First hypertext fictions were written in the early eighties, and the first commercially distributed hypertext fiction was Michael Joyce's Afternoon. A Story. It was published by Eastgate Systems in 1987 one and a half decade ago. I would like to take a look at hypertext fiction, its history and present, and try to make some predictions of its future. Also, other topics like electronic reading devices ebooks are discussed.
In the mid-eighties there were already an amount of different kinds of computer based cybertexts alongside hypertext fiction (especially adventure games and Multi User Domains; the latter being a multifarious upgrade to the first multi user adventure games). Hypertext fiction (hyperfiction) uses hypertext authoring environments to create "interactive" and "non-linear" fictive texts, which were said to offer the reader an unheard of power over the unfolding of the story. The fact that hypertext, a medium mostly targeted to educational and technical writers, also thrilled fiction authors is not a surprise - what is surprising is that it took so long for fiction authors to get interested in a tool which, according to its founder, Ted Nelson, was to transform computers into "literary machines".
According to Ted Nelson, hypertext is "non-sequential writing -- text that branches and allows choices to the reader, best read at an interactive screen. As popularly conceived, this is a series of text chunks connected by links which offer the reader different pathways." From the definition by Nelson, the first type of interaction is clear: the reader can choose between the different paths through the text corpus. When the interaction is limited to this only, we can, after Michael Joyce, speak of "exploratory hypertexts". But there is more to it: some hypertexts allow the reader to rearrenge the links, make new links, and even to write his/her own lexias. This type of hypertext we call "constructive hypertexts". 
A lot of expectations has been put upon hypertext fiction, with its capability to empower readers' interaction with the text, because of the new kind of spatial writing it makes possible, and because of the possibility to integrate text with audio-visual materials, thus creating a new kind of "Gesamtdatenwerk".
Although the significance of hypertext for present society cannot be overestimated (World Wide Web, for example, being based on hypertext techniques), the point of hyperfiction still remains unproven. At the same time as MUDs have been expanding in number, styles, and users, and while adventure games have been developed with the aid of the latest computer hard and software towards interactive cinema and virtual reality, the number of hyperfictions has stayed exceptionally small, their publicity is almost none, and the tools haven't been keeping up with, for example, WWW page composers. The main problem, however, lies in the generic reader interfaces the hyperfictions employ. The interaction happens mainly through anchor words/sentences, and menu windows
We are, in fact, in a situation where we really have to consider the question if hyperfiction is, in the ten years from Joyce's Afternoon to Twilight (1997), come to the end of its road. This would not mean the total extinction of the genre, but rather the shift of focus to different areas, especially towards virtual reality worlds (or, narrative story worlds). With the examples of a couple of more original new (and forthcoming) hyperfictions, I try to sketch an alternative, a kind of hyperfiction which surely tends toward virtual reality, but without losing its characteristic nature as based on written language. I would like to take as main my starting point J. David Bolter's article "Ekhprasis, virtual reality, and the future of writing" and polemically confront his opposition between text and image, and the strenghtening dominance of visual over textual:"Digital graphics call into question the future of alphabetic writing itself".
In Joyce's Afternoon (in the original MacIntosh version) the reader interface, or, the navigation device, was simple and clearly separated from the fictive content. There was no map for the reader depicting the spatial structure of the text - even though the story was built according to a strict hierarchic structure as shown in many essays written about Afternoon.
Stuart Moulthrop's hyperfiction Victory Garden went one step further. In Victory Garden there is a map of the hypertext - the map isn't too detailed and its worth as a navigation device is almost none, but still, it made explicit the spatialized structure behind the text (which, not too surprisingly, can be described as "a garden of forking paths", or better yet, "a garden of intersecting paths").
With Shelley Jackson's Patchwork Girl, we can go still further towards the spatialization of hyperfiction (even though Patchwork Girl's status as "fiction" is not self-evident). In Patchwork Girl the spatial and conceptual arrangement of the hypertext lexias is made accessible for the reader. The navigation device is now more complex than it was with the previous hyperfictions mentioned here (the situation with Windows versions of these texts is a bit different, since they all employ the same technique) - but here lies unfortunately also one of the main weaknesses of this interface: the toolbar mixes spatial and conceptual functions in a very confusing way. Jackson also employs the spatial map as a site of signification - that is, in some passages the arrangement of the lexias and their colors work as a meaningful composite. In Patchwork Girl, which can be best described in terms of postmodernist poetics, the metastructure - that is, the conceptual map - is an active part in the reading process. Because of this, the cognitive leap from text to navigation and back is not that big. The same technique is used by Deena Larsen in her Samplers - nine vicious little hypertexts. This text is based upon a quilt with nine different images: each of the images lead to their own story, which spatial arrangement more or less resembles the according image.
The new work from Joyce, Twilight. A Symphony has all the benefits - as well as drawbacks - of the spatialized reader interface, and also an amount of audio-visual materials but still it does not, in any significant way, exceed the boundaries laid out with Afternoon already. Thus, from Afternoon till Twilight seemed possibly to comprise the long day's journey for one mode of writing.
But then, the possibilities hinted at in Patchwork Girl and Samplers, as well as some qualities borrowed from computer games, have come to enliven the genre in works like Califia by M.D. Coverley, and Reagan Library by Stuart Moulthrop. They are both works which have a strong visual aspect. In Califia the different visual materials help to create a very emotive virtual space in which the story takes place. With the use symbolic signs, like star charts, the virtual space is scripted it needs to be read as well as appreciated.
The solution in Reagan Library is quite different. It is a web text which uses Quick Time VR plug-in. With the plug-in, a 360 degree panorama can be attached to the web pages (see picture). Each page consists of a navigable space and of text (with hyperlinks). Each text fragment is connected to a certain place in the virtual landscape, but the exact relation between place and text is quite vague. In this case the spatialization of the hypertext is much more concrete than with the highly abstract spaces of cognitive hypertext mappings. One of the results is that navigation is much more intuitive this way.
A Screen from Reagan Library by Stuart Moulthrop
Both Califia and Reagan Library are still clearly hypertexts, but they do push the envelope a bit. One of the problematic points in them from the hypertextual point of view is the notion of lexia. What exactly, is one lexia in Califia, or in Reagan Library? Should the text and image be seen as a single unit, one lexia, or are they two seprate, although closely connected, lexias? The additional programming and random procedures in Reagan Library are already quite long steps towards cybertextuality. The tendency seems to be moving further and further away from pure hypertexts.
There is also another, very different and very interesting approach to visualisation of hypertexts. Jim Rosenberg has written extensively about the idea of "hypertext taken into the lowest level of language" and "channeling the syntax outside of a sentence". In his approach, the visual layout is put to provide a syntax for the hypertext structure. This allows both linguistic structures otherwise impossible (like simultaneity), and hypertext taken to its logical conclusion: not only linking text lexias which themselves are not hypertextual, but creating texts that are wholly hypertextual. Rosenberg himself has employed this strategy in his Intergrams series of poems, but so far the idea hasn't been truly adapted to hyperfiction. Rosenberg has also been active in reminding us that the link-node structure is not the only possibility for hypertext; in addition, structures like nesting, relations, and sets can be used.
Time and space are two fundamental categories with which we structure the world around us. When the fictional space has been extensively concretized in hyperfiction, it seems odd that temporal issues have been on a sidetrack in the hypertext discussion so far.
Stuart Moulthrops Hegirascope and Mark Amerikas Grammatron are by far the best known and most satisfying examples of temporally conditioned hypertexts. In both, the lexias follow each other in certain rhythm and the reader just has to adjust her reading speed accordingly. Hegirascope offers also hypertext links, which can be used if the reader is fast enough.
One of the important temporal levels is the time of the text. With constructive hypertexts the text may undergo radical changes through different readings - that is, the text itself "lives", or, evolves. In hypertext editor StorySpace this is possible in two ways: first, there are so called guard fields which put conditions to some links (the link will not be activated before a certain lexia is visited first - for the reader this means that at some point in time a link exist, at another it doesn't). Because of the pre-planned and stable structure of this device its status as a temporal device is weak. In other cases the reader can make changes to the whole structure of the text, which makes it potentially changeable. This certainly introduces a temporal aspect for the text, but once again, in a limited sense: the changing text is approachable only to the specific reader. Deena Larsen, for example, with her Marble Springs, has exceeded this limitation by asking readers to send her their comments and insertions to be included in the upgraded versions of Marble Springs. Fully effective this quality is clearly with WWW-based texts.
Web-fictions are always, at least potentially, subject to changes beyond authorial control.For example, if there are links outside the author generated material, the destination page may change (this is nicely used in Matti Niskanen's web-fiction Leporauha, in which, for example, there is a link to the web front page of a yellow press paper which naturally changes daily). Another web fiction (actually a hybrid of print novel and web fiction), Markku Eskelinen's Interface, grows twice a year by new lexias written by the author or active readers. One especially interesting web fiction is Gonzalo Frasca's web text, the reading of which is limited to only one session at a time. That is, once you stop reading it, you can never return to the exactly same text - the time of logging in, for example, has effects on the text. This is a very interesting application in its way to fuse real time and narrative time.
In texts like the ones mentioned above, and in all texts subject to considerable change, the text time is the dominating one in relation to textual (fictive) time: at different phases of text time the cognitive time categories may vary even radically. The distinction between textual time and text time may be one of the crucial differences between spatial hypertext and cybertext, since the latter often aims at conflating these into one temporal dimension.
Even with this crude outline of different temporalities involved in hyperfiction, it should be clear that there are huge potentialities not yet realized in hyperfiction. Better yet - even though the temporality of hyperfiction is undertheorized, the situation is much better in traditional fiction studies. With all the useful formalisations of narratology at their disposal, the hypertext (fiction and otherwise) tool developers should be able to produce navigation devices which are as capable of manipulating the temporal dimensions as the present devices are in handling spatial ones.
When the spatial and temporal aspects are developed in the digital texts, the step from hypertext to cybertext is inevitable. While works like Hegirascope are, in Espen Aarseths theoretical framework, definitely already in the category of cybertexts, I see it useful, purely from a practical viewpoint, to stretch the concept of hypertext to include them. In Hegirascope, for example, the hypertextual structure is clearly the dominant aspect it is the basis on which the temporal push effect is built. The link following is temporally restricted, and if no choices are made, the default link activates. This is not so far from ordinary hypertexts, after all. On the other hand, we can quite easily imagine an infinite number of highly functional cybertexts which are lightyears away from pure hypertexts. Thus, cybertext could have more of analytical power, if (almost) the whole of hypertextual field would not be submerged to it.
Cybertextuality should not be confused with multimediality. Even though there seems to be a strong tendency towards multimedialisation in cybertexts, that is in no way a necessary connection. One needs only to look at Aarseth's text typology to realise that there are hundreds of possibilities for functional texts, just using the seven variables of traversing texts. That is, there are hundreds of genre positions available, without using any multimedial features at all. Add to that all the multimedial possibilities and we start to get a glimpse of what we are dealing with.
Genre is clearly a word which should be used with great care here. If we adopt Aarseths terminology of genre positions, then genres are functional categories which can be formally described. This is naturally a very different approach to genres than the traditional use either the classical way of distinguishing between the genres of drama, epic, and lyrical poetry, or the modern categorization of genres like mystery, science fiction, romance, etc. Despite confusing the already diverse use of the word genre, the functional genres should be added to our conceptual tool box. That is the surest way to avoid the unhappy situation prevailing presently, treating all the modes of digital literature as one undifferentiated genre of its own usually called, to increase the confusion, hypertext literature. A relatively simple functional classification is necessary in order to see the obvious differences between hyper and cybertexts and one should be aware, though, that there are always hybrids which are not easily fit to any one genre.
A Screen from Core by Peter Berlich
The old genre notions, naturally, are not useless with hyper and cybertexts. The genre categorizations according to the content type (science fiction, mystery, romance etc.) can be well enough applied here too. There seems to a tendency, though, towards weakening the boundaries between genres. Quite simply, when a text is subject to changes caused by programming, or by the reader, it is always suspect to step over any rigid classifications.
Other boundaries in jeopardy are the thresholds between fact and fiction, between prose and poetry, and, between fiction and drama. Also, the countours of an individual work are getting more and more vague. A good example of hardly classifiable cybertext LaFong's (the writing/directing team of Michael Kaplan and John Sanborn) cyber version of the movie "North By Northwest." In this "work" people are drawn to the story world by an enigmatic e-mail message. Soon other messages follow, and soon the person finds herself being intricately a part of some organization named Dysson (of course, it is always possible just to ignore the messages, but it is made quite difficult). The game (if you can call it that) consists of e-mail messages sent by the organisers, but also by the readers/players, and of web-pages. The readers are drawn to the Dysson organization (which turns out to be some kind of a cover for a cult or conspiracy) and they are put to play an active role in that organisation. The organisers direct the "play" somewhat, but the readers can also contact each other directly. What has started as a game may turn to a real organisation, if the readers do play along their roles. While this may seem as quite a transparent plot, it is not, after all, that easy to make sure if it is all a fake (as witnessed in Eric Idle's account of his own experience). On the other hand, what goes on in the web is "real" in some sense - in cases like this, the division between fact and fiction seems unsuitable; rather, it is a question of simulation. Also, are we dealing with a dramatic work, or with an interactive narrative? And are there any boundaries to a work like this, where everything can be connected (and linked) as parts of it, by the organisers or the readers themselves?
There are no limits to hypertextual linking. The realisation that (almost) anything can be linked to anything else is the main force in undermining any rigid boundaries. The linking can occur between independent documents (like in WWW), internally between parts of a document, or even, externally, from text to the surrounding everyday world (cybertect which makes a call to your cellular phone etc.)
The situation where each and every conventional boundary is subject to transgressions, it is still totally possible to conform to the conventional forms. One should not confuse descriptive and prescriptive statements. To describe the possibilities to overcome certain conventions does not mean that there could not be writing according to the old forms any more. On the other hand, the history of literature has been evolutionary like any other historical phenomena forms change, new forms are born and some forms are outdated. A good example is metric poetry. It clearly got outdated during the first half of 20th century. This doesnt mean that it was totally extinct there are no prohibitions to write metric poetry even today (or even in a digital medium), but it cant be used in an innocent way anymore. Metric poetry written today bears inevitably marks of the present situation; it is an intentional act in itself, which cannot but affect the whole text and its reception and interpretation. Likewise, the birth of hyper and cybertexts doesnt annihilate any old forms of writing, but it most probably makes many old forms outdated in the same way as metric poetry.
Cybertextuality is a challenge for the electronic books. There are two alternatives for ebook development. They may take the remediating way of mainly doing what books have done, but in a digital format. That is, they can be limited to reading digital texts, which use only sparsely cybertextual qualities. The most used e-book at the moment, Rocket Book, is clearly designed for reading simple text it can handle simple hyperlinks but that's all. Clearly, for better and worse, it is still a book.
The other possibility, then, is to abandon the book concept, and take as starting point the notion of cybertextuality. In this direction we are dealing more with reading devices than books there are no needs for buttons to turn the page, for example, but a flexible way to handle programmable texts. Also the interface should be programmable, that is the only way to accommodate all the infinite types of cybertext.
It is quite possible that there will be two basic types of ebooks. The one type is not much different from the models we have at the moment they are simple, book-like (in every way) things meant to be used in reading traditional literature; probably they will be recycling platforms for pulp fiction. The other type is a more elaborate device which can run any kinds of cybertext, and possibly is wirelessly connected to the web continuously. With a device like this, Ted Nelsons vision of instantaneous literature is finally reality.
The older media end up becoming the content of newer ones, said Marshall McLuhan in his Understanding Media. This is certainly the case with digital literature presently. Most of the digital texts are repurposed, or, remediated forms of previous literature. While it seems clear that literature in the traditional sense will continue to be a part of the contents of new digital media, there will also emerge a new type of literature, one which is not only remediation of older literature. Jay David Bolter and Richard Grusin have claimed that in each media there are two parallel tendencies simultaneously at force. The other tendency is towards immediacy, or, the transparency of the media. In realistic literature this is the goal - the text should be as transparent as possible, so as not to disturb the make believe of fictional experience. The other tendency is towards hypermediacy, the stressing of the media itself. Postmodernist metafiction, calling attention frequently to textual-fictional nature of the narration, as well as to the material aspects of a book, is clearly hypermedial.
Hypermediality should not be confused with hypermedia or multimedia as such, as the examples above tried to exemplify. Written language can fulfill both functions perfectly well - but still, there seems to be a tendency to break away from textual presentation - Califia and Reagan Library use visuality for immediacy, Patchwork Girl and Samplers for hypermediacy. As Jay David Bolter claims:"Words no longer seem to carry conviction without the reappearance as pictures of imagery that was latent in the words." (1996, 260) Even if this was a valid description of the overall cultural situation it does not mean that language as such had lost its power to evoke worlds. The visualisation and spatialisation of hyperfiction does not mean its merging into virtual reality - text may maintain its status alongside visual information and in a new symbiotic relation to it. Thus, I think we already have hyperfiction expanding the limits of the genre in spatial terms, and further, I believe we'll see new works expanding the limits also in the field of temporality - for starters, there are Aarseth's 576 genre positions to try out.
 Nelson 1993.
 Ibid., 0/2.
 Joyce 1995.
 Rötzer 1995, 127. About the spatiality of hypertext writing see Bolter 1991.
 See the work of Brenda Laurel and Marie-Laure Ryan especially.
 Bolter 1996, 256.
 See especially Douglas 1992; 1994; Walker 1999.
 I have borrowed the term scripted space from Soeren Pold (1998).
 If, however, someone happened to have problems in imagining them, she should see Aarseths typology of texts (which serves as a heuristic help in theoretically defining 576 different genre postions) or Markku Eskelinens more hypothetical visions of cybertextual possibilities (Aarseth 1997; Eskelinen 1999).
 This is nicely illustrated in Monthy Python comic Eric Idle's account of his involvement with "Dysson"- see: http://www.pythonline.com/fakeopen/dysson/