Umberto Eco, small worlds, and the assumptions we bring to simulations

Ive been reading Umberto Ecos book, “The limits of interpretation” (1990), and especially the essay on “Small Worlds”, by which he means, in effct, fictional worlds. I take this to mean simulated worlds as well. Its very enlightening on the ways in which simulations work for us.

He discusses the philosophical reality of “possible worlds”: “a fictional possible world is a series of linguistic descriptions that readers are supposed to interpret as referring to a possible state of affairs where p is true if non-p is false (such a requirement being flexible since there also are, as we shall see, impossible possible worlds.”)

In this sense it occurs to me that the difference between reality and simulation may simply be that the creator of a simulation understands the rules of the simulation, whereas nobody understands the rules of life. (For several reasons;
– metaphysically: Kant suggested and the Existentialists shouted that human reason can never understand reality;
– in terms of complexity: I have to discern not only the metaphysical validity of my interpretation of the rules, but also the psychology which makes the human mind arrive at any sort of interpretation, and also the socio-economic factors which made my particular mind arrive at my interpretation, the semantics of the words I use to express my interpretation for myself and to you, and so on.)

This definition starts to break down though once the simulations become interactive. Designers of multi-player simulations seem to be constantly surprised by things the players do: the richer the options open to the players, the less the designers can anticipate every outcome. But still, to a certain extent, if there is a computer programme, somebody has the source code, and this is not the case with life. (Or at least, not the case for me!)

Euclidean geometry is not a possible world but an abstract representation of the actual world. (Although Flatland of course is a possible world.) Simulations may start off as an abstract representation of the actual world, and in some cases (eg mathematical simulations of possible networks, used to optimise designs, or demonstrate the acceptability of one design against another) may not go far beyond this. The interpretation of the possible world is largely mathematical, or done by machine.

Any more complex possible world relies on the reader (viewer, user, etc.) adding his/her own cultural assumptions to the textual (or artistic) representation, and taking many shortcuts. (If Conan Doyle says that Baker Street was “gas-lit” he can rely on us to fill in the meaning of the word, even though we will all fill it in rather differently.)

“Possible worlds” are not as complex as the real world: Eco quotes Dolezel: “fictional worlds are incomplete and semantically unhomogeneous: they are handicapped small worlds”. Particularly in a computer simulation, there is a model, which exists (as a programme or algorithm) and which somebody can (and hopefully does) understand completely and perfectly.

However Ecos insight is that the simulation begins to depend on the cultural references the user brings to it. He uses examples from fiction: when an 18th century English novel mentioned a southern European olive plantation, the English reader might never have seen such a thing, but would supply his own conception of the scene: “probably the world he or she had in mind is different …[from the one the author had in mind>… but this does not matter. For the purposes of the story, every cliche-like conception of a French landscape can work.” In fact the reader does not have to imagine conscientiously each scene: “it is sufficient that he or she pretends to believe to know them.”

Cultural accretions by the user can be triggered by tiny references or clues: if I begin a story “He reached for his cutlass…” you guess at once that the story may be about pirates, the best known of all cutlass users.

This ability to pick up tiny clues, but at the same time to gloss over whole areas of ignorance, is essential to simulations. See the article from Presence, referred to earlier, about the way that users pick up visual signs from avatars, make social assumptions, and tailor their behaviour accordingly. Equally, combat simulations may demand that the trainee accepts that there is no realistic smell to the simulated battlefield, when much war literature suggests that there are very distinct and striking smells (of munitions, petrol and decomposition, for instance) that are remembered as an essential part of real battle experience.

Going back to p and not-p, Eco identifies what he calls s-necessary properties – these are the properties which are essential to the fiction: if Gertrude is the mother of Hamlet, then Hamlet is the son of Gertrude. In combat simulations, the enemy does not suddenly turn into Roger Rabbit. Without consistent s-necessary properits, the possible world becomes impossible.

Eco contrasts impossible drawings (eg the Penrose triangle) with impossible fiction. Both portray impossible worlds. But the medium makes a difference: it is easier to spot impossibility in drawings, because you comprehend all the facts at once. Texts (or simulations) take time to unfold; to spot impossibilities we must notice incongruities which only appear over time, and compare these with memories of earlier events: so it may take a while to realise that the simulated world is impossible. Or the incongruities may be so complex or so subtle that we never do realise them. (This, put baldly, is my concern about many business simulations: they may have more hype than s-necessary properties.)

Eco also uses a kind of truth table approach to compare different worlds within a literary text. (Oedipus Rex by Sophocles.) He draws up matrices to represent the properties of two worlds, and discusses how it would be possible by a kind of algebra to derive the properties of one world from the other, in which case we can say that either world is conceivable from the view of the alternative one, or not. Oedipus is particularly suitable for this demonstration, because Oedipuss world-view does not include (until the end) that he is Jocastas son as well as her husband, and so on. So his world has the property oMj (Oedipus is married to Jocasta) but not oSj (Oedipus is the son of Jocasta); whereas the world of the oracle Tyresias includes both properties. Eco draws this all out as a diagram. This may sound irrelevant, but look at the uses of truth tables in formal logic and their many applications.

It seems theres a whole area of thought about possible worlds. Fodder for another entry.

Leave a Reply

Your email address will not be published. Required fields are marked *