1Obsolescence and lability. These two keywords define the digital apparatus and lead us to think that every artistic work created with the latter is completely and inherently ephemeral. Such a point of view certainly has a deep impact on the esthetic characteristics of a work, as well as on its chances to be preserved. It would then be impossible for a creator to choose the way of digital creation without acknowledging this fundamental condition. It should yet be noted that many authors carelessly embark on digital literary forms, thinking that they are but a byproduct of books and that, although they offer more possibilities, their cultural characteristics essentially remain the same. By the way, don’t we talk about “digital books” and “augmented books”?

2In digital literature, the issue of the ephemeral is hardly ever discussed and sadly, the slightest modification of the prevailing digital paradigm leads authors to complain about the fact that their work cannot be made sustainable. Such was the case when both MACs and PCs supplanted family computers, then when the appearance of the web marked the end of the cultural CD-ROM, as browsers entered an “incompatibility war” in which every new generation of browsers would invalidate the previous generations’ screen display of HTML files, or more recently as smart phones began to supplant computers. As well, HTML 5 is now gradually ousting Flash, this software tool in which so many authors believed to guarantee the digital durability of their works.

3Maybe we should try to determine the exact temporal behavior of these productions before we draw general conclusions. Let us start with the analysis of the preservation solutions currently proposed: are they likely to stop the ephemeral or cancel it by making the same things happen again?

The preservation of digital works

6The notion of variability cannot be dissociated from that of state. If a variable apparatus must be maintained or re-established, what is the moment, the state to freeze and preserve? In short, what is the reference state of the work? Is it the material state that “makes it” a work? This issue is explicitly tackled by the variable media theory, according which the creator’s intent is the reference state. It invites creators to describe this state in detail.

10The latter approach is now widely adopted in digital literature. There are currently many international databases (ELO directory, ELMCIP, Hermeneia, to mention but a few) trying to get together and form a consortium (CELL) to ensure their interoperability. Documentation seems to have supplanted copying.

11All these approaches are yet based on the same axiom: preserving a work amounts to preserving its materiality at one or several precise moment(s), either by copying this materiality or simulating it, or in other words, by copying some of its aspects. This approach is ill-suited to works that are designed for what I call “private reading,” meaning works that are inherently destined to be read by everyone in various, widely differing, digital contexts and reception conditions. Then, we are to observe that variability is expressed by a phenomenon of lability —which I consider as a constituent of the very materiality of the work— rather than a media that would embody it. Incidentally, some works operate through lability. As we will see later on, it is sometimes part of the author’s project, sometimes not.

12To consider variability as a fundamental characteristic of the work and media leads us to reconsider the role of preservation: is it really a specific materiality that must be preserved or restored? The Spinozist ontological model will provide us with the necessary heuristics to tackle this issue. Let us now introduce the latter through a reflection on the role of time.

Ephemeral or temporary?

13The terms ephemeral and temporary are often regarded as synonyms. They both refer to a temporal system. However, the dictionary of the Académie française gives two distinct definitions for these words. The definition of the word ephemeral is very close to its Greek etymology: “Which lasts only a day. By extension: which lasts a short moment, without a future.” As for the word temporary, it is defined as follows: “Which exists only for a short time.”

14The notion of temporary does not convey this idea of a tomorrow. In other words, ephemeral is the antonym of eternal (which is endless), while temporary would be the antonym of permanent (without interruption). What is apparent here is the dichotomy between eternity and immortality that Spinoza has developed in his ontology. Now, Spinoza proposes a solution to this dichotomy. Let us summarize it briefly.

16This concept leads us to wonder about the power to act and its relationships with the materiality of productions: we know for sure that obsolescence and lability have an effect on the materiality of productions, but what about the productions’ power to act? Is the work defined only by materiality? On the other hand, materiality and perception are intimately linked: what status does perception acquire when the work’s power to act is disconnected from its materiality? Then, what does the preservation of a work mean?

17To answer these questions, we have to examine in greater detail what lability and obsolescence consist in and what they really have an effect on.

Observations

Lability

18We call “lability” a property of the result that is displayed, most of the time on a screen, as the program is run: for a given program, this result depends on the technological context in which the program is run. Therefore, it is a phenomenon that first and foremost affects the digital reading. Lability is above all a technical phenomenon. It was first observed in 1990, through the behavior of the works published in the Alire magazine, a historic magazine in the field of digital literature. We soon realized that the display of the work would vary greatly depending on the machine on which it was run, regardless of generativity or interactivity. This process is inherent to the functioning of computers and a direct consequence of IBM’s invention of compatibility.

19This phenomenon would not, or hardly, occur on family microcomputers, as programs would mostly be related to a single machine and always to a given manufacturer.

20With the invention of the PC, IBM also invented compatibility, or in other words the possibility of running a program on all compatible machines. However, this possibility does not mean that the displayed result will be identical on all machines: this difference constitutes the technical aspect of lability. The capacity of the average IT equipment has kept on increasing, which has resulted in the appearance of a diachronic, automatic and irreversible lability: within a few years, it became impossible to find machines that would run the works’ programs as initially designed by their authors.

22In this sense, technical lability goes together with semiotic lability, the most problematic being the latter. Semiotic lability occurs when the technological context influences the semiotic decision and construction of meaning, as well as the emotional or esthetic feeling.

Properties of lability

23Technical lability is in no way comparable to noise. The program’s behavior is perfectly reproducible in a given technological context, without inducing any jamming or noise phenomenon. It just changes the observable properties of the result. Therefore, readers will generally detect no sign of lability and will become aware of its existence only by comparing the various displays obtained through the execution of the program on different machines, which they usually cannot do. A fortiori, they cannot compare what they observe on their screens with what authors can see on theirs. Moreover, this comparison will not prove convincing if the program is generative, in which case the program induces different results from an execution to the other, or interactive, in which case the observed result depends on the reader’s actions.

24Technical lability is unpredictable and out of control. The program cannot measure it, because there is no technical feedback between the program and the result displayed on the screen. Technical lability can even go as far as to break up the executed program, so that the latter can no longer be considered as a description or model of the observed result. I have already witnessed the temporary interruption programs’ parts, while others would continue, which would result in a display against the program’s logic, with no apparent mistake and perfectly reproducible. Obviously, that kind of technical behavior has a deep impact on the construction of meaning, as semiotic lability is extreme.

25However, one can demonstrate that lability does not affect all programs in the same way. In a program, three usually intertwined categories of instructions are to be observed: generative instructions which create a memory representation of media (text, music, image), instructions which manage the interaction with the machine (management of inputs, outputs, memory, disc reading…) and instructions which ensure that media are displayed on the screen. Technical lability may affect all three instructions categories. On the contrary, semiotic lability will mainly impact the third category, as it is closely related to perception. Undoubtedly, this is the main reason why it was not evidenced before the development of animated poetry. This particular form of poetry mainly resorts to this third category of instructions.

Lability and perspective

26Although the identification of lability results from the development of animated poetry, it is not exclusively linked to the technical properties of these productions. For technical lability to induce semiotic lability, the elements undergoing this technical lability must be ascribed with textual meaning, which is not necessarily the case.

27Here is an example taken from the field of generation. Let us note that the following observations, relating to authors or readers, can be applied to all forms of digital literature. Jean-Pierre Balpe is one of the most important authors of the “texts generation” branch of digital literature. In the 1990s, his generators would take a lot of time to build a memorized text and nothing would be displayed on the screen during that time. The author then had the idea to display parts of the elements involved in the generative program during the generation process. Readers would try to predict the generated text before its display. With the improvement of computers’ processing speed, this calculation phase became shorter and shorter, while the display of these elements grew more and more “ghostly,” mostly because their duration was too short to allow for the anticipation of the generated text. However, Jean-Pierre Balpe has never ascribed these transitory states with meaning: for him, the text of the work (in the semiotic sense) was limited to the alphanumeric text that was generated, which was not influenced by lability. Simultaneously, he attributed a paratext value to the transitory elements undergoing lability. Therefore, semiotic lability is not only a matter of perception, it also depends on the value ascribed to the element undergoing lability.

Lability and obsolescence

28Technical lability leads us to make a distinction between what falls within the perennial, i.e. the source-program written by the author, and what falls within the temporary, i.e. the result that can be observed during the execution. While this result is undoubtedly a temporary state, the source-program remains a perennial object which is still productive as long as the technological context allows it. Let us note that several technical operations are performed from the time the author writes the program until the reader executes it. The source-program is the object created by an author in their programming context. It may be a simple text when the program is directly coded (in Java, C programming language…) or a more complex object manifesting itself in the programming environment in a series of texts (scripts), or images when this environment is a graphic one (Flash, Pure data…). In any case, the source text cannot be self-executed. However, it can easily be printed or captured (image capture) and then transferred to another medium. In this sense, it is perennial as the variability of the media does not affect it. In order to be executed, this source text must be interpreted or compiled. When interpreted (a HTML file, for instance), the functional dimension of the code, and thus lability, is transferred to the interpreting program. When compiled, the first step consists in coding it in a hexadecimal language before adding the hexadecimal libraries that it uses. As a result, we are provided with an object file that is not yet executable. The last step consists in linking this object file to the operating system, so as to obtain the executable file. The latter is no longer perennial, as it is the one undergoing lability.

29It is precisely this perennial characteristic of the source program that allows certain preservation techniques to prevent obsolescence. Obsolescence only affects hardware. It may be offset through programming, in the form of various artifacts: simulation, emulation, virtual machines… This is the reason why the very first digital poem that I programmed on a family computer in the mid-1980s was made available to the public only recently: there is now a PC emulator of the machine I used back then. Obviously, the execution of the program in the said emulator does not reproduce the exact esthetic behavior of the initial machine, but it is quite similar and neutralizes obsolescence (not lability) anyway.

Perception and reading

30In the light of the above, the material system of digital productions can clearly be split into two orders: the first one is a perennial element consisting of the source program and its data, taking a non-binary and intelligible form that the author can manipulate —though not directly the machine— which we shall call the “source.” The second one consists of transitory elements: the machine that uses and usually transforms this source in order to create an executable file, as well as the states resulting from its execution, which we shall call “observable transitory.” The prevailing view often has the work mistaken for its states, because the reception of the work is often mistaken for the perception of these states. Although dominant, the relationship between the reception of the work and perception of part of the apparatus is sometimes ineffective.

31Mistaking the work for the observable transitory is not always justified. Some processes do not produce an observable transitory, for the poetic nature of the work lies only in the program. This is the case of perl poetry (digital poetry written in the programming language called Perl) and certain obfuscation approaches, a technique which consists in making the source incomprehensible to any human, while preserving its computer relevance to the machine. In other instances, the source program may be decomposed to show that it contains esthetic representations which are neither updated during its execution, nor written in the program.

33In this way, without even appealing to a potential conceptual dimension that is sometimes part of writing projects, the relation between the work and perception of the execution’s result is not obvious at all. We too often refuse to admit what is yet considered as obvious in printed texts analyses: the materiality of the work must be decomposed, so as to reach degrees of perception that are unattainable through the usual reading apparatus. In the case of digital productions, this means that one must experiment on the programs, decompose them, isolate parts of them, make video captures of the produced states, process these recordings to retrieve data that we cannot directly perceive through our senses.

Perception and power to act

Work and perception

35So, the perennial and transitory only concern the perceptible material elements —states or objects— and depend on the perspective taken on these elements. Is the work likely to be identified through these elements? Obviously, there again the answer depends on the point of view. One cannot ignore that certain approaches, mine included, integrate lability as one of the work’s fundamental esthetic components. In this case, it is not technical lability which is concerned, but semiotic lability and its consequences on reading, since no one can pretend that one “has read this text.”

36Lability prevents us from limiting the work’s surface esthetics to our perception of the work at a given time. The transitory dimension constitutes a perennial property of the work. Consequently, any attempt at crystallizing a perceptive identity, like preservation approaches tend to do, only hampers its power to act, or even destroys the esthetic project.

38Another part of passage resorts to an adaptive generator. It is a program that takes measurements on the machine and adapts its functioning according to the results. Consequently, I was able to design a program that none of the then available machines could run entirely, which would yet produce a result on every machine. I have not yet seen what a full running of the program would result in, as none of the machines I have run it on was capable of executing all of its parts. In his review of A hundred thousand billion poems, a journalist was surprised that Raymond Queneau had produced a work “he himself had not read.” In the same way, he could be astonished that I have created a program “the result of which its own author has not seen.” Lability is here clearly accepted as a future inherent to the work. Obviously, it is not accepted for its perceptive result, which remains utterly unpredictable. In this case, perception is but a modality semioticizing lability.

The power to act in the work

39Perception can be exploited in the work. It can constitute a component of esthetic representation, in which case one no longer reads the work, but reads within the work. The reader becomes an extendible part of the work, through a relation that is specific to its essence, updated in the reading process, be it a narrow reading or meta-reading.

40The traditional material aspects of the work remain largely inadequate to define the power to act, which may also consist in a power to contest, a subversive and radical questioning of common practices, or a unifying power which, in any case, is subjected to a singular relation to the language defining its poetic nature. From then on, preserving the work no longer consists in preserving its materiality, but in allowing for updates of its fragmented power to act. I highly doubt that we will ever be able to simultaneously update all aspects of the power to act, because they are expressed through relationships endowed with different temporalities. Of course, the power to act is partly implemented through narrow reading and although legitimate, the will to save a readable and perceptible state of the observable transitory is not sufficient. It does not even ensure the preservation of the work’s initial “acting,” because the cultural context of reception is transformed in an irreversible and non-reproducible way. We also have to admit that other aspects of this power are operated through the decline of perception —the “perceptive death”— and that this destructive process must be preserved too if we are to update these aspects of the power to act. For instance, a detailed analysis of Jean-Marie Dutey’s mange-texte (Philippe Bootz, 2012) shows that the unsolvable loss of the perception of the observable transitory meets the work’s narrative objective, while its author could not have planned it, as the lability phenomenon had not yet been discovered at the time this work was created.

Conclusion

41Thus, far from being trivial, the issue of the ephemeral has led us to question the relation between perception and reception, to tackle that of preservation and ultimately, the very definition of the digital work.