu ideal. 5. Deep Hypertext: The Xanadu Model http://xanadu.com/xuTheModel/ This states the Xanadu model for a popular audience. 6. Theodor H. Nelson, The Future ofInformation. ASCII publishers (Japan), 1997. Available in scanned form at http://www.sfc.keio.ac.jp/-ted/INFUTscans/INFUTscans.html This is one of the better summaries of the Xanadu project and its objectives, from a somewhat literary perspective. Unfortunately not published in the USA. 7. Theodor H. Nelson, Philosophy ofHypertext. Thesis submitted for the Ph.D. in Philosophy at Keio University, 2002. (This document.) Outlines the history of the Xanadu project and its relation to the philosophy of General Schematics-- relating hypertext to the indeterminacy of description, the indeterminate boundaries of ideas and subjects, the indeterminacy of writing, and the interpenetration of relationships in all directions. Suggests that the abstractive process of identifying identities and equivalences is closely related to the issue of transclusion. 8. Theodor H. Nelson, "As We Will Think." Proceedings of Online 72 Conference, Brunel University, Uxbridge, England, 1972. This was an early paper on the Xanadu concept, containing possibly the first published illustration oftranspointing windows. 9. Theodor H. Nelson, "Virtual World Without End." Proceedings of Cyber Arts International Conference, Sept 1990. - -- - - - - -··- - PHILOSOPHY OF HYPERTEXT page 131 This summarizes the Xanadu ideals and plan as of 1990. 10. Theodor H. Nelson, "Replacing the Printed Word." (S.H. Lavington, ed., Information processing 80 (Proc. IFIP 80 World Computer Conference), NorthHolland Publishing Co., 1980, 1013-1023). This was an abbreviated presentation of the Xanadu concepts, and the first for a professional audience; it was later expanded into the book Literary Machines. 11. Theodor H. Nelson, Literary Machines, early editions ( 1981 to 1986). Published by the author. This first edition, appearing in slight variations over several years, contained various inspirational materials on the Xanadu ideas and ideals, but not the technicalities which were presented in the later editions after 1987 (12). 12. Theodor H. Nelson, Literary Machines, recent editions (1987-current). This contains numerous technical details about the xu88 functional concepts and addressing system. Mindful Press; available from Eastgate systems at www.eastgate.com/catalog/LiteraryMachines.html (Translated also into Japanese and Italian.) This major revision of (11) is generally considered the definitional document for Project Xanadu (and, by some, for the Web)-- especially on the basis of its early chapters on the idea of a world wide hypertext system. The later chapters are an introduction to the architecture of the xu88 Xanadu system, designed by Roger Gregory, Mark Miller and Stuart Greene. The tumbler universal addressing scheme (by Gregory and Miller) is presented. What the book does not reveal is the crucial inner mechanisms (revealed in (26)), specifically-• the nature of the enfilade (a tree structure having upwardly propagating characteristics (such as byte count) and/or downwardly imposed characteristics (such as sequence); • the three interior enfilades of xu88, specifically the Granfilade, the Spanfilade and the Poomfilade, which are not yet adequately documented. 13. Theodor H. Nelson, "Xanalogical Structure, Needed Now More than Ever: Parallel Documents, Deep Links to Content, Deep Versioning and Deep Re-Use". Available at the author's site, PHILOSOPHY OF HYPERTEXT page 132 http://www.sfc.keio.ac.jp/-ted/XUsurvey/xuDation.html Also available at http://www.cs.brown.edu/memex/ACM_HypertextTestbed/papers/60.html (The official ACM version is not available on-line except to members.) This is a long paper on the structure and history of the Xanadu project, most of which has been included in various parts of this thesis. Xanalogical Structure and the xu88 Design These pieces stress the symmetry of links and transclusions- a principal aspect of most Xanadu designs, especially the xu88 design (now Udanax Green). 14. Theodor H. Nelson, "The Xanadu Paradigm" http://www.sfc.keio.acJp/-ted/ARCIIlVE/xuposter87 .tif (poster), 1987. At (Only a poor scan is currently available.) This states the whole Xanadu concept very briefly. 15. Theodor H. Nelson, "The Heart of Connection: Hypermedia Unified by Transclusion." Communications ofthe ACM, 38:8 (August 1995), 31-33. Also posted at http://www.cs.uct.ac.za/courses/CS200W/Resources/HCI/Resources/nelson.ht ml This summarizes the arguments for transclusion, and introduces the term "transpointing windows." [Also in this category: 13. Theodor H. Nelson, "Xanalogical Structure, Needed Now More than Ever: Parallel Docwnents, Deep Links to Content, Deep Versioning and Deep Re-Use".] [Also in this category: Appendix F of Theodor H. Nelson, Philosophy of Hypertext. (This thesis.) ] - - -- - PHILOSOPHY OF HYPERTEXT page 133 Parallel Documents, Transpointing Windows, Parallel Textface™ Transpointing windows (trademarked as "Parallel Textface") are a crucial system of visualization for understanding relations between documents, as yet not popular. 16. Theodor H. Nelson, "My Parallel Universe", at http://www.sfc.keio.ac.jp/-ted!TN/PARALUNE/paraluniverse.html General remarks on parallelism and the need to represent and show it. 17. Theodor H. Nelson, "Documents are Parallel by Nature" at http://www.sfc.keio.ac.jp/-ted!TN/PARALUNE/paradoxx.html x Included here as part of Appendix D. This presents my fundamental argument that documents are parallel. I believe parallelism is a :fundamental relationship among documents (among other things) that needs representation, and is not considered in today's XML fad. 18. Theodor H. Nelson, "Examples of Parallel Documents", at http://www.sfc.keio.ac.jp/-ted!TN/PARAL UNE/parexamples.html Included here as part of Appendix D. This gives examples for my fundamental argument that documents are parallel. 19. Theodor H. Nelson, "Parallel Visualization: Transpointing Windows" at http://www.sfc.keio.ac.jp/-ted!IN/PARAL UNE/paraviz.html Included here as part of Appendix D. This presents the transpointing-windows visualization. 20. Theodor H. Nelson, "Parallel Data Structures" at http://www.sfc.keio.ac.jp/-ted!TN/PARALUNE/paradata Included here as part of Appendix D. Brief remarks on parallel data for parallel documents. [Also in this category: 8. Theodor H. Nelson, "As We Will Think."] PHILOSOPHY OF HYPERTEXT page 134 [Also in this category: 15. Theodor H. Nelson, "The Heart of Connection: Hypermedia Unified by Transclusion. "] [Also in this category: 27. Theodor H. Nelson, "Xanadu Technical Description." ] Transclusion Transclusion, or the same content recognizably in more than one place, is a crucial relationship among documents which is central to Xanadu concepts. 2L Theodor H. Nelson, "Borrow." At http://www.sfc.keio.ac.jp/-ted/ARCHIVE/brw68.tif Specs for what is arguably the first implementation of transclusion, at Brown University, 1968. (Probably programmed by Steven Carmody.) Transclusion was handled here as an operation rather than a condition. The transclusion concept had not been not well articulated at the time, but is implicit in this note. [Also in this category: 8. Theodor H. Nelson, "As We Will Th.ink."] [ Also in this category: 15. Theodor H. Nelson, "The Heart of Connection: Hypermedia Unified by Transclusion. "] [ Also in this category: 13. Theodor H. Nelson, "Xanalogical Structure, Needed Now More than Ever: Parallel Documents, Deep Links to Content, Deep Versioning and Deep Re-Use".] [ Also in this category: 7. Theodor H. Nelson, Philosophy of Hypertext. (This thesis.) ] [Also in this category: 23. Theodor H. Nelson, "A File Structure for the Complex, the Changing and the indeterminate."] History of the Xanadu Project Here are some pieces about the history of the Xanadu project, some of which represented key stages in our thinking. PHILOSOPHY OF HYPERTEXT page 135 22. Theodor H. Nelson, "Project Xanadu History (lo-res)". At http ://xanadu.com/xuhistory.html This is a very brief summary of our work over the years. 23. Theodor H. Nelson, "A File Structure for the Complex, the Changing and the indeterminate." Proceedings of the ACM National Conference, 1965. This was the first published paper on the Xanadu concepts, especially parallelism of documents, though the concept was not fully elucidated. This represents an early and very simple design of the Xanadu family, refuting the notion that "Xanadu needed vast computing power." Tue design proposes the list of sequential text items as a fundamental unit for personal work. Such lists are connected side-by-side, either by either simple links or transclusion (the same items in two or more lists). 24. Theodor H. Nelson, Hypertext Note 10: "Xanadu." http://www.sfc.keio.ac.jp/-ted/ARCIIlVE/htnlO.tif Included in Appendix C. This is a summary of the Xanadu plan as of 1967. Note that the project was being referred to in the past tense, news of its death being greatly exaggerated even then. 25. Mark S. Miller, "Xanadu Secrets Become Udanax Open-Source", at www.udanax.com/ This was a 1999 announcement of the availability of the code for xu88 and xu92. [Also in this category: 21. Theodor H. Nelson, "Borrow."] PHILOSOPHY OF HYPERTEXT page 136 Xanadu Technical Aspects There is no end to the technical concepts of the Xanadu project. Here are some sources. 26. Theodor H. Nelson, "Xanadu Technologies-- An Introduction", at http://xanadu.com/tech/ This is a preliminary introduction to enfilades and the internals of the xu88 and xu92 designs. It presents a brief statement of the technical history of the project, with a walkthrough of the original Model T enfilade structure (later included in xu88 as the Granfilade). A principal revelation is that the tumbler addressing scheme presented externally at the programming interface of the xu88 system (30) was in fact the principal internal reference scheme as well: enfiladic tumbler permutation matrices manage the entire space of 88.1. 27. Theodor H. Nelson, "Xanadu Technical Description", 10 Feb 1971. Published in four sections, at http://www.sfc.keio.ac.jp/-ted/ARCHIVE!XTD--ppl-9.tif http://www.sfc.keio.ac.jp/-ted/ARCHIVE!XTD--ppl0-19.tif http://www.sfc.keio.ac.jp/-ted/ARCHIVEIXTD--pp20-29 .tif http://www.sfc.keio.ac.jp/-ted/ARCHIVE/XID--pp30-40.tif (Note that page 7 is missing from this scan.) This is the last major Xanadu design done by Nelson alone, before the discovery of the enfilade. It is the last non-enfiladic Xanadu design between 1971 and 1992. It is designed around the concept of the ring buffer ("bed") through which a stream virtually moves ("babbles"). Note the illustration oftranspointing windows (p.4). This internal design would have supported transpointing windows for two versions at a time. 28. "XS-7", a 1972 Xanadu design document at http://www.xanadu.com/xu72des.tif These are the technical internals of the first enfiladic Xanadu system, as noted for our use in a typical working document (9 April 1972). Several of the enfilade illustrations and examples for (26) were taken from this document. PHILOSOPHY OF HYPERTEXT page 137 29. Cover page of xu88 FEBE Protocol Manual when the software was still called "Xanadu 88. l ". The 'confidential' notice is now inoperative. At www.xanadu.com/XOCDOC-cover.tif 30. xu88 FEBE Protocol Manual chapters: Front End-Back End [now called in the industry "server-client"] interface protocol ofxu88 design (now Udanax Green): This is the original manual for the xu88 design delivered from XOC, Inc. under Autodesk, before the decision to redesign which resulted in xu92. •Original cover page when it was still called Xanadu 88.l (148), http://www.xanadu.com/XOCDOC-cover.tif • Front page, www.udanax.com/green/febe/ •Introduction, www.udanax.com/green/febe/intro.html • Philosophy and Motivation, www .udanax.com/green/febe/philosophy.html • Implementation, www.udanax.com/green/febe/implementation.html • Designing Frontends, www.udanax.com/green/febe/frontends.html • FeBe Example, www.udanax.com/green/febe/example.html •Technical Overview, www.udanax.com/green/febe/overview.html • Addressing, www .udanax.com/green/febe/addressing.html •Tumbler Arithmetic, www.udanax.com/green/febe/tumblers.html •Links and Link Types, www.udanax.com/green/febe/links.html • Versions, www .udanax.com/green/febe/versions.html •Fe vs. Be, www.udanax.com/green/febe/fe-vs-be.html • F eBe Protocol, www .udanax.com/green/febe/protocol.html • Bibliography, www.udanax.com/green/febe/bibliography.html • Appendix A - FeBe Protocol Syntax, www .udanax.com/green/febe/syntax.html • Appendix B - Useful Tables, www.udanax.com/green/febe/tables.html •Appendix C - Manual Pages, www.udanax.com/green/febe/manpages.html •Appendix D - Glossary, www.udanax.com/green/febe/glossary.html 31. "Current Status of the Udanax Green Code" at www .udanax.com/green/status.html This is a statement (not necessarily current) of the status of Udanax Green code (referred to in this thesis as xu88). See also (32) and (33). 32. Jeff Rush, "Sunless-Sea CyberArchaeology Site", at http://www.sunless-sea.net/ This remarkable site is devoted to sorting out the xu88 and xu92 code, PHILOSOPHY OF HYPERTEXT page 138 especially through on-line discussion. Code change management is handled at (33). 33. Tau Productions' public CVS repository, at http://www.taupro.com/cvs This site handles change management for · the xu88 and xu92 code, as discussed at (32). The Xanadu Copyright Model The Xanadu project was always conceived as a copyright solution based on transclusion and permission to include contents indirectly by reference. Here are some sources. (See also "Transcopyright", below.) 34. Theodor H. Nelson, "Xanadu: Document Interconnection Enabling Re-Use with Automatic Author Credit and Royalty Accounting." Information Services and Use 14:4 (1994), 255-265. This summarizes the Xanadu copyright plan as of 1992, when Project Xanadu still expected to be the worldwide hypertext system. 35. Pamela Samuelson and Robert J. Glushko, "Intellectual Property Rights for Digital Library and Hypertext Publishing Systems: An An\alysis of Xanadu" Available from the ACM at http://wwwl.acm.org/pubs/articles/proceedings/hypertext/ 122974/p39samuelson/p39-samuelson.pdf as well as http://www.sfc.wide.ad.jp/-yoko/maui/Xanadu.html and available in what is apparently a pirated, incorrect edition at http://www.sfc.wide.ad.jp/-yoko/maui/Xanadu.pdf This is a serious and respectful analysis of the Xanadu ownership and accounting model. Unfortunately it suffers from factual errors. It appears to be based on an edition of Literary Machines and other sources which were well out of date when the article was published. It presents the Xanadu copyright model as a pay-per-view system and states that we were going to charge for links, which was not part of the general plan by that time. (These and many other possibilities had been considered earlier in attempting to design a viable economic model for a network.) PHILOSOPHY OF HYPERTEXT page 139 Further, by not distinguishing between links and transclusions Gust as (38) does not), this article continues to perpetuate certain misunderstandings. Transclusion of content means its virtual inclusion in on-line documents, as portions delivered (sold or donated) by various on-line publishers. To conflate transclusion with the link (a connection between unlike contents) continually causes confusion in understanding Xanadu concepts. 36. Theodor H. Nelson, "Transcopyright: Dealing with the Dilemma of Digital Copyright." Educom Review 32: 1 (January/February 1997), 32-5. On line at www.educause.edu/pub/er/review/reviewArticles/32132.html The original paper, "Transcopyright: Pre-Permission Republishing", may be found at http://www.sfc.keio.ac.jp/-ted/transcopyright/transcopy.html for Virtual This is the definitional paper for transcopyright permission. "Transcopyright: A simple legal arrangement for sharing, 37. Theodor H. Nels re-use and republication of copyrighted material on the Net." In Takashi Masuda, Yoshifumi Masunaga and Michiharu Tsukamoto (Eds.), Worldwide Computing and Its Applications. Springer-Verlag, Berlin, 1997. ISBN 3-54063343-X. Pp. 7-14. This expresses the transcopyright ideas. 38. Peter J. Denning, "ACM INTERIM COPYRIGHT POLICY", Version 3, 12/18/98, currently at http://www.acm.org/pubs/copyright_policy/. This includes an informal endorsement of transcopyright by the official world-wide computing society (ACM). Unfortunately the ACM does not actually implement transclusive portional delivery as the idea requires. 39. Terje Norderhaug and Juliet M. Oberding, "Designing a Web of Intellectual Property." At http://www.igd.fhg.de/archive/l 995_ www95/proceedings/papers/95/webip.htm I This paper argues that rightsholders are irrational to resist "inline linking" (transclusion without permission), saying this is holding back "technology". It appears to take the argue that copyright law is obsolete and should merely go away. PHILOSOPHY OF HYPERTEXT page 140 Moving the Xanadu Model to the Internet From 1994, moving the Xanadu model to the Internet has been a central (See also next category, "Transquotation, aspect of my work. Transpublishing, and the Keio Hypertransaction System.") 40. Theodor H. Nelson, "Literature to Last: Design for a Universal Digital Medium." In Ute Hagel (ed.), Labile Ordnungen: Netze Denken, Kunst Verkehren, Verbindlichkeiten. Hans-Bredow Institut, Hamburg, 1997. ISBN 387296-0849. Pp. 98-102. This article is a preliminary update of the Xanadu structure and concepts, moving them into the Web era. [Also in this category, and included here as part of Appendix I: 59. Theodor H. Nelson, "Deep Hypertext: The Xanadu Model". 60. Theodor H. Nelson, "The New Xanadu Structure for the Web,". 61. Theodor H. Nelson, "Transcopyright for the Web". 62. Theodor H. Nelson, "The Virtual Literary Format". 63. Theodor H. Nelson, "Problems of the Virtual Literary Format".] Transquotation, Transpublishing, and the Keio Hypertransaction System As an approach to putting the Xanadu model on the Web, we have worked at Keio University to allow quotation on ordinary Web pages. These are some of the papers. 41. Transpublishing "poster" page at http://www.sfc.keio.ac.jp/~ted/TPUB/TrranspubPoster .html This is intended to show visually the concept of transpublishing. 42. T.Nelson, "Transquotation Demo", at http://www.sfc.keio.ac.jp/--tedffPUB!fQdemo.html This gives specifics of Keio permission formats. 43. Theodor H. Nelson, "Transpublishing: A Simple Concept", 1999, at http://www.sfc.keio.ac.jp/--tedffPUB/TPUBsum.html This is a brief statement, intended for the public, explaining the transpublishing concept (Xanadu-style transclusion) as it could be applied PHILOSOPHY OF HYPERTEXT page 141 to the Web. These ideas are continued in (44). 44. Theodor H. Nelson, "Transpublishing for Today's Web: Our Overall Design and Why It Is Simple". Working paper for Keio-JIPDEC Hypertransaction Project, 1999). At http://www.sfc.keio.acJp/-ted!TPUB/TQdesign99 .html A working paper on transpublishing for the Web, mainly for the working group at Keio University, presenting the details of the design ideas in (43)especially the (hypothetical) TQtag and (workable) TQstring for implementing the Keio Format of text quotation. Interactions with Microsoft and Netscape browsers are key issues. 45. Theodor H. Nelson, "Transpublishing Structures for Today's Web". Working paper for Keio-nPDEC Hypertransaction Project, 1999. Available at http://www.sfc.keio.ac.jp/-ted!TPUB/tpubstrux.html. Included here as Appendix H. Briefer than (44), this presents the (hypothetical) TQtag and (workable) TQstring for implementing the Keio Format of text quotation. 46. Yoshihide Chubachi et al., "HTS" page at green.iprs.sfc.keio.ac.jp/-hts/ This is the entrance page for the transpublishing server created by the Keio Hypertransaction Project under Mr. Chubachi's supervision. OSMIC OSMIC is a simple miniature of the Xanadu versioning model, implemented at Keio University by Ken'ichi Unnai. 47. Theodor H. Nelson, "OSMIC: Open Standard for Media lnterConnection" (1996). At http://www.sfc.keio.ac.jp/-ted/OSMIC/OSMI Cd 1m.html This is the original OSMIC defining document. It discusses the hypertime model and the Xanadu-type "trouser link". 48. Theodor H. Nelson, "Virtual Editing and Microversions: OSMIC" at www .sfc.keio.ac.jp/~ted/OSMIC/OSMICpage.html This shows the microversioning strategy of the OSMIC demo. PHILOSOPHY OF HYPERTEXT page 142 49. Theodor H. Nelson, "Models of Time, Backtrack and Groupware" at http://www.sfc.keio.ac.jp/~ted/OSMIC/osmicTime.html The general concept of hypertime backtracking- forward and sidewaysas implemented in the OSMIC demo. 50. Theodor H. Nelson, "The Two Spools of OSMIC", at http://www.sfc.keio.ac.jp/~ted/OSMIC/osmicTwoSpools.html Explains the Xanadu model of text editing as implemented in the OSMIC demo, explained in terms of two spools: the accumulating text, and the accumulating list of user operations. 51. Theodor H. Nelson, "Installing the OSMIC Prototype", at http://www.sfc.keio:ac.jp/~ted/OSMIC/osmicInstall.html Instructions for installing the OSMIC prototype programmed by Ken'ichi Unnai. 52. Theodor H. Nelson, "Using the OSMIC Prototype", at http://www.sfc.keio.ac.jp/-ted/OSMI C/osmicUse.html Instructions for using the OSMIC prototype programmed by Ken'ichi Unnai. CosmicBook™ CosmicBook appears to be the first publicly available product offering transpointing windows. 53. Download page for transpointing windows prototype at ted.hyperland.com!TPWdemo/ This is the original demonstration of transpointing windows done by the author and Ian Heath. 54. Caerdroia, the Journal of Mazes and Labyrinths, published by Jeff Saward, caerdroia@dial.pipex.com; produced by Labyrinthos, the labyrinth research center, photo library and archive, 53 Thundersley Grove, Thundersley, Benfleet, Essex, SS7 3EB, England. On line atwww.labyrinthos.net This magazine gave permission to use their material for the original demo of transpointing windows (53) on which CosmicBook is based. 55. CosmicBook(tm) PHILOSOPHY OF HYPERTEXT page 143 http://xanadu.com/cosmicbook/ This is the download page for CosmicBook parallel documents. Proportional Micropayment and Hypercoin™ The Xanadu model of proportional payment assumes fine divisibility for quotation. No deployed on-line payment system currently has sufficiently fine granularity. 56. "HyperCoin Executive Summary", at www .xanadu.com/hcoin/HcoinSum9906.html This explains briefly why micropayment is important, giving especially the specs designed for the HyperCoin(tm) Project. 57. Theodor H. Nelson, "Transpayment Methods", at www .sfc.keio.ac .jp/~ted!fPUB/TransPayMethods.html This discusses payment methods suitable for proportional micropurchase. 58. Theodor H. Nelson, U.S. patent 6,058,381: "Many-to-Many Payments System for Network Content Materials", 2000. This is a micropayment scheme designed for proportional payment under transcopyright. The VLIT Format The VLIT format is inteded as a Web-compatible file model, allowing people to re-use and quote in any new context and any quantity, with delivery from the originator. Unfortunately the current standards make this extremely difficult, so that overcoming those standards is the main problem. 59. Theodor H. Nelson, "Deep Hypertext: The Xanadu Model", 2001 at http ://xanadu.com/xuTheModel/ Included here as part of Appendix I. Explains the Xanadu Model. 60. Theodor H. Nelson, "The New Xanadu Structure for the Web," 2001 PHILOSOPHY OF HYPERTEXT page 144 at http://xanadu.com/nxu/ Included here as part of Appendix I. Discussion of Web-compatible Xanadu structure. 61. Theodor H. Nelson, "Transcopyright for the Web", 2001 at http://xanadu.com/tco/ Included here as part of Appendix I. A brief introduction to transcopyright for the Web. 62. Theodor H. Nelson, "The Virtual Literary Format", 2001 Included here as part of Appendix I. The intended alternative file format for putting the Xanadu model into a standard browser. 63. Theodor H. Nelson, "Problems of the Virtual Literary Format", 2001. Included here as part of Appendix I. Problems of XML for implementing (62). Hypertext in General From a single declared practitioner in 1965, hypertext has become a field with vast amounts of publication, on line and off. A very few noteworthy items are: 64. Bush, Vannevar, "As We May Think". Atlantic Monthly, July 1945; posted at http://www.theatlantic.com/unbound/flashbks/computer/bushf.htm This article is generally taken as the starting point for hypertext. However, it should be noted that Bush's notion of hypertext was based on transclusion rather than linking. 65. Theodor H. Nelson, "Hypertext Notes" (1967). At http://www.sfc.keio.ac.jp/-ted/ARCHIVE/htnO.tif http://www.sfc.keio.ac.jp/-ted/ARCHIVE/htnl.tif http://www.sfc.keio.ac.jp/-ted/ARCHIVE/htn2.tif http://www.sfc.keio.ac.jp/-ted/ARCHIVE/htn3 .tif - -- - ----- -- PHILOSOPHY OF HYPERTEXT page 145 http://www.sfc.keio.ac.jp/-ted/ARCHIVE/htn4. tif http://www.sfc.keio.ac.jp/-ted/ARCHIVE/htn5 .tif http://www.sfc.keio.ac.jp/-ted/ARCIDVE/htn6.tif http://www.sfc.keio.ac.jp/-ted/ARCIDVE/htn7.tif http://www.sfc.keio.ac.jp/-ted/ARCHIVE/htn8.tif http://www.sfc.keio.acjp/-ted/ARCHIVE/htn9.tif http://www.sfc.keio.ac.jp/-ted/ARCHIVE/htn 10.tif Hypertext ideas as of 1967, when there was no one else in the field. 66. Jakob Nielsen, "Short History of Hypertext", at http ://useit.com/alertbox/history.html A very short list of 18 events, 1945-1995. 67. Jorn Barger, HyperTerrorist's Timeline of Hypertext History, 1996. At http://robotwisdom.com/web/timeline.html A list of about a hundred events, 3000 B.C.-1996. 68. Leslie Carr, Wendy Hall, and David de Roure, "The Evolution of Hypertext Link Services." A CM Computing Surveys, Vol. 31, Number 4es, December 1999. Th.is is an excellent survey of deeper hypertext systems based on pooled linkbases (instead of the embedded one-way links of the World Wide Web.) 69. Berners-Lee, Tim, Weaving The Web: The Original Design And Ultimate Destiny Of The World Wide Web. 2000. This gives Berners-Lee's personal history of the Web, and of course is imbued with his point of view. PHILOSOPHY OF HYPERTEXT page 146 ZigZag® and FloatingWorld™ The ZigZag system is a radical redefinition of all abstract structure with a principled nonhierarchial system of elements. In computer terms, this means the possibility of doing away with files, applications, monolithic programs and the difficulty of programming, creating instead a unified crystalline structure of units connected at different angles. As a connective system for data and programming, it offers a possible new universe of simplicity, where user functions and data become a unified, visualizable environment. 70. Theodor H. Nelson, "What's On My Mind", at http://www.sfc.keio.ac.jp/-ted/zigzag/xybrap.html This is an introduction to the ZigZag structural mechanism. 71. ZigZag home page http://xanadu.com/zigzag/ This is the main page for ZigZag structure. 72. ZigZag tutorial http://xanadu.com/zigzag/tutorial/ZZwelcome.html This is an introduction to ZigZag structure, with illustrations. 73. The ZigZag prototype (version 0.70, programmed in Perl by Andrew Pam and Bek Oberin) may be downloaded from the page ted.hyperland.com/zigzag/zz070.txt This is best used by Linux adepts. 74. Theodor H. Nelson, "How to Make the ZigZag Boot Floppy", at http ://xanadu.com/zigzag/zzBootFloppy.html Instructions for making the ZigZag boot floppy, which can run on a standard Intel-type machine (loading in bare-metal mode). 75. Keyboard commands for use of the ZigZag prototype http ://xanadu.com/zigzag/zzDirexCondensed.html This gives the standard keyboard commands for using ZigZag . ---- PHILOSOPHY OF HYPERTEXT page 147 76. Genealogy demo download at ted.hyperland.com/zigzag/famdemo/zigzag.zz This demo data, for use with the ZigZag boot floppy or Perl prototype, shows the use of ZigZag structure for representing genealogy; instructions at (77). 77. Genealogy demo instructions at www.xanadu.com/HolmFamilyDemo.html This is a walkthrough of the family demo (76). 78. Theodor H. Nelson, "Floating World: a New User Environment", at www.xanadu.com/FW99/ The 1999 ZigZag vision paper. This is the design document for a sweeping new user environment to be based on ZigZag structure. These are the Floating World specifications for a whole new interactive world, to be based on a ZigZag engine, as of November 1999. (The system was then called ZX). 79. Les Carr's ZigZag-type pages http://www.ecs.soton.ac.uk/~lac/zigzag/ These show an alternative implementation of ZigZag-like functionality, data representation in a browser; implemented by Prof. Carr in using XJ\.1I., Javascript. 80. The Gzigzag project at http://www.gzigzag.org/ This is an interesting and highly motivated version of ZigZag as implemented in Java under the direction of Tuomas J. Lukka. (He is currently using the ZigZag trademarks with my permission.) 81. Downloadable of Gzigzag at http://www.gzigzag.org/download.html This Java-based system works best under Linux, is said to be stable on the Macintosh, has more problems in Windows. 82. Theodor H. Nelson, U.S. patent 6,262, 736: "Interactive Connection, Viewing, and Maneuvering System for Complex Data." U.S. Patent No. 6,262,736, 2001. This is the basic U.S. patent for the ZigZag system. PHILOSOPHY OF HYPERTEXT page 148 P.chology, Artificial Intelligence and Knowledge Representation The 1958 Schematics paper may have foreshadowed, in peculiar ways, today's burgeoning field of knowledge representation. Knowledge representation is in tum expected to be the center of artificial intelligence. This is all rather peculiar. 83. J.S. Bruner, J.J. Goodnow and C.A. Austin, A Study of Thinking. Wylie, 1956. This pioneering work essentially redefined cognitive psychology, previously a behaviorist and Piagetian enterprise, in terms of structure and strategic payoff. 84. Michael W. Eysenck and Mark T. Keane, Cognitive Psychology. Lawrence Erlbaum Associates Ltd, Publishers, 1990. This is an excellent survey of the field of cognitive psychology as of 1990. 85. Raymond Kurzweil, The Age ofIntelligent Machines. MIT Press, 1990. 86. Ronald J. Brachman and Hector J. Levesque, Readings in Knowledge Representation. Morgan Kaufman Publishers, 1985. 87. Patrick J. Hayes, "Some Problems and Non-Problems m Representation Theory". Proc. AISB Summer Conference, U. Sussex, 1974. Classic article discussing what can and cannot be represented. 88. Ronald J. Brachman, "On the Epistemological Status of Semantic Networks". In N.V. Findler (ed.), Associative Networks: Representation and Use of Knowledge by Computers. Academic Press, 1979. 89. William A. Woods, "What's in a Link: Foundations for Semantic Networks". From Bobrow and Collins, Representation and Understanding: Studies in Cognitive Science (Academic Press, New York). Distinguishes between structural links (which set up the parts of a proposition or description) and assertional links (which assert something within the structure). 90. Marvin Minsky, "A Framework for Knowledge Representation". In J .Haugland (ed.), Mind Design, MIT Press, 1981; foundation of Minsky's later The Society of Mind. PHILOSOPHY OF HYPERTEXT page 149 Introduces the "frames" concept, now central to Al. 91. Martins, J.P. and Shapiro, S.C., "Reasoning in Multiple Belief Spaces". Proc. IJCAJ-83, Karlsruhe, 1983. 92. Anthony S. Maida and Stuart C. Shapiro, "Intensional Concepts in Propositional Semantic Networks. Cognitive Science, 1982, 291-330. 93. Hubert L. Dreyfus, "From Micro-Worlds to Knowledge Representation: Al at an Impasse". In J .Haugeland (ed.), Mind Design, MIT Press, 1981. An excellent overview of knowledge representation and artificial intelligence as they developed in the 1970s. 94. A. Newell, "The Knowledge Level". Artificial Intelligence, 1982, 87-127. A survey of the views in knowledge representation as of that time. 95. Martin, W.A., "Descriptions and the Specialization of Concepts." In P.H. Winston and R.H. Brown (eds.), Artificial Intelligence: an MIT Perspective, vol. 1. MIT Press, 1979. 96. Zadeh, L.A., "PRUF-- A Meaning Representation Language for Natural Languages", Memo No. ERL-M77/61, Electronics Research Lab, College of Engineering, U.C. Berkeley, 1977. 97. Stottler Henke Associates, Inc., "A Chronology of Artificial Intelligence." At http://www.shai.com/ai_general/history.html This is a useful short summary of events in Al, starting with an "expert system" on head wounds discovered on an Egyptian papyrus of 3000 B.C. 98. M. Ross Quillian, "Word Concepts: A Theory and Simulation of Some Basic Semantic Capabilities." Behavioral Science, 1967, 410-430. Quillian is credited with introducing the concept of "semantic net". This is a summary ofhis 1960 Ph.D. thesis. 99. Tim Berners-Lee, James Hendler and Ora Lassila, "The Semantic Web". Scientific American, May 2000; available at http://www.sciam.com/2001/050lissue/050lberners-lee.html This offers Berners-Lee's sweeping vision of a machine-explorable network of meaning that he believes will supplement the Web as we know it and perfect access to information. PHILOSOPHY OF HYPERTEXT page 150 Other Sources What follow are a number of other sources bearing on this thesis. 100. David Baltimore, personal communication, 2000. In this conversation, Nobel laureate Baltimore agreed that "solving the genome" led only to knowing the start-and-stop points of the various strips of genetic information. 101. Lauren Hammond, personal communication, ca. 1970. This extremely enjoyable conversation took place when Mr. Hammond was a passenger in the taxicab I was driving in New York City. He died shortly thereafter. 102. Michael A. Hiltzik, Dealers ofLightning. HarperBusiness, 2000. This is essentially the official biography of Xerox PARC. While it presents the standard view that PARC was a magnificent play-pen for the finest computer scientists in the worl. it is possible also to understand its harsh and abrasive atmosphere, and the ways in which founder Bob Taylor guided the group to a world of paper simulation. 103. Alfred Korzybski, Science and Sanity: An Introduction to Non-Aristotelian Systems and General Semantics, 1933 (5th Ed., 1993). New York: Institute of General Semantics. The defining document of Korzybski's General Semantics. 104. Thomas Jefferson, "The Unanimous Declaration of the Thirteen United States of America." Philadelphia, 1776. Jefferson's earlier draft obtained from http://odur.let.rug.nl/-usa/D/l 776- l 800/independence/doi.htm (Presented by Ka-Ping Yee via xu88 software in Fig. 10). A key example of parallel versions. 105. Brewster Kahle, personal communication, ca. 1999. In this discussion with the author, Brewster Kahle, founder of the Internet Archive, proposed to imitate transclusion using "trigram" search-- searching for triples of characters-- to find substantial string identities across the entire space and time of the Web. We did not get into extensive design discussion, however. PHILOSOPHY OF HYPERTEXT page 151 106. J. Kawakita, The Original KJ Method, Kawakita Reaserch Institute (1982) 44pp. The defining document for the KJ method of free-form idea work leading to document structure. 107. Theodor H. Nelson, "Clearing the Boulders from the Catacombs: Principled Alternatives to the Tradition ofNarned Stuck Files in Hierarchical Directories", at www .xanadu.com!fYRF/BoulCat.htrnl This briefly discusses the lurking computer tradition of hierarchical directories and lump files, and its deleterious effects throughout the computing world. 108. Theodor H. Nelson, "Embedded Markup Considered Hannful." In XML: Principles, Tools, and Techniques (World Wide Web Journal 2:4, fall 1997). Table of contents at http://www.w3j.com/xml/ Complete article at http://www.xml.com/pub/a/w3j/s3.nelson.html This articulates the view that such formats as SGML, HTML and XML are fundamentally mistaken, as they pollute the data and artificially impose hierarchy. 109. Theodor H. Nelson, "The Tyranny of the File." Datamation, 15 December 1986. This explains that conventionalized file structure is a tradition which has shaped most computer usage, and perhaps even warped most computer usage. 110. Jakob Nielsen, "The Death of File Systems", 1996. Online at http://www.useit.com/papers/filedeath.htrnl. This is the only piece of writing I have found in which some other computer professional criticizes the prevailing paradigm of file systems and structure. 111. H. Ohiwa, K. Kawai, M. Koyama: "Idea Processor and the KJ Method," J. Information Processing 13 (1990) 44-48. This presents software to support the KJ method of free-form idea work. PHILOSOPHY OF HYPERTEXT page: 152 112. H. Ohiwa, N. Takeda, K. Kawai, A. Shiomi: "KJ editor: a card-handling tool for creative work support," Knowledge-Based Systems 10 (1997) 43-50. This presents software to support the KJ method of free-form idea work. 113. JeffRulifson, personal communication, 1999. In this conversation, Rulifson took personal credit for the renaming of "cut" and "paste". (Compare 117.) 114. Daniele C. Struppa, personal communication, 1998. Prof. Struppa explained to me in this conversation that microlocal analysis is a completely general representation for information dealt with traditionallly by the familiar notations and methods of the calculus. 115. Yuzuru Tanaka, "A Synthetic Media Architecture And Its Prospects." At http://nrdf.meme.hokudai.ac.jp/ip/bib/smaSec.html This describes IntelligentPad, Prof. Yuzuru Tanaka's ingenious component software system, designed for easy programming by untrained users. It involves the visual stacking of functional units ("pads") on top of one another. 116. Yuzuru Tanaka, "From Augmentation Media to Meme Media: Towards a New Knowledge Media Architecture". At http://www.genome.ad.jp/manuscripts/GIW95/IL/GIW95I05.html Prof. Yuzuru Tanaka presents a scenario for IntelligentPad, his component software system, to create a new form of evolution. He wants the IntelligentPad structure to develop into an independently evolving universe of ideas, instantiating Dawkins' "meme" concept. 117. Larry Tesler, personal communication, 1998. In this conversation, Tesler took personal credit for the renaming of "cut" and "paste". (Compare 113.) 118. P.N. Johnson-Laird, Mental Models: Toward a Cognitive Science of Language, Inference and Consciousness. Cambridge University Press, 1983. - - - -- - - - - - -- - - PHILOSOPHY OF HYPERTEXT page 153 Acknowledgments PHILOSOPHY OF HYPERTEXT page 154 Acknowledgments This work could not have taken place without the help, moral support and good will of many people. There is only room to mention those who contributed either emotionally or directly to the work itself. Most important in the last decade have been the help, warmth and inspiration of my collaborator and life partner Marlene Mallicoat, who has worked awesomely to make the continuation of the Xanadu work, and now this thesis, possible. The others I must list here in reverse chronological order. For the last five years, the generous support and understanding of Keio University have enabled this work to go on. In Japan over the last six years, the friendship and support (in the order I met them) of Yuzuru Tanaka, K.Ohkubo, Edward Harter, Kay Nishi, Hajime Ohiwa, Nobuo Saito, and Kenji Naemura have added cheer and life to these proceedings. Numerous students at Keio University have especially participated in this work. I must especially mention the work of Yousuke Igarashi, who wrote the first transpublishing script; Ken'ichi Unnai, who implemented the OSMIC microversioning system; and Yoshihide Chubachi, who implemented the Hypertransaction system. I am sorry that at this moment I am unable to sort out the numerous others who have contributed in various ways. Since 1992, Andrew Pam of Xanadu Australia and Katherine Phelps of Glass Wings Press have been special friends and collaborators at many levels of this project. PHILOSOPHY OF HYPERTEXT page 155 Since 1988, Stan Dale, founder of the Human Awareness Institute, has been an inspiration, friend and deep supporter, whose contributions to my life I cannot begin to enumerate here. Throughout the years, from the nineteen-seventies, my U.S. collaborators on the Xanadu Project have been boon companions and inspiring teachers. Some were best friends who became collaborators; some were collaborators who became best friends. One must I now single out for particular appreciation and gratitude: Roger Gregory, whose devotion to these our shared ideals is matched only by his cantankerous brilliance. The intellectual companionship and moral support of my son Erik Nelson, and the friendship and support of his mother Deborah Stone, have been a great blessing through nearly all of the Xanadu years. Four educators had a particularly strong personal and intellectual influence on me: my great-grandfather, Edmund Gale Jewett, provided both his science teaching and his poetry, a sense of human rightness and creativity at the center of an awesome world teeming with life and pain; Leo Rosten, whom I knew as a family friend and who told me many things on many subjects; Michael Scriven, whose sparkling intellectual swordsmanship (and instant understanding of whatever I said) enlivened my college years; and Thomas C. Schelling at Harvard, who combined nuanced and multilevelled strategic insight with the ability to express it in few and powerful words. From my birth to that turning point in 1960, and then for the remainder of their lives, my grandparents Theodor Holm and Jean Parke Holm were a source of strength and inspiration. And of course my parents, Ralph Nelson and Celeste Holm, provided inspiring examples and remarkable lessons in media.