Central Saint Martins, London · dissertation, 2001 · recovered from a Macromedia Director projector and published, August 2026
Dissertation
Tree Beard
Tracing web navigation as a tree grown in three dimensions
Central Saint Martins College of Art and Design,
London.
2001
treebeard · a Director projector for Classic Mac OS
Web browsers of 2001 recorded history as a flat list and offered a Back button, discarding the three cues by which people actually re-find things: time, space and movement. This dissertation proposes a companion program that traces a browsing session as a tree you can turn and inspect in three dimensions. Each site visited becomes a node; the routes between them become branches whose length encodes time spent and whose width thickens with repeated use, so habitual paths grow into a trunk. Leaves mark favoured sites, fruit signals that a site has changed, unwanted growth is pruned, and, seeded from a keyword, the tree grows on its own. The argument is developed through memory and recollection, the linear nature of language, growth and emergence, cartography, and information-foraging theory.
Browser history keeps a list of where you ended up. This was an attempt to keep the journey instead, and to let it grow into something you would recognise on sight.
I made this at Central Saint Martins in 2001, in the final year of a BA in Graphic Design. It wasn’t handed in as a document; it was software. You read the essay by clicking nodes rather than turning pages, because I was arguing that thinking isn’t linear, and a linear essay felt like the wrong shape for saying so.
I found it this year while going back through my archive for the IDEA magazine piece on twenty years of working with my brother. It was on one of a stack of old drives, in a program only an old Mac will still open. The text has been pulled back out of the file and edited for reading: spelling, punctuation, a few sentences that ran on, and three claims I would not make now. Everything that changed is recorded.
I’m publishing it because the idea still interests me.
Looking back
Written in 2026.
How it survived
It survives as a single file called treebeard: a self-running Macromedia Director projector for Classic Mac OS. It isn’t a document about an interactive piece; it is the piece. It still runs on the old Mac, which is where the film on this page comes from. What I can’t do is run it on anything modern, or open the authoring files it was made in.
It was my final dissertation, so it belongs to the end of a course that ran from 1998 to 2001. I could not have told you that from memory. It feels much longer ago than it is.
The text and the code both came back out of it. Thirty-six scripts, ten of them mine and still carrying the names I gave them. One of the ten is the engine that draws and turns the tree: Director 8 had no 3D of its own, so the perspective is worked out and drawn by hand, in Lingo.
What I was proposing
A summary.
The problem. In 2001 a browser offered two ways back: a Back button and a list of addresses. Both discarded what people actually navigate by. We remember a place in relation to other places, to the route taken, to how long we stayed. A list of addresses records that you were somewhere without recording anything about the going. The evidence I gathered was not marginal, though every figure in it is now a quarter of a century old, and it has not kept equally well. So here it is graded.
- Source held Roughly 58% of navigation was returning to pages already seen, and the Back button accounted for some 30% of everything done in a browser. Tauscher and Greenberg, and the paper is still where it was.
- Source probable 17% of surveyed users could not find sites they had visited before; 87.7% were browsing with no goal at all. Almost certainly one of Georgia Tech’s GVU surveys. They were the only surveys of that size running then.
- Source lost Which one it was. My essay calls it “a global WWW usability study” and nothing more, which isn’t enough to go and find it. I don’t remember which one I read.
The shape of it held either way: most of what people did online was trying to get back to something, with tools poorly suited to it.
The proposal. A companion program alongside the browser that draws a session as a tree growing in three dimensions. Not a map of the web but a trace of one person’s movement through it; I was insistent that mapping a network which won’t hold still is futile.
The part I would put first now, and did not put first then: it was never meant to be a diagram. It was meant to be alive. It grows while you work, thickens where you go often, fruits when a page changes, and gets pruned when you have finished with a branch. A branch can be taken as a cutting and replanted in another tree. A tree can be seeded from a single word and left to grow while nobody is watching. Drawing your history as a tree turns out not to have been new. Giving it a life was the part I have still not found anywhere else.
The mechanism, in full. Each site becomes a node, carrying its name, dates, visit count, time spent, and where that is not enough to jog memory, images and fragments lifted from the page itself. Branches record the route from one site to the next, so the structure holds paths rather than events. Height encodes when. Branch length encodes time spent, damped by the stated equation length = c × e−kt so a long session never outgrows the frame. Branch width encodes habit: a branch thickens with the number of nodes it carries, so well-travelled routes become a trunk and one-off detours stay thin at the periphery. The rule comes directly from the brain, where a synapse strengthens with repeated use. A site reached again by another route is never duplicated; a translucent link is drawn instead. Leaves are placed by hand to mark favoured sites. Fruit appears by itself when a site has changed, removing the need to keep checking. Growth can be pruned, a useful branch taken as a cutting and replanted, and a tree seeded from a keyword and left to grow unattended.
What it is like to use. I start reading about cartography, and a node appears. I follow a link to Tufte, and a second node appears, joined to the first. From there I wander into information visualisation and lose an afternoon in it, so that branch grows long. Two days later I want the Tufte page again. I never bookmarked it and I cannot remember the address. What I do remember is that it came off the cartography reading, early, before the long afternoon. On the tree that is a short stub near the trunk, just before the branch that runs away into the distance. I do not have to read anything to find it. I turn the tree and it is there.
That is the entire argument in a paragraph, and my essay never makes it. It specifies the machinery thoroughly and leaves you to assemble the experience yourself.
The case I built around it. Five chapters do the theoretical work. Memory and recollection argues that human recall is associative rather than sequential, so a history should be shaped like a network. Complexity observes that the web already resembles a nervous system in scale, and proposes the tree as an external memory, and further as a record of how a train of thought developed. Linear language argues that speech and writing force non-linear thinking into sequence, shaping how we think and not merely how we speak. Growth, by way of Kevin Kelly, argues that a system modelling a living network must itself grow and be capable of being cut back. Cartography and foraging close the case: maps describe stable territory and the web is not stable, so tracing beats mapping; and by way of Pirolli and Card at Xerox PARC, people are cast as informavores hunting information as animals forage, weighing what a source costs against what it yields.
What became of it. The thing I was proposing never arrived: a persistent, personal, weighted trace of your own path through information, in which habit shows as structure. Browser history in 2026 remains a flat searchable list. The problem was not solved so much as routed around.
What I did next. I kept going with it. Two years later I took an MSc in Virtual Environments at UCL and built the tree properly, to test it: the same structure shown on a monitor and through an augmented reality headset, timed, with users finding folders. The headset was forty per cent faster than the screen. Everyone preferred it. And both were far slower than the plain file browser they already knew. The interesting part turned out to be the display, not the drawing.
What I didn’t know at the time
The idea did not come out of the literature (see the reading list). It came out of making things. I had been teaching myself Flash ActionScript and Director Lingo, building interactivity and generative systems to find new ways of making, and the tree came out of that: something I built my way into rather than read my way into.
Which is also why I did not know there was a field. I got there through a close friend, Chris Lawrence, who was studying computer science at UCL. He listened to me describe what I was making and told me it had a name, and a literature, and people who had been at it for years. Everything in my bibliography I found after that conversation. My reading list is the map of what one friend pointed me towards.
What I found was that the visual history problem already had its own research, most of it out of Xerox PARC and a handful of university labs: Lamping and Rao’s hyperbolic browser (1996), Andrews’ Harmony browser (1995), Narcissus (1995), Card’s WebBook and Web Forager (1996), Cockburn’s WebView (1999), Tauscher and Greenberg on revisitation (1997).
Since then I have found it ran closer still. Peter Dömel’s WebMap (1994) drew a browsing journey as a graphical map, and Ayers and Stasko’s MosaicG (1995) modified the Mosaic browser to build a tree of visited documents as you went. Most pointedly, PadPrints (1998), a browser companion building a zoomable tree of visited pages with thumbnails on the nodes, already encoded dwell time, visit frequency and recency, and measured people refinding pages in 61% of the time it took them without it. I cite none of those three, because I never found them. Software was ahead of me too: Netscape shipped RSS 0.90 in March 1999, two and a half years before this was built, and TheBrain had been drawing personal knowledge as a navigable graph since early 1998.
So none of the individual mechanisms were new, and I am not claiming them. What I think was mine is the frame around them: the tree not as a diagram but as an organism with a lifecycle, pruned and cut back, taken as a cutting and replanted, cross-bred, seeded from a keyword and left to grow while nobody is watching. The systems I found visualise what you did. Mine was trying to be something that lives and acts.
What strikes me most now is what became of all of it. The field published working systems and measured real gains, and browsers shipped almost none of them. Which makes the conclusion I reached, that navigation had been left to a monopoly and needed independent hands, the part that has aged best, and not for a reason I would have wanted.
How it worked
The tree was never flat. You turned it. Reading the essay back I noticed I never say so anywhere, which looked like an omission until I remembered that nobody read it the way you are reading it now. You were inside the tree the whole time. It was the interface, not an illustration, and you do not describe the room you are standing in.
Nodes nearer the front are drawn larger and more strongly coloured; the ones behind fall away small and pale. Branches that would sit on top of one another in a flat drawing come apart as soon as you move it.
The tree at the head of this page is that same engine, ported out of my recovered Lingo. One thing the code settled: the essay says a branch thickens with the number of nodes it carries, but the program never did that; a branch is drawn thick or thin according to which way it runs. The proposal described a system more careful than the prototype I actually built.
The rest of what survives is photographic: the title screen, the tree itself, an essay node open for reading, my working folder, and a related experiment sitting alongside it.
The essay
Written 2000–01. Edited for reading.
Central Saint Martins College of Art and Design
Tree Beard
Tracing web navigation as a tree grown in three dimensions
J. Hudson-Powell
2001
Introduction
Do we interface and gain the most of the internet's vast capabilities or are we only utilising the surface of this remarkable information source? There are so many specified fields of study within this subject that its relevance spans a huge range of topics, ideas and theories. I have defined my range of study to concentrate on the proposal of a web tracking system and its relation to information, communication and interaction. The essay works from the basic notion that you understand the whole by understanding its individual components. The proposed system answers the question in terms of usability and brings into the picture the downfall of current navigation programs and the limits they set upon us as web users. These limitations in terms of navigation are a central problem between complete ease of access to information and communication, and the limited navigation we enjoy at present.
The problem
The web is the fastest growing communication medium and has grown at a rate which exceeds the expansion of the radio or television. It took thirty-eight years for the radio and thirteen for the television to reach 50 million users, it only took four for the internet to catch on to the same degree. This is a relatively short period of time for the adaptation and development of the user interface. Usability is fundamental to all information-based media, whether technological or not. Information and communication are the two main attributes of the web. Trade or any other possibilities are redundant without complete ease of access to its fundamental properties. The web's ability to store information is immeasurable as it is distributed globally, but more importantly it is stored electronically. The access and retrieval of this information is flawed only by the interface with the web. We perceive or visualise the internet through web browsers and web sites. Web sites on one level give or limit as much informational content as the creator/ designer allows in terms of content and site structure. But the problem is, no matter how great the site's content, design, or structure of it is confined within a browser. At present, as a web user you only have two options; either Microsoft's Internet Explorer or Netscape's Navigator. So usability is defined by the browser program and its interface. The program also defines navigation and this is where I feel the real problems lie. According to a global WWW usability study the browsers have problems with this:
17% of subjects report that they were unable to find previously
visited sites.
8.8% of subjects report not being able to visualise where they have
been and where they can go.
6.4% of subjects report not being able to determine where they are.
87.7% of subjects report browsing in an opportunistic manner,
i.e. by following links from page to page with no
premeditated goal.
These figures, coupled with the fact that 58% of all navigation is actually revisiting previously seen sites and that the use of the browser's 'Back' button accounts for 30% of all actions made, prove this underlying need for a capable history system. At present history facilities within the current browsers use two mechanisms:
1. A history menu that contains the last URL (sites) addresses visited, with newer sites appearing on top. The main problem with this stack system is that sites are easily pushed off the bottom of the stack and information is lost when backtracking or browsing new branches off previously visited sites.
2. A history facility that contains a searchable history of all URL addresses visited from installation with date and amount of times visited.
There is a huge underlying problem with both of these systems, which is the lack of visualisation. These lists of textual information lack the cues to allow any real recollection of a site. Web browsing is sporadic and intense. This burst mode of browsing doesn't allow a familiarity of every site, so it is important to understand where less familiar sites lay in relation to the more familiar. Neither of these mechanisms rely on or allow scope for memory and recollection. The stack system records all sites visited, this may seem to be problem solved, but it neither gives nor relays any of the most vital notions of navigation: time, space and movement. An understanding of all three of these aspects allows re-navigation. The stack system gives a rigid sense of linearity it portrays none of the non-linear browsing behaviours of taking multiple paths from one source point. Without these paths and connections finding sites relative to other sites is made into an unnecessarily laborious task.
The solution / function
The development of the program I propose has been raised as a direct solution to the problem of re-navigation, which I feel, is one of the major problems with current web browsing applications. This is a solution to more than just the subject of web re-navigation, the program pulls on many different subject matters and brings in to question others issues. Cartography, data visualisation, growth, memory and the linear/non-linear nature of communication are the main areas used to answer the question of a greater ease of navigation. The tree, I feel, will create a greater freedom of overall navigation. The program I propose is represented using a tree analogy. The tree as a visual metaphor is well used for representing progressive data. It allows a great understanding of time, space, movement and decision making. The program won't map but rather trace your movement through the network, recording vast amounts of what would seem abstract data and visualising into an application which would aid a web user. The program would run along side your current browser adding an extra depth to web interaction. The tree comprises of several different functions to build a strong understanding of all the aspects to allow re-visitation and navigation:
-All growth is in a vertical direction and represents the constant variable of time. This allows the user to gauge or cross reference sites, and understand the relationships between sites.
-Nodes are representations of URL addresses (web sites). A new node will appear as a new site is visited. Nodes hold direct links to the web sites they represent. On a mouse rollover a node would produce information like site names, addresses, date created, date modified, total visits and time spent on the site. This is the most basic data needed to identify a site. However, if this info were not enough to allow for a site's recollection then information like images, sounds and sections of text would be drawn from a site automatically when visiting. These attributes are more food for the memory, as these would more likely be the clues to remembering a site.
-Branches are formed to link nodes and to create the tree's structure. Branches become paths between sites allowing an easy route to find the not so familiar sites through the more familiar ones. Branches are gauges of time; the length of the branch below a node represents the total amount of time spent on the site. The tree's growth rate is controlled with a simple algorithm: Length = constant × e^(−constant × time). This means the length of the branches does not grow linearly with time. Growth is initially fast, then slows as time increases, following the natural exponential function. (e is the number 2.718, the natural exponential constant, which has some special mathematical properties.) The result is that the branch does not grow to a size that makes it out of proportion with the rest of the tree. This makes it easier to present this information within a confined space. These branches give the notion of movement and space. If a node is back tracked to and another link taken the source node will have two branches and become a junction. The width of a branch is defined by the amount of nodes that appear above it within that branch structure, i.e. The trunk will grow thick and visually stronger, while smaller branches on the periphery will appear lighter and be less defined. As a site may be visited from many different links and angles it would be likely that a site would be visited many times. No site may appear twice in any one tree so semi-transparent branches link these connections and navigation/growth will continue from the original node.
-Leaves may be placed upon nodes as visual markers allowing the user to highlight favoured sites. Fruit may also grow upon the tree. Fruit will grow on a node when a site has been modified. This will eliminate the need for constant revisiting, therefore making navigation of new sites a more important issue.
-All of the programme's properties translate and correlate all of the primary aspects of navigation to allow and guide a user to any information needed. It quantifies all of these aspects and places them in relation with each other. It singularly groups all of this data into one visualised formation. If you know where one site is it is as simple as following routes and paths to the required site/ information. The tree is also capable of self-growth.
The tree would understand to a basic level browsing behaviour and patterns. It would understand contextual information within its body and grow with a small understanding of sites you may want to visit and information you may want to find. Growth parameters could be set, such as growth rate, connections per site along with key words or subjects. If sites on branches are disliked they may easily be chopped off. This will allow for a constant search for related information and allow a real sense of multi directional paths of navigation.
These same growth parameters may be set with a greater emphasis on a
defined subject matter for a seed. For example at the start of a new project you could type in a brief key word or subject, thus defining the life-task of the germinating seed, and within a minute it would be possible to have a structure which would contain hours of standard browsing time. This data mining would allow for a broadened span of research. The final function would be a search engine to deal with the more complex information structures within the tree. Although this may seem to go against the primary objectives of the tree, the user would be meta-searching a localized information source with the knowledge that the required site/ information does exist within the localized tree structure.
Memory and recollection
We have obtained fluency in communication through experience, repetition and adaptation. This fluency and autonomy is also the product of the way in which our brains abstract and recall information when it is needed. Given an input/ subject the brain will try to understand it by referring to memory. It will also look for associated information or any link it calculates is relevant. It will not follow a linear path from intake of information to recognition. The brain finds subject matter demanded by scanning a huge network of ideas memories and abstractions of information. If the subject is familiar the brain can divulge information quickly and more accurately. This is because the paths it follows to locate stored distributed information are strengthened by repeated use. The less familiar or well remembered a subject, the more the brain will resource distributed information it thinks will help construct an accurate perception.
Any recollection we have has not always been found in its specific location in the brain, but more normally formed from fragments of distributed thoughts. In conversation when presented with a subject, we try to recall any personal experience of it in order to respond. The brain will produce information of varying accuracy and relevance depending on how well its network could provide it with the necessary components. It will also produce related subjects, which were realised by association when routes through the brain were stimulated. In the interaction (discussion for example) these related thoughts provide scope for progress in the conversation. So many of these complex biological functions are imperative to all interactions. The process of information storage and retrieval is how we communicate and utilise information.
Complexity
The physical nature of the internet is central to advantages it has over communication systems or information storage / retrieval systems of the past. The internet is a distributed system it has characteristics of an natural organism. It consists of a huge collection of nodes working autonomously. These nodes are constantly conducting an exchange of information, they also have a high level of connectivity with the rest of the organism. In theory there is no centralised governance. This gives the web an ability to evolve in a similar way to an organic form. Problems overcome by re-routing of information, efficiency can be increased as the system can learn to build on its self. " " The complexity of the web has reached a level which is comparative to the network of the human brain. Connections in this electronic system appear to mimic the synapses in our own biological information storage /retrieval system. In a sense what emerges from the internet could be perceived as a collective consciousness. As noted earlier, most navigation is re-navigation, which points to the same underlying need for a capable history and memory system. The web tracing program I propose will build a structure, which will act like a virtual memory or mind. The latter of these two functions will only emerge through artificial intelligence and the data mining. The notion of the tree structure becoming a virtual memory is not in question. The system visualises every movement you make when on a web browsing session, tracing and remembering all links taken autonomously. The more revisited a site or a subject the more dominant the branch will become visually and structurally. This happens in much the same way as the retrieval and storage in the brain when a connection or pathway strengthens through repeated use. If a subject or site were rarely visited then it would be possible for it to get lost in the complexity of a tree. But I feel we deal with this problem in much the same way as we do when retrieving less frequently needed information in human memory. We use the stronger branches as anchor points, then work through the pathways, linking, cross referencing and remembering. This works on the knowledge that all sites sit within the structure in relation to all the other sites. If you can remember the nearest site or subject to the needed site it is only a matter of following pathways until your site is found. But the tree becomes more than just a re-navigation tool. It becomes a think map, just by looking at this tree it would be possible to view thought processes and how ideas are formed. Nodes become more than just representations of web sites visited, they become these individual sources of information, plotted points on a train of thought. The length of each branch allows you to not just know how much time was spent on a site but also how much information you drew from that particular information source. These are the topics which allow the system to become more than just a web application. But instead it becomes more of a tool and aid to your natural sense of navigation.
"These ever-shifting hyperlinks bear a remarkable resemblance to connections that grow and fade in a human brain. If one neuron in the brain is activated shortly after another neuron, the synapse connecting the two gets stronger. In the end the strength of the connection grows with the degree and rate of activity"
Linear language — non linear thought
Throughout evolution we have formed a vast amount of different types of communication. Communication to communicate thoughts and ideas from individual to individual and from individual to communities. Speech and writing are the two main structural types of communication and both are linear. By linear I mean they use icons which have a set meaning and work in specific sequences. Thoughts, which are constructed in a non-linear way, are converted into the linear (letters, words, sounds) to be passed between individuals in a form of mutually understood encryption. As language has formed the basis of all communication it has allowed us to form relationships with individuals, through to forming governments and multi national relations. Language is the foundation of all civilisations it has allowed the trade of goods through to the trade of views and ideologies. We have been wholly reliant on it for centuries and as a result it has formed the way we think and construct thoughts. As we translate our thoughts to speak and write we have developed a more language oriented thought process. I find I do not always fully understand what I am thinking until I have spoken it. 'Putting your thoughts down on paper' is a commonly used phrase, but it seems that putting your thoughts down is to restrict these thoughts.
Our minds are non-linear; you don't just start a thought in one corner of your mind and continue to the other and finish with a conclusion. So why do we communicate in this fashion, because the basis of all communication is linear. With the start of a line or the start of a conversation we continue till we finish in a direct manner. So we communicate outwardly in this manner but this is not how we process or collect information. We constantly hoard information input from as many sources as possible. Although we seem to start with predetermined goals or subjects, we try to correlate this information the same way we think, linking referencing and analysing and again it still doesn't always become clear until it's written or spoken.
We use these linear communication methods to construct and form ordered thoughts and ideas. We piece together what we are thinking and try to build a solid mesh of fragmented views and factual information to export so that another person may understand enough to interact, in conversation for example. These interactions are complex; they appear simple to us as we have mastered the skill observation, feedback response and improvisation. So it seems that essentially the complexity of our minds, memory, how we interact and respond, is hindered and possibly restricted by the necessity of linear communication.
The internet is linear since you have to start at one point and end at another, but what happens in-between these two points is essentially multi directional. The web in its entirety could be perceived as a global brain, in its physical and non-physical attributes. So will the internet displace the linear fashion in which we process and relay information? On the most specific levels in terms of language (written and spoken words) things have to remain linear. For example sentences only make sense with the words in the correct order. The internet will alter the linear fashion of communication in terms of subject fields or areas of information. Because of the distributed nature of the web, processing information using it takes on unique characteristics, such as greater scope for associated research and more obviously the speed of information and its diversity. Because of this influx of information and the new methods of accessibility, a net logic must be developed. This logic must increase the efficiency of the interaction between the information source and the individual. I say individual because although the logic must be adopted universally it should also be able to adapt to the individual's needs. But as the internet stands it is not this free roaming space it is envisaged to be, it is limited not in its capacity to store information but by its capability to relay information. The way in which we interface the internet and likewise the information it holds, is not in the most efficient manner. Navigation and interface seem to be the only obstacles between complete ease of access to the required information.
Growth
Growth of an organism can be perceived as a continual self-improvement or adaptation over time. Growth patterns can be seen in stages which relate to feedback and control. Progression and self-improvement are responses to variables over the course of time. No living organism may stop the effects of time as much as we try to, so biological growth is inevitable. Technology is different. Most people perceive a constant self-progression within the world of the non-biological. The world of technology to most people seems to move on inevitably with its own momentum. There is an awareness that technology escalates or evolves, but most people are not aware how this happens. This out of control state suggests to many that the technology is self-governing.
We use such studies as cybernetics to understand the control mechanisms and electronic and biological systems. The word cybernetics is derived from the Greek for "steersman". The global telecommunication network is the largest man made structure in the history of the human race. It is so vast, and so quick to change and adapt, that it seems we are fast losing a concise comprehension of its capabilities. Another question is what is it and could it become? The web is an evolving distributed network. It evolves in the sense that it has no central governance and is constructed on a framework of individual components. Each individual component inputs independently and each individual demands output independently. This creates a state of constant fluctuation.
The system I propose does not map the network as a whole, as this is fundamentally flawed due to its constant shape shifting nature. So with such an obvious emphasis on adaptation, evolution and growth it would be a meaningless task to create a system which didn't comprise of all of the above aspects. The visual metaphor of a tree allows great scope for all of these properties. The tree formation is grown as you navigate and traverse the web. This sense of growth aids our perception of movement and space, it allows the user to feel as if he/she is making a physical progression and moving through real space.
This is the first and most vital notion of growth but the tree will also have the capability to grow without the need for constant interaction. The tree system will be programmed with a small amount of artificial intelligence in that it will learn and understand at a basic level the user's navigation behaviour. We all navigate in our own personal manner, although this may seem obvious, this is a meagre area of study. If a company can understand individual navigation behaviours it is possible to create friendlier user interfaces and structures. It also makes it easier for a company to target brand markets. The program will also understand what subject matter it contains within its body structure. With this information and the information the user could set into growth parameters, such as key words, subject matter, growth speed, connections per site, the tree will be able to grow of its own accord. It would be able to find new sites, paths and connections. This will allow a continuous progression and a continuous multi-directional search for information. Each branch will continue to trace extending paths through the network. If after scanning through the new growth you find paths and sites you don't like or are not interested in, you can just prune your tree; chop it back. This form of data mining can be extended by being able to take an off-cut of an existing tree, maybe a small branch of liked sites or subjects and replanting it. Being able to breed and cross breed subjects to create whole new trees for research in new areas.
Another possibility is seeds. At the start of a new project you could simply state the nature of its growth, setting properties similar to those of a developing tree but with a more specific emphasis on subject matter, and let it grow. The romantic idea that you could leave it overnight to grow would not be necessary, within a minute you could have hours of research time contained within the one tree. The emergent nature of this tree system allows for a real sense of growth. A URL creates a node so each node stands as an individual. No node is a tree and the tree formation will not emerge without nodes, neither will the function of a tree change from the function of a node. The tree becomes more than an individual node and more than just a collection of nodes. It forms into something else and this is where my system stands apart from other history programs. It allows the function of its individuals to create the greater function of the whole.
"One drop of water is not enough for a whirlpool to appear in, just as one pinch of sand is not enough to hatch an avalanche. Emergence requires a population of entities, a multitude, a collective, a mob, more."
Kevin Kelly, Out of control
(London, 1994) pp. 26,
Data visualization & visual metaphors
Categories such as time, space, cause and number represent the most general relations which exist between things; surpassing all our ideas in extension, they dominate all the details of our intellectual life. If humankind did not agree upon these essential ideas at every moment, if they did not have the same conception of time, space, cause, and number, all contact between their minds would be impossible....
Émile Durkheim, Les formes élémentaires de la vie religieuse (Paris, 1912), pp. 22-23.
The processing of quantitative data such as URL's, dates, times, modification dates and route directories is how we understand the internet. But without any form of visualisation such raw data bears little or no relevance to our perception of the internet as a virtual environment. To visualise such abstract data is to transform our understanding of it from data, which would take a specialist understanding to a representation, which reflects our most basic visual vocabulary. We all have a visual vocabulary which is described in the above quote. We have to understand elevated maps, timetables, which use a complex parallel logic, just to catch a bus. Now these two examples seem to us quite basic visual communication systems. But to understand your position from such an abstracted view, compared with how we actually interact with our environment at ground level, is complex. The information that the most basic timetable translates is huge - it places you within a new time frame or a relative time frame to the information you wish to abstract. It conveys a concise representation of time and can dwarf a week of times and routines into a dozen lines of text and numbers. We apply this understanding to a day by day need to be able to translate symbols, numbers and signs, to be able to navigate and interact with environments of all types.
Much of this works on recollection and memory. So how do we translate specific data to specific data which we can utilise and understand? Much of this cartographic communication lays in symbolism. One way is to use one of many visual metaphors, such as a tree or a web, which directly relates to a form or structure of which we all have a basic understanding.
These types of metaphor are more than just symbols. They don't just relay one specific piece of data, but communicate an intricate mass of data. The tree is the analogy I have adopted for my system. The tree is one of the most efficient methods of representing and communicating large amounts of complex data. The oldest application of this metaphor would be the family tree. The family tree sits perfectly with the nature of this metaphor. The tree conveys time, movement, and most of all it displays decisions and interaction. It displays multiple decision-making allowing for a multitude of different paths and directions. Dividing and subdividing, progressing its limits, expanding with time, the nature of the tree is growth or progression and this is why it portrays a progressive data source so well. It is so simple that the time frame of the data is translated into the time frame of a tree's growth. It becomes evident at first glance how each node of information sits in relation to the entire structure. If a train of thought changes, diverts or back tracks when it continues - a junction is made; a visual and maybe a physical division. The organic nature of the tree allows for development, patterns and cycles; information, fluctuation and change. It doesn't imply a linear information source or dwarfs its non-linear aspect, but allows for a freedom of the distribution of information. It also gives itself a broader range of symbolic aspects, such as leaves and fruit. I don't feel you would have to look very hard to fit aspects such as seasons, roots and photosynthesis for the system to translate more clearly into the
web's behaviour.
Cartography
Maps are a medium of communication. The cartographer communicates spatial information to the reader/user, such as geographical concepts, locations, distributions and relationships. To do this he must encode the facts and concepts he wishes to express. This happens in just the same way as when you wish to convey information by writing, by the medium of mathematics or by the distinctive combinations of sounds in the spoken word. The cartographer uses the graphic tools of marks, hues, lines, patterns and tones as his visual language. These are his visual tools but to communicate information is the application of the media.
To have a true scale map of a town you would need a map for the map and so on, so maps are scaled reductions. So the real world needs to be symbolised to appear with clarity and legibility. A map is a system of encoding a variety of different types of information and data, so that a trained reader can retrieve the original concept or information. They record, calculate, display, analyse and generalise the interrelation of subjects and their spatial relationship. Maps have allowed man to extend his normal range of vision and make it possible for him to see a broader spatial relationship with his environment.
Maps have made us spatially aware and globally aware. We all have a global understanding about how the world is pieced together. Maps of the world divide its space not only by seas, but also in terms of territory, control and politics. States are used as building blocks to piece the great puzzle together. At a global level projections are distorted, a two-dimensional representation of a three-dimensional environment, the curved globe is not flat. So to visualise this within a rectangle is fundamentally flawed. This represents the nature of modern printing over the need for a universally correct representation of the globe. The flatness of the image, although it fits neatly on a double page spread of an atlas, does not give a correct understanding of the world's circularity.
Media has already displaced global positioning you are more likely to be aware of the news half way across the globe than the news in the next town. This is one of the issues I feel the web will impact on in the extreme. Global displacement is evident on the web since information is distributed over a global network. Every connection is connected as much as any other and global positioning gives no greater access privileges. It creates a mutual space; a virtual infinity where the perfect site is only a click away or this is how it's perceived. The internet is a distributed network and has no apparent organising structure, no determining points of origin or notion of centrality. But stabilising points do dominate such as the US dollar is the choice for on-line trade and English dominates as the universal language. There will have to be a compromise of national identity and communicational dialect. There is:
... a power take-over by a dominant language within a political multiplicity. Language stabilises around, a bishopric, a capital. It forms a bulb. It evolves by subterranean stems and flows, along river valleys or train tracks: it spreads like a patch of oil.
Gilles Deleuze and Félix Guattari, A Thousand Plateaus, "Introduction: Rhizome", trans. Brian Massumi.
Maps represent real environments and actual space. And this is why the use of conventional methods and systems to map the internet will ultimately be flawed. The web as an environment is in perpetual fluctuation, shifting developing and evolving. All is independent, but creates the whole. A map of global communication lines and physical networks is and has been done many times. Mapping the emergent nature of the internet, a virtual environment and a virtual space, is a task many have tried to take on. All attempts fail to present the dynamically changing ecology of users and documents on the web. Researchers at Xerox created a program called 'Time Tubes and Disk Trees'. Disk Trees represent time slices of the web ecology and collections of Disk Trees form Time Tubes. These visualise small slices in time which can be built to create longer periods of time. This can be used to govern and order data, but still this and many other programs are dwarfed by the sheer amount of data which is dismissed by not creating a global visualisation. We have become reliant on maps as tools to understanding space and likewise to navigate it. The program I propose won't map the whole of cyber space, but will record the user's interaction with it and then allow a full understanding of navigation and re-navigation through these charted waters. This slower methodical approach to the problem will I feel, allow a much greater understanding of navigation and how to navigate efficiently with a more defined skill.
Maps, traces, paths & tracks
It is evident that the tree is not a mapping program, although it does create a map of a sort, but more a tracing program recording movement and decisions forming paths, tacks and alleys. Passages of thoughts and ideas. Two researchers at Xerox's Palo Alto Research Centre in California (Peter Pirolli and Stuart Card) are using foraging theories from ecology and anthropology to understand how people search and find information in data rich environments, such as the internet. Biologists came up with a foraging theory in the 1970's as a way of explaining animal behaviour; it could explain the decisions a fox would make between the choice of catching a juicy rabbit or a tiny vole. It decided that animals choose to maximise their "benefit per unit cost". They'll gather food in a way which yields the best energy returns. A rabbit may have high-energy units, but costs a lot to catch; the vole has fewer units, but will also cost less.
So, if food equals information we have to make the same decisions. Do we stay and overgraze exploiting one patch which will eventually lower returns and render the patch worthless or move on to search for a new patch or site which may or may not exist. This points to ideas of designing a site like a flower to attract the 'informavores'. Like a bee may pick up a scent, can a designer create a site that will instinctively draw you in and can we learn to track information, hunt it like our ancestors once sought the woolly mammoth. I think we already do these things instinctively, we spend more time foraging the web than we do for food.
An animal will follow tracks and scents, he understands his environment, and understands his interaction and the interaction of his prey and uses small disruptions as clues. Skills and techniques are used and it becomes apparent that a water hole will attract many information sources, so in the same way as a hunter learns methods in hunting behaviour we can also learn navigation behaviour. However the problem we face is an adaptable environment shifting and changing on every hunt. The tracing system I propose will allow the hunter to follow tracks and maybe learn to anticipate information. It won't just allow re-navigation of old territory, but a greater understanding of all navigation allowing a more efficient hunt with a greater "benefit per unit cost".
"Animals have been solving problems for millennia, and natural selection has made them good at it. It follows that we can learn something from them."
David Stephens, Foraging Theory
(Minnesota, 1986)
Conclusion
It is hard to speculate with any concise reliability how the internet will affect, divert or change our most basic understanding of communication and information. How we interact with these two fundamental traits is going to make vast changes. Maybe it will demean the information-powerful as the availability and vastness of information expands. Or maybe we will become information-displaced, needing to learn something only when it is specifically needed. I think the role of the independent programmer will need to be called into action to take the huge monopoly on navigation abilities away from the Microsofts of the market. We need to learn navigation skills from more than one interface or single angle. Not to do this will always limit the input, output capabilities of the web. If we are to learn to utilise the internet to its utmost capability, we need to be less passive, and to question. We are receptors to these companies and abide by their laws and limitations. As our practice as designers is to conceptually and visually communicate notions of every kind, it seems a futile task to merely constrain and limit our ideas to the gray windows.
Although the system I propose doesn't bring down the walls and limitations of navigation, I feel it does, however, broaden the picture and give a wider scope for a fuller multi directional interaction. I feel the tree will personalise interaction allowing a greater sense of familiarity, thus lessening the autonomous procedures of web browsing. The real application of this kind of program will exceed just re-navigation. It could be used to build navigation behaviour studies, information databases, and a greater understanding of the web as an environment, on a greater level than just where you are sitting within it. This non-perception of where you are going, have come from or want to find leaves you open to all persuasion. When navigating in this manner it is easy to become waylaid, lost and stumble around. This can be fun, but there is a real need for the web to step up a gear.
To be able to have complete access to the information you want will demand a greater flow of credible output which will in turn demand a higher value of imput. This will affect the contextual content, and likewise the visual content. It will allow the designer to have control over more than just the type face, marks, hues, lines, patterns and tones, it will give scope for advanced structural design and allow for the user to have a real interaction. Not just the option of forward and back but maybe both at the same time. If a greater multiplicity in the nature of navigation and web behaviour grows the user will reap the rewards and the standards will rise. It will make way for some real application for a diverse range of ideas and concepts.
At the moment computer science spear heads the conceptual boundaries of the webs capabilities and not the designer. We adapt our boundaries to the progress of the latest programs and have become the masters of them but it feels that not enough of us really step off the beaten track. And this is why I feel giving the user more freedom will in turn create a real need for more than just capable design, but require new visual and conceptual systems which adapt, evolve, and change like the environment it occupies and the users it houses. The tree alone may not bring all of these bright futures, but I do feel that it is an aid to other tools and an individual component to a larger pool of tools will bring about a vastly needed change.
Bibliography
Only four of these are actually named in the essay. The rest I read rather than cited, which makes this less a citation list than a record of what I had on the desk. Every marked title opens a card here on the page: the book or the paper itself, what it argued, why it is on this list, and where to read it. Nine are free in full.
Kevin Kelly. . Fourth Estate,
London. 1995 ISBN 1-85702-308-0
John Armitage. . Sage,
London. 2000 ISBN 0-7619-5901-7
D’Arcy Wentworth Thompson. . Cambridge University Press,
Cambridge. 1981 ISBN 0-521-09390-2
G.R. Crone. . Dawson, Archon Books. Great Britain. 1978 ISBN 0-208-01724-0
Arthur H. Robinson, . John Wiley & Sons, Inc,
United States of America. 1969 ISBN 471-72805-5
Jeremy Black. . Reaktion Books,
London. 1997 ISBN 1-86189-012-5
Edward R. Tufte. . Graphics Press, Connecticut. 1997
Leo Bagrow, R. A Skelton. . C.A Watts & Co.LTD,
London. 1966
Stephen Hawking. . Bantam Books,
Great Britain. 1995 ISBN 0-553-17521-1
Michael Talbot. . Harper Collins Publishers,
London. 1991 ISBN 0-586-09171-8
J. Lamping, R. Rao. . Journal of Visual Languages and Computing, London. 1996
E. Rennison. , MIT Media Lab. Cambrige. 1994
Tim Bray. , Fifth International World Wide Web Conference ‘96. Elsevier Science, UK. 1996
Keith Andrews. , Austria. 1995
R. J.Hendley, N. S.Drew, A. M.Wood, and R. Beale. . University of Birmingham, UK. 1995
Stuart K. Card, George G. Robertson, and William York. . The Association for Computing Machinery, Inc.
California. 1996
J. S.Risch, D. B.Rex, S. T. Dowson, T. B.Walters, R. A. May, and B. D.Moon. . The International Conference on Information Visualisation ‘97.
Washington. 1997
Kim M. Fairchild, Steven E. Poltrock, George W. Furnas. , Cognitive Science and It’s Applications for Human-Computer Interaction. Lawrence Erlbaum Associates, Inc. 1988
S. G.Eick, G. J.Willis. . Information Visualisation ‘93.
USA. 1993
R. A.Becker, S. G.Eick, A. R.Wilks. , Transactions on Visualisation and Computer Graphics. USA. 1995
S. K. Card, P. Pirolli, J. D. Mackinlay. : Display Evaluation for Direct Walk Information Visualizations. The Association for Computing Machinery, Inc. California. 1994
A. Cockburn, S. Greenberg, B. McKenzie, M. Jasonsmith, and S. Kaasten. . University of Canterbury, University of Calgary. New Zealand, Canada. 1999
L. Tauscher, S. Greenberg. . University of Calgary, Canada. 2000
Web references
Almost all of these are gone. Every marked address opens a card saying what stood there; most can still be visited in the Internet Archive as they were in 2001. One was never captured at all.
If these reference sites are clicked on the relevant URL will be loaded automatically.
Twenty-three books and papers, three of them named in the essay. Each one opens a card.
Treebeard is the Ent from The Lord of the Rings. He is the oldest thing in the forest, he remembers everything that has happened in it, and he will not be hurried. I was a student who read a lot of fantasy, and I thought it was a decent pun on a tree that keeps a record. Tolkien spells it as one word. I used two, and I have no idea why.
Twenty-five years later, I still want a browser that does this.