By taking a number of texts and linking them together in a form reminiscent of footnoting, a 'Hypertext' is formed. An example: a student puts Eliot's "Wasteland" on a computer, together with all its related texts. The student sets up the system whereby the reader of the computerized text can call up the related texts while reading the main poem. So if the phrase "April is the cruelest month" is related to some essay on Chaucer, then the reader can call up this essay by clicking the mouse on the phrase. The words/phrases which have links to related texts are called 'Anchors', and are often indicated by being underlined.
A determined student could then go on and enter into the computer all the texts related to the related texts; e.g. the academic papers referenced in the Chaucer essay. The student would then set up anchors going from the Wasteland related texts, to the Wasteland-related related texts. In fact, this process could go on ad infinitum, until the full Universal Hypertext is produced: all possible texts linked to all possible texts appropriately. This is the 'Docuverse'. Actually contemporary theory tends towards the idea that no text has meaning on its own, but only in relation to its endless tree of related texts.
As well as its instant referentiability, Hypertext also includes multimedia. So Eliots 'Wasteland' could also be linked to a digitization of Eliot's radio performance of 'Hollow Men'; and the text of 'Hollow Men' could be linked to digitized video clips from 'Apocalypse Now'.
A final fundamental property of Hypertext is its adaptability. A reader of Hypertext is also an author. He can write and link in new essays to the piece he is reading. Thus Hypertext overcomes the domination of the author, as well as the domination of the central document.
A subset of the Universal Hypertext, is the Universal Hyperpoem. When is Hypertext a Hyperpoem? This question is up to the individual writer, but could probably be answered in analogous ways to the question "when is prose poetry?" To keep it simple: when a bunch of poems are linked together (or when a bunch of texts are linked together 'creatively'), a Hyperpoem (HP) is formed. The individual poems that make up the Hyperpoem are called 'Subpoems'.
Outside of subpoetica, questions of Form (i.e. Hyperform) are central to Hyperpoetry. Hence one approach to an analysis of Hyperpoetics is a taxonomy of Hyperforms.
HPs can be roughly divided into two approximate classes: FOOTNOTE HPs and NSEW HPs.
The Footnote HP gets it name from the way that Hypertext has been used in the past. A Hyperpoem can have a number of subpoems that run in parallel, i.e. there is no set order of reading. In a sense, the subpoems can be seen as footnotes of a main poem, or of each other.
The simplest footnote HP involves a HOME poem, which has a number of anchors to (usually) shorter subpoems. These subpoems may themselves contain anchors - but in the end everything depends upon returning to the home poem. This type of footnote HP is called a CENTRE HP.
At the other end of the footnote spectrum lies the Hyperpoem without a home poem; where every part of the HP is a subpoem with links to other subpoems. This is called a WEB HP. (A very complex crossover of the Centre and Web HP is a Centre poem whose home poem is a hyperpoetic web.)
The NSEW (North South East West) HP has more to do with linearity and choice. When writing a poem, a poet makes many decisions (sometimes subconscious) concerning the direction which a poem may take. By stopping at some of these decision points, the poet can offer the reader a number of directions in which to follow the poem. The poet can write two possible subpoems to follow the decision point, and let the reader chose which one comes next. The poet can go on to write two possible subpoems for each of these two possible subpoems, and so on, thus creating a branching out of possible paths through the Hyperpoem. Although the reader choses the direction through the Hyperpoem (hence the name 'NSEW'), the poet still enforces some degree of linearity.
An OPEN NSEW HP only allows upward movement through the tree of possible directions. A NON-OPEN HP allows some movement accross the tree. A CLOSED NSEW HP has at least one route back to the bottom of the tree, back to what is called its ROOT poem (c.f. the home poem of a Centre HP). The ORDER of a NSEW poem is the maximum number of links in its subpoems. So a NSEW HP which literally had the direction options 'North, South, East or West?' as links in it, would be a fourth order NSEW HP.
Obviously these two types of HPs, Footnote and NSEW can blur to become CROSSOVER HPs. A poet can design varying degrees of linearity into a footnote HP, and into a NSEW HP.
The vast majority of Hypertext is on the World Wide Web (WWW), a multi-media Internet. The texts on the WWW are stored on various connected computers around the world called SERVERS. Links between texts are shown in the form of underlined words.
A NON-DISTRIBUTED HP has one author, and is contained on a single server. Because of a poet's changing views on a poem, and the ease with which editing may be done on a computer, the HP may go through a number of revisions. But these revisions will be essentially linear in nature, being done by one poet.
A DISTRIBUTED HP is contained accross a number of servers around the world, and may have more than one author, thus making revisions non-linear. Even if it has only one author, the fact that this author only has physical access to one of the servers will lead to a more frequent editing of the subpoems on that server: hence the revisions may still be non-linear. Also, some of the servers may go down, slow down, etc.
A MALLEABLE HP, which includes the Hypertext facility for allowing the reader to add to it, is particularly sensitive in this area. Distributive malleable HPs may grow and mutate accross the servers of the world, fed by various authors.
In reality, good Hyperpoems are much harder than normal poems. By a 'good' Hyperpoem, I mean a Hyperpoem that doesn't lean too heavily on its Hyperform. This may seem contradictory (why write Hyperpoetry unless you want to utilise Hyperform?) but there is a temptation to take any old poem, chop a few bits out to give a home poem, then put in some semi-relevent anchors to the chopped-out subpoems. This is a pointless (but common) approach, and will be seen for the worthless thing it is once the novelty factor of the WWW and Hyperpoetics dies down.
This chopping technique is an example of a poem parasitically feeding off its Hyper aspect to give it credibility. What is needed is a Symbiosis between a poem's Hyper and linear aspects for there to be anything of value in the HP. It is finding this symbiosis that makes Hyperpoetry so hard.
As a case study in hyperpoetics, we will show one method that can be used to construct a simple, but symbiotic, Centre HP.
Write the home poem first - make it work on its own. But bear in mind, the better a home poem works, the harder it is to make it Hyper without disappointment. Next, chose your anchors. Remember the psychological effect that anchors have on the Internet user (assuming you are producing the HP for the WWW). The anchors are not seen as simply underlines, but as saying 'this word/phrase is Significant, it leads to something New!" A badly positioned anchor can have a nullifying effect on the text around it. It is often a good idea to anchor to strong words and phrases. This will seem more natural to the Web user.
Next write a series of poems with titles the same as, or related to, each of the anchored words/phrases. Write (create) them in the order that the anchors appear in the home poem. Try not to pause for long between each writing. When writing these subpoems bear in mind the home poem, but beware of totally dropping into its style (or else your subpoems will just feel like footnotes).
Write a subpoem, bearing in mind while you're writing it that it is a subpoem and not a home poem. Chose a strong anchor word/phrase from it.
Next write another subpoem, and try and steer its path in such a way that it eventually contains the anchor word/phrase chosen from first subpoem. Link the two poems together using these words/phrases.
Now chose a new anchor from the second subpoem, and repeat the process.
Write a single stanza root poem. Choose two strong words/phrases in the stanza: make them anchors. Write two second-stanzas, one linked to each anchor, each concentrating on, or sparking off from, apsects of the word/phrase it is linked to. Now repeat this whole process for each of the new stanzas. Keep going: never stop.
There may seem to be no point in writing 1st order NSEW poems. After all what's the point of writing a poem with choices, when there's only one choice in each subpoem? A DECAY HP is a symbiotic 1st order HP. The anchor is one of the words in the subpoem. Selecting the word takes the reader to a copy of that subpoem, but with the selected word missing, and a new word from the subpoem underlined. Selelecting this new word takes the reader to a subpoem where both of selected words are now gone, but which resembles the original root poem in every other way - with another word underlined. This process continues until no words are left. Or until a meaningful conclusion is reached in the remaining words. The Hyperpoet's job is to make this process semantically coherent, or at least poetically relevent.
Hyperpoetics is intimately related to Connectionism, especially as its most common breeding ground is the WWW (which can be viewed as a connectionist database). A major figure in the history of Connectionism was Donald Hebb, a neuropsychologist. He put forward a hypothesis to explain Long Term Memory (LTM) in the brain. The brain consists of cells called NEURONS which communicate with each other by firing out electronic impulses. It is from this interaction of neurons that much of our thought comes. Hebb suggested that if two connected neurons fired simultaneously, then maybe the connection between them strengthened (or if one fired when the other didn't, then the connection between them weakened). This is the HEBB LEARNING RULE, and has been used as the basis for many Neuroscentific theories and for the development of computational intelligence using artificial neural networks and evolutionary algorithms. Apart from its relevence to memory, it is found that when the Hebb rule is applied to a web of many interconnected neurons, order spontaneously emerges from the chaos.
By viewing each subpoem as a neuron, and the links between the subpoems as connections between neurons (i.e. those connections which are strengthened or weakened by the Hebb Learning Rule), one can develop Hyperpoems whose structure is affected by the way they are read.
Suppose a home poem has links to three subpoems, and the readers chooses the first subpoem and reads it. As a result the connection between the home poem and the sub poem is 'strengthened' in some way. For example, perhaps the underline for the anchor to that subpoem becomes thicker/bolder. (And maybe the underlines for the anchors leading to the other subpoems are made fainter/weakened.) If this process of strengthening/weakening is repeated as various readers explore the Hyperpoem, then the Hyperform will emerge which is chosen by the readers. A form of 'natural selection' is occurring amongst routes through the Hyperpoem. The HP is SELF-ORGANISING.
There are a wide variety of artifical neural network architectures. When looking for new hebbian hyperforms to experiment with, the Hyperpoet could do worse than to look for inspiration in these networks. Some examples are Hopfield, Back-propagation, Neo-cognitron, Kohonen and Adaptive Resonance. The Back-propagation net extends nicely into Hyperpoetics, leading to Input, Output, and Hidden subpoems. Given a multiple centres HP, the Input subpoems are the home poems, the Output subpoems are those on the periphery of the hyperform; and the Hidden subpoems are those in-between. In this case, the altering of anchor strengths does not happen until the reader reaches an output subpoem. Then anchor strength changes are back-propagated to the home poems.
Artificial Neural Networks are often seen as just being one branch of COMPUTATIONAL INTELLIGENCE. Another major branch of Computational Intelligence are EVOLUTIONARY STRATEGIES (e.g. Genetic Algorithms, Genetic Programming, Ant Colony Searches). By creating multiple copies of a Hebbian Hyperpoem and making them accessible to a variety of people, a POPULATION of Hyperpoems emerges. The members of this population will evolve as various groups of people read them. Ideally this Hyperpopulation would be distributed.
Perhaps, given access to the WWW, other aspects of evolutionary strategy could be applied to Hyperpoetics. However, one must bear in mind that current versions of HTML (the computer language used to write Hypertext in the WWW) do not have the ability to weaken or strengthen the underlining of anchors. Perhaps other methods such as blinking, boldified, different sized and different fonted texts will have to be used in the mean time. (Interestingly enough, the latest version of Netscape on the PC actually highlights an anchor on a page you're reading if you have used that anchor in the past.)
A well documented aspect of Hypertext is its dependence upoen the observer/reader. Usually (as in 'The Wasteland') the reader is supplied with a central text which takes dominance over the subtexts it refers to. In Hypertext however, the observer has instant access to all the subtexts, and may become more interested in one of them than in the main text. Thus the dominance of the central text is destroyed.
Similarly, the observer of a hyperpoem decides their own route. The observer affects the characteristics of the observed hyperpoem (especially if it is a hebbian HP). This is similar to the situation in quantum theories. Another side of the same coin in quantum theories, is the Uncertainty Principle. This can be applied to Hyperpoetics by creating SEMI-STOCHASTIC HPs. Here the same link will not take the observer to the same subpoem each time they select that link. The uncertainty involved, and the precise number of possible subpoems for each link, is up to the poet. One can even apply the theory of different PROBABILITY DISTRIBUTIONS. Given a semi-stochastic HP, the probability with which a possible subpoem is selected from a particular link can be governed by a standard probability distribution such as Possion or Gaussian. Thus one can have Poisson and Gaussian HPs. (Once again, this semi-stochasticity is not directly implementible in HTML: perhaps somebody needs to design HPML, a HyperPoetry Markup Language.)
A semi-stochastic hebbian-web HP, can truly call itself a 'Quantum' Hyperpoem.
The major stumbling block for Hyperpoetics, and indeed for Hypertext in general, is the Cathode Ray Tube. The only practical realisation of Hyperpoetics currently is on a computer screen. Computer screens are incompatible with human beings in two simple ways:
(1) Unlike books, they are bulky and difficult to transport (and the smaller ones have bad resolution qualities).
(2) They are hypnotic, and stop the blinking reflex in the eye - leading to a lack of luid on its surface - and then to eyestrain.
Item (1) is becoming less and less important as time goes on. There is every reason to expect that a decent cheap lightweight screen will become available. Perhaps even a screen that you can fold up and put in your pocket, like a book. Item (2) however requires a total rethinking of the way that we put our text onto a screen. There seems to be no sign of a move away from Cathode Ray Tubes at the moment: Liquid Crystal Displays are not a feasible alternative.
Thus the study of Hyperpoetics is currently also the study of Cathoetics and the psychology of perception. (Any future serious studies of Hypertext will need to look at this aspect as well.)
Attention span is cut by a factor of 3 on the WWW. Try it: so much to see and so little time to see it. Combine this short attention span with (1) and (2) above, and you have a recipe for your HP becoming 'The Tree That Never really Fell Because No One Saw It Fall'. Actually, more often than not, it will become 'The Tree That Everyone saw Beginning to Fall, But Got Bored and Left Before They Saw It Hit The Ground'.
At the beginning of Italo Calvino's 'If on a winter's night a traveller', the author describes various ways of reading a story. My favourite is to lie on a bed, listening to music, and reading the story from a book. Calvino fails to include the modern option of sitting in a bright white computer room, back straight, staring at an opaque screen with minute electron glows dancing across the glass - and with eyes wide unblinking: no tactile sensation of paper, just the irritatingly almost-friction-free glide of a mouse over the table - the only sound being the groaning of computer cooling-fans. Generally, an experience of Hyperpoetry will include more than one of those five unattractive aspects of digital technology. This is a far cry from 'lying on a bed, listening to music, and reading the story from a book'.
Lets take a quick dip into that most famous of literary ghettoes, Science Fiction.
A Hyperpoetical medium which could compete with books in the realm of recreational reading, would perhaps be a nano-technological Radio-Internet Transceiver. The pages would be colonies of molecules that could diminish and multiply, growing into black marks and pictures very quickly when nessecary. It would connect into the Internet via radiowaves, and subpoems would be selected by toughing the page. The book would have a spine, physical pages, and be a manifestor of the Universal Hyperpoem.
The nano-book is pure science fiction. However, it makes a serious point. Until hyperpoetry is less effort to read, much of its theory will be fiction. To overcome these limitations in the mean time, the Hyperpoet must not only be aware of the necessity for symbiosis, but also of Cathoetics: the potentially 'blinding' visage of the E-muse.