Halting, Sphexishness, and Analysis, Terminable and Interminable
ACM SigWeb Newsletter, 2000
Cognate problems in computation, animal behavior, and psychotherapy are employed as lenses into issues confronting hypertext narrative. The suggestion is that a similar need for jumping out of the system drives self-transcendence in these areas — with potential significance for the construction of hypertext narratives.
Hypertext fiction, hypertext narrative, hypertext literature, artifactual hypertext
With a title like "Halting, Sphexishness, and Analysis, Terminable and Interminable," I feel obligated to begin by explaining just what the heck this paper is actually about. It is about hypertext fiction. In the next few pages, I will make a quick foray into Sumerian mathematics, explicate parallel self-transcendence problems in the three domains of my title, invoke E.O. Wilson's consilience, talk about roller coasters, explain why hypertext fiction has failed to catch on, and suggest what it must do to succeed going forward.
I hope, in this, to provide enough hooks into hypertext fiction to persuade at least some of you that the problems with which this nascent medium grapples can be useful lenses into some of the larger issues facing hypertext systems in general.
SUMS AND SUMER
First, to Sumer. Denise Schmandt-Besserat has argued convincingly that written text was "the byproduct of abstract counting." The linearity of text, which we take as a primary characteristic, is rather a side-effect of the need for a fixed sequence of mark-making in the recording of numerical transactions.
What begins as a literal series of clay tokens, each representing a unit of some good, is gradually abstracted further and further. First, the tokens are gathered up in an envelope. Then images of the tokens are impressed on the envelope as a redundant encoding of the tokens therein. Finally, in what Baudrillard would glibly dismiss as the precession of the simulacra, the tokens are realized to be unnecessary, and the marking on the envelope acquires symbolic status.
Critical to the success of this ascendancy of the abstract is a fixed order of operations in mark making. When you can no longer break open the envelope and physically count the number of objects, one must have rules — among which Schmandt-Besserat enumerates semanticity, discreteness, systematization, codification, and syntax. "With the invention of numerals," she adds, "pictography was no longer restricted to accounting...From then on, writing could become phonetic and...store and convey any possible idea."
So, if we find Schmandt-Besserat's research compelling, we must agree with her that in writing, "the linear and hierarchical sign order continued the way that accountants had formerly manipulated tokens."  The directionality and linear connectedness of written language can thus be seen as an epiphenomenal manifestation of its origin. Let's keep this in mind, while we fast forward seven thousand years, to 1936, and pick up the first word from the title of this talk, "halting."
In computer science, one of the theoretical limits of computation is defined by what is called the halting problem, so named because it asks if, whether given a specific input, a program can be determined in advance to reach an answer in a finite number of steps and halt.
First posed by David Hilbert, it takes the form of a hypothetical challenge: given a specific theorem derived from the functional calculus K, determine whether the theorem is provable.  Alan Turing, in his 1936 paper "On computable numbers," demonstrated that, in principle, it is impossible for a computer to make this determination. Turing accomplishes this by envisioning a most simple computer, a Turing machine, which attacks problems in a series of finite steps. By coding up the description of the machine itself as a proposition, Turing establishes that there must necessarily be theorems which any given machine cannot print out.
While often lumped together with Godel's Theorem, the halting problem is actually a much more concrete result. Where Godel posits theoretical limitations on the provability of statements within axiomatic systems of a given robustness (arithmetic being one of them) the halting problem intrudes directly upon the capacity of computers to algorithmically generate a given statement. Put in human terms, for each computer, there will be a sentence which the computer can never, in principle, speak.
What the halting problem adds to our meandering path thus far is the making visible of an implicit limitation imposed by text. The semantics and syntax of mark making in Sumer — the propositions of formal arithmetic — when coded up in a computer system, generate an intractable theoretical barrier. Or at least — and we shall come back to this — when considered from within this system.
From the characteristics of abstract formal systems, I'd like to move on now to animal behavior, and the second part of my title, "sphexishness." This adjective, coined by Douglas Hofstadter in "Godel, Escher, Bach,"  describes a peculiar fixity of behavior which some animals exhibit. Wooldridge, in his book "Mechanical Man," describes the experiment  — conducted with the species of wasp called the sphex — which gives rise to the term.
The female sphex wasp appears to be genetically hard-wired to perform a certain series of steps in preparing its egg chamber. It will search out and paralyze a particular breed of caterpillar, and bring it to the mouth of a burrow which it has previously prepared. It then enters the burrow to check to be sure that everything is okay before dragging the paralyzed bug inside, where it will implant its eggs and seal up the nest. If one interrupts this cycle — by moving the paralyzed insect a few inches away from the mouth of the nest — the sphex will not simply pick up where it has left off.
Instead of dragging the insect into the burrow which she has just inspected, the sphex will instead move the bug back to its position at the mouth of the burrow, and leave it there as she enters the burrow to perform her inspection. Wooldridge describes an experimenter moving the bug forty times, with the sphex repeating the same behavior each time.
Hofstadter makes much of this, assigning the label "sphexishness" to those characteristics of behavior — whether animal, human, or machine — which exhibit this pre-programmed character. Not that this is always a bad thing. Rather, as Hofstadter points out, there are many times when the pre-programmed response is correct. And to slide a little E. O. Wilson in at this point, the evolutionary forces which selected for deep behavioral traits are likely to have produced successful strategies. Sphexes unencumbered by prey are less likely to be overpowered by potential predators in their nests.
But we humans like to think of ourselves as creative — what Hofstadter calls "antisphexish." We are not bothered by Godelian sentences. I have no trouble with the sentence, "this is a sentence which I will never speak." I just, well, speak it. And unlike the sphex, we do not mindlessly repeat behaviors. No matter what our therapists say.
ANALYSIS TERMINABLE AND INTERMINABLE
Well, okay, maybe we should think about what therapists say, since we covered the first two-thirds of the title of this talk, and what remains now is the question of analysis, terminable and interminable. As you may be aware, this was the title of one of Freud's last papers on analytic theory, in which he grapples with the question of completion in the analytic process.
He offers two measures for a terminable analysis. First, a practical one: "when so much repressed material has been made conscious, so much that was unintelligible has been explained, and so much internal resistance conquered, that there is no need to fear a repetition of the pathological processes concerned."  The second, more rigorous test, is the possibility of having "such a far-reaching influence on the patient that no further change could be expected to take place," or what he describes as "a level of absolute psychical normality."
Freud finds the second, more interesting criterion, unlikely for a variety of reasons. Significantly, he singles out etiologies hinging on a "constitutional strength of instinct"  as being highly resistant. (But the sphex knows that...) Freud also calls into question the process of analysis itself. Analysis can only be prophylactically effective "in so far as a possible future conflict is turned into an actual present one upon which influence is then brought to bear."  That is, analysis can only treat what it is presented with in the current situation, and while Freud remains vague about the degree to which future recrudescence of a treated complex may be prevented, he explicitly disclaims the potential for analysis to activate a complex. "If an instinctual conflict is not a currently active one, is not manifesting itself, we cannot influence it even by analysis." 
Given the generic framework of the analytic enterprise, which might be summed up as guided insight into narratized behavior, it would appear, then, that analysis is in principle interminable. Freud stops short of ceding this point, though I hasten to add that I am in no way considering him the last word on this subject. Freud's paper is employed here as a probe into yet another system exhibiting a kind of interminability, this one all the more troubling since it suggests a limitation which may be operative in the kind of consciousness which we are all exhibiting at this very moment.
THE UR-SYSTEM OF NARRATIVE
What is the common thread which connects these examples? To try and answer this question, I will first call upon science, and then turn to a consideration of amusement parks.
First, the science. E.O. Wilson, in his book "Consilience," argues passionately for a coherent vision of the intellectual enterprise, grounded ultimately in our scientific knowledge of the external world. Now I know that to invoke Wilson is to risk being labelled a reductionist. To those who would make that claim, I would suggest that a close reading of the book does not support such an easy categorization. Consilience, from the Latin for "jumping together" is, as Wilson defines it, the search for commonalities of law across levels of organization — not reducing everything to the physical or biological level, but not refusing to acknowledge this component.
Wilson takes as a given that "our brain and sensory system evolved as a biological apparatus to preserve and multiply human genes. But they enable us to navigate only through a tiny segment of the physical world whose mastery serves that primal need. Instrumental science removes the handicap."  What Wilson is arguing for here is not "seeing everything reduced to biology," but rather understanding the character of any epigenetic (by which we mean interacting genetic and cultural) biases which may be operating. Physics tells us that light exists as a spectrum; the evolution of the eye and brain limit our perception to highly determined bands of color.
Wilson does not believe that higher functions in thought are controlled by such low-level biases, but that they are influenced. "Rational choice is the casting about among alternative mental scenarios to hit upon the ones which, in a given context, satisfy the strongest epigenetic rules. It is these rules and this hierarchy of their relative strengths by which human beings have successfully survived and reproduced for hundreds of millenia."  An example of this kind of influence would be the pan-human response to snakes. Evolutionarily, we appear to be genetically programmed to respond; on the mental and cultural level we convert this into myth and image consilient with these genetic constraints.
I would argue, following Wilson, that our ur-narrative constructs (selected for by evolution at some remove) suffer from similar gaps. To come back to Schmandt-Besserat: narrative, like accounting, is all about successorship. This happened and then this happened. The queen died and then then king died of grief. This leads to the interminable behavior which surfaces in formal systems such as those that Hilbert and Turing study, but also in the recursive linguistic maze of analysis. And always, there is the sphex, reminding us that interminability lives deep down in our animal brains — a part of us which is never eradicated, since evolution does not proceed by substitution, but rather accretion.
If all that this was saying is that humans have an innate tendency to resist change, you would be forgiven for saying that I've wasted the last ten minutes of your time. That is not what I'm up to. Rather, I'm arguing that the notions of narrative which we are epigenetically predisposed to create are self-limiting, circular, and tautological. A million years of experience in the macroscopic human lifeworld engenders a reification of cause and effect, of actors and agency. Of textual satisfaction as linear trackability.
We like linear narrative because it codes up deep epigenetic theories about experience. Theories which were selected for because they were verified, time and again, by tests conducted within the constraints of the system that produced them. Only as the boundaries of experience have been instrumentally expanded have these theories been problematized. Ernst Cassirer  called relativity "the shattering of the highest law of motion taught us by experience." Even more troubling notions like quantum tunneling and non-local causality threaten the very core ideas of narrative. If things are neither here nor there, and information can travel faster than light, the satisfaction and utility of linear storytelling may finally be facing a challenge it cannot solve from within its circular system.
THE CONEY ISLAND OF THE MIND
In the spirit of consilience and the scientific method, I will wrap up this paper with an attempt to convert my speculations to a testable hypothesis. To do that, since one should always couch science in the most entertaining framework possible, I will talk about amusement parks. Particularly those most paradigmatic amusement parks, the worlds of Disney.
Michael Joyce, in discussing the differences between the linear text and the hypertext, has invoked the image of Disney. "Print culture is as discretely defined and transparently maintained as the grounds of Disney World. There is no danger that new paths will be trod into the manicured lawns."  He goes on to talk about, by extension, "disney texts," which, even though they may be enacted digitally, retain their control of the reading experience, and continue to provide the safe, stable experience of the printed word. In contradistinction, he offers the "wonka" text — a name drawn from the book by Roald Dahl and the film "Willy Wonka & the Chocolate Factory."  Joyce notes the distinction between disney order and wonka chaos, between expectations deliberately shaped and met, and expectations deliberately violated.
But I'd like to push the analogy even further. Everything about disney — whether world or text — is hermetic. One goes to Disney World to enjoy the constraints; to revel in the relaxation which comes from simplification and order. One wishes, when visiting "the World," as it is called in Florida, to stay on the property, where everything is taken care of by the self-sealing system. The World continues to accrete new exhibits — like the addition of the Animal Kingdom this past year — to make it possible to have a wide variety of experiences, all carefully constructed to ensure appropriate satisfaction, without needing to jump out of the system. An appropriate emblem for this might well be Disney's signature roller coaster, Space Mountain. A roller coaster — that most relentlessly entropic of amusement park rides — enclosed completely within the monadic darkness of a Disney-made mountain. Such is the self-contained world of the text.
In Willie Wonka — and here I am speaking about the film — there is no such hermeticism. For those not familiar with the story, a young boy, named Charley, wins a raffle run by the eccentric chocolatier Willie Wonka, and joins a group of children invited to tour Wonka's factory. The factory is a fantasy universe much on the model of Disneyworld, but with important differences. As mentioned above, the function of the tour is not to satisfy the expectations of the raffle-winning children, but rather, in a series of disturbing vignettes, to eliminate them by attrition, trapped, each in their own way, by their petty human vices. Charley, as the final remaining winner, is guided into Wonka's inner sanctum on a device called the wonkavator.
In the final narrative move of the film, the wonkavator blows through the roof of the factory, to float over Charley's hometown. The final message of the film is that the ultimate goal of the journey into the center of the labyrinth is to find ourselves back out in the world with newly open eyes. One might note that this is also the lesson which Gilgamesh learns from Utnapishtim in the first written epic. But I digress. I promised a testable hypothesis, and here it is.
Disneytext — or if you prefer, linear narrative — is a self-reinforcing, circular system which will only, can only ever give rise to more linear narrative. To jump out of the narrative system, the mind must be allowed back out of the frame, back into a world of things and processes experienced now with a wider set of perceptual possibilities, in order that new, non-linear notions of narrative may emerge, and be sustained at a different point of dissipative equilibrium. Given that consciousness seems able to jump out of the limiting frames which have been here posed — we utter Godelian sentences, break habits, and do, eventually, leave our therapists behind — I suggest that we will in fact succeed in getting the narrative monkey off our backs.
The evidence will consist in digital storytelling which proceeds not through the use of conventional narrative, but rather, in interactions with the world to be taken as real. Not the pale, poly-agonal representation of the real supported by VR, although the attractiveness of immersive worlds counts as evidence. Rather, if I am correct, the future of fiction will comprise arrays of artifacts, in whose selection and manufacture will be manifested the art and artifice of a new kind of narrator.
And with that, I halt.
1. Schmand-Besserat, Denise. Forerunners of writing. In Writing and its use. Gunther, Hartmut and Ludwig, Otto, Eds. Walter de Gruyter, Berlin, 1994, p. 267
2. Schmandt-Besserat, Denise. Art, Writing and Narrative in Mesopotamia. In Collectanea Orientalia. Gasche, H., and Hrounda, B, eds. Civilizations du Proche-Orient, Series 1, Neuchatel, 1996, p. 318
3. Turing, Alan, On Computable Numbers, with an application to the entscheidungsproblem. Reprinted in The Undecidable, Davis, Martin, ed. New York, Raven Press, 1965, p. 116-151.
4. Hofstadter, D. Godel, Escher, Bach. New York, Basic Books, 1979.
5. Wooldridge, Dean. Mechanical Man. New York, McGraw-Hill, 1968.
6. Freud, Sigmund. Analysis, Terminable and Interminable. In The standard edition of the complete psychological works of Sigmund Freud, Strachey, James, Ed., New York, W.W. Norton, 1976. volume 23, p. 219.
7. Ibid. p. 220.
8. Ibid. p. 230.
9. Ibid. p. 231.
10. Wilson, E.O. Consilience. New York, Random House, 1998. p. 52.
11. Ibid. p. 180.
12. Cassirer, Ernst. Structure and function and Einstein's theory of relativity. New York, Open Court, 1923.
13. Joyce, Michael, The ends of print culture. In Of Two Minds, University of Michigan, 1995. p. 179.
14. Willy Wonka & the Chocolate Factory. David L. Wolper Productions, Warner Brothers. 1971.