… But often such an object, which because of one or several of its properties has been placed in one class, belongs to another class by virtue of other properties and might have been placed accordingly. Thus, the general division remains of necessity somewhat arbitrary.29
D’Alembert’s discourse still suffers from an unresolved tension between the model of the tree and the model of the map. It becomes clear that the sum of our knowledge (present, but also, as it was for Leibniz, future) extends like a geographical map without borders, within which infinite itineraries are possible. But, given that the Encyclopédie, in its printed form, is in alphabetical order, one knows one will need to resort to a number of reductive strategies.
What we already have, however, is a first hint at the ideal model of an encyclopedia, that is, a hypothetical compendium of all of the knowledge available to a given culture.
1.4. The Maximal Encyclopedia as Regulatory Idea
The encyclopedia is potentially infinite because it is forever in fieri, and the discourses we construct on its basis constantly call it into question (in the same way in which the latest article by a nuclear scientist presupposes a series of encyclopedic notions concerning the structure of the atom, but at the same time introduces new ones that render the old ones moot).
The Maximal Encyclopedia is not content with merely recording what “is true” (whatever meaning we may choose to give to this expression). It records instead everything that has been claimed in a social context, not only what has been accepted as true, but also what has been accepted as imaginary.
It exists as a regulating principle: yet this regulating idea, which cannot constitute the starting point for a publishable project because it has no organizable form, serves to identify portions of encyclopedias that can be activated, insofar as they serve to construct provisional hierarchies or manageable networks, with a view to interpreting and explaining the interpretability of certain segments of discourse.
This encyclopedia is not available for consultation in toto because it is the sum total of everything ever said by humankind, and yet it has a material existence, because what has been said has been deposited in the form of all the books ever written and all the images ever made and all the evidential items that act as reciprocal interpretants in the chain of semiosis.
Having become transformed over the centuries from an (attainable) utopia of global knowledge into an awareness of the impossibility of global knowledge, but with the certainty of the local availability of the elements of this knowledge, no longer the project for a book, but a method of investigation addressing the general and omnivorous library of culture in its entirety, the Maximal Encyclopedia was envisaged in poetic terms by Dante, when, in Canto 33 of his Paradiso, as he finally attains the vision of God, he is unable to describe what he saw except, precisely, in terms of an encyclopedia:
In its profundity I saw—ingathered
and bound by love into one single volume—
what, in the universe, seems separate, scattered:
substances, accidents, and dispositions
as if conjoined—in such a way that what
I tell is only rudimentary.
I think I saw the universal shape
which that knot takes; for, speaking this, I feel
a joy that is more ample. That one moment
brings more forgetfulness to me than twenty-
five centuries have brought to the endeavor
that startled Neptune with the Argo’s shadow!30
The encyclopedia is the only means we have of giving an account, not only of the workings of any semiotic system, but also of the life of a given culture as a system of interlocking semiotic systems.
As I have shown elsewhere (see, for instance, Eco 1975), from the moment one takes the route of the encyclopedia, two theoretically crucial distinctions are lost: (i) in the first place, that between natural language and other semiotic systems, since properties expressed in nonverbal form can also constitute part of the encyclopedic representation of a given term or corresponding concept (in the sense that a potentially infinite number of images of dogs are part of the encyclopedic representation of the notion “dog”); and (ii) in the second place, the distinction between semiotic system as object and theoretical metalanguage. It is impossible in fact to create a metalanguage as a theoretical construct composed of a finite number of universal primitives: such a construct, as we have seen, explodes, and when it explodes it reveals that its own metalinguistic terms are nothing other than terms of the object language—though they may be used provisionally as not susceptible of further definition.
The encyclopedia is dominated by the Peircean principle of interpretation and consequently of unlimited semiosis. Every expression of the semiotic system is interpretable by other expressions, and these by still others, in a self-sustaining semiotic process, even if, from a Peircean point of view, this flight of interpretants generates habits and hence modalities of transformation of the natural world. Every result of this action on the world must, however, be interpreted in its turn, and in this way the circle of semiosis is on the one hand constantly opening up to what lies outside and on the other constantly reproducing itself within.
Furthermore, the encyclopedia generates ever new interpretations that depend on changing contexts and circumstances (and hence semantics incorporates within itself pragmatics). Therefore we can never give it a definitive and closed representation: an encyclopedic representation is never global but invariably local, and it is activated as a function of determined contexts and circumstances. The expression “dog” occurring in a universe of discourse regarding fireplace furniture generates different interpretants from the same expression occurring in a universe of discourse regarding animals; while, within a discourse on animals, the same expression generates different ramifications of interpretants depending on whether the subject is zoology or hunting.
1.5. Labyrinths
D’Alembert spoke of a labyrinth, and he naturally attempted to express the concept through that of a map, without, however, being able to speak of the topological model of a polydimensional network. The Porphyrian tree represented an attempt to reduce the polydimensional labyrinth to a bidimensional schema. But we have observed how, even in this simple classificatory instrument, the tree regenerated the labyrinth (of differences) at every fresh step.
We must first reach a consensus on the concept of labyrinth, because labyrinths come in three varieties (cf. Santarcangeli 1967; Bord 1976; Kern 1981). The classic labyrinth of Cnossos is unicursal: there is only one path. Once one enters one cannot help reaching the center (and from the center one cannot help finding the way out). If the unicursal labyrinth were to be “unrolled,” we would find we had a single thread in our hands—the thread of Ariadne which the legend presents as the means (alien to the labyrinth) of extricating oneself from the labyrinth, whereas in fact all it is is the labyrinth itself.31 The unicursal labyrinth, then, does not represent a model for an encyclopedia (Figure 1.15)
The second type is the Mannerist labyrinth or Irrweg. The Irrweg proposes alternative choices, but all the paths lead to a dead point—all but one, that is, which leads to the way out (Figure 1.16). If it were “unrolled,” the Irrweg would assume the form of a tree, of a structure of blind alleys (except for one).32 One can take the wrong path, in which case one is obliged to retrace one’s steps (in a certain sense the Irrweg works like a flowchart).
Figure 1.15
Figure 1.16
The third kind of labyrinth is a network, in which every point may be connected with any other point (Figure 1.17).
Figure 1.17
A network cannot be “unrolled.” One reason for this is because, whereas the first two kinds of labyrinth have an inside and an outside, from which one enters and toward which one exits, the third kind of labyrinth, infinitely extendible, has no inside and no outside.
Since every one of its points can be connected with any other, and since the process of connection is also a continual process of correction of the connections, its structure will always be different from what it was a moment ago, and it can be traversed by taking a different route each time. Those who travel in it, then, must also learn to correct constantly the image they have of it, whether this be a concrete (local) image of one of its sections, or the hypothetical regulatory image concerning its global structure (which cannot be known, for reasons both synchronic and diachronic).
A network is a tree plus an infinite number of corridors that connect its nodes. The tree may become (multidimensionally) a polygon, a system of interconnected polygons, an immense megahedron. But even this comparison is misleading: a polygon has outside limits, whereas the abstract model of the network has none.
In Eco (1984b: ch. 2), as a metaphor for the network model, I chose the rhizome (Deleuze and Guattari 1976). Every point of the rhizome can be connected to any other point; it is said that in the rhizome there are no points or positions, only lines; this characteristic, however, is doubtful, because every intersection of two lines makes it possible to identify a point; the rhizome can be broken and reconnected at any point; the rhizome is anti-genealogical (it is not an hierarchized tree); if the rhizome had an outside, with that outside it could produce another rhizome, therefore it has neither an inside nor an outside; the rhizome can be taken to pieces and inverted; it is susceptible to modification; a multidimensional network of trees, open in all directions, creates rhizomes, which means that every local section of the
