CREATIVITY AND CONSTRAINT

Yazdani (1989) describes the process of story writing as an author's presenting a multi-faceted description of an imagined world under the constraints of seriality of the text medium. The author achieves artistic effects by controlled revelation of the world view through ordering of selected elements and the texture of expression. Educators and microworlds designers might likewise inquire what are the constraints of our medium against which we struggle to realize our art.

J. R. R. Tolkein, author of the heroic fantasies "The Hobbit" and "The Lord of the Rings", one of the most popular writers for the literati in this half century, is a world class creator of virtual worlds. Tolkein's achievements and ideas can help illuminate the challenge before the would-be microworld designer. In the course of his academic career, Tolkein described his sense of the roots of fantasy in fiction:

"The incarnate mind, the tongue, and the tale are in our world coeval. The human mind, endowed with the powers of generalization and abstraction, sees not only green-grass , discriminating it from other things (and finding it fair to look upon), but sees that it is green as well as being grass . But how powerful, how stimulating to the very faculty that produced it, was the invention of the adjective: no spell or incantation in Faerie is more potent.... The mind that thought of light , heavy , grey , yellow , still , swift , also conceived of magic that would made heavy things light and able to fly, turn grey lead into yellow gold, and the still rock into swift water. If it could do the one, it could do the other; it inevitably did both. When we can take green from grass, blue from heaven, and red from blood, we have already an enchanter's power... But in such fantasy, as it is called, new form is made; Faerie begins; Man becomes a sub-creator...."
from Tree and Leaf

Changing the process itself leads to new creations, a new microworld with its own unusual beauties. When the increment of the POLYSPIRAL procedure is applied to the angle value instead of to the side-length of figures drawn, the POLYSPI becomes the INSPI, capable in its turn of its own independent elaborations and semantics. Changing a process creates more variability (with new kinds of confusion and chaos) as well as new kinds of order. POLYSPI-like designs were made before computing. I doubt that this is true of INSPI designs. So there is something new under the sun.

The creation of new things is more striking than the attribution of aberrant qualities. Most of the specific creatures of Tolkein's fancy can be seen as mixtures of qualities. But there is one new creation comparable in novelty to the INSPI where one can examine the process by which the creature became who or what it is: Smeagol (aka Gollum, the riverine hobbit gone-wrong). In this specific sense, different character development through varied experiences is the literary analog of process modification creating INSPI.

What are the constraints ? For the literary genres, the processes are worked out in tales, the telling of which takes time. For computer microworlds, processes are less time-constrained. They go forward faster and can be repeated. In literature, the revelatory process is development in time. In computer microworlds, the revelatory process is an unpacking of component parts through analysis in detail.

FATE AND LEARNING

The primary satisfactions of literature derive from aroused expectations and their resolution by the end of the work. One measure of the quality of a literary work is the sense of necessity developed as the story unfolds. If we seek to design microworlds through which people can learn without instruction, we should be able to specify what is the deep content that ultimately will be discovered by people who have the experience and do not miss its meaning. What is "fated" to be discovered ? What is the equivalent of "fate" in a timeless medium ? Here is one answer to consider. One element of cognitive power is the ability to deploy alternative descriptions of circumstances and problems; further this deployment is enhanced by integration of the different perspectives made possible by the different descriptions. Consider increasing complexity in Logo turtle geometry as an example. One of the best slogan's for the Logo microworld ideal is "no threshold; no ceiling." One can make a circle with the simplest composition of primitives. One can make a 3-dimensional projective sketch of the Piazza San Marco, if you know how.[1] One should not assume, however, that traversing that path is trivial simply because there is no threshold. There may be giant steps along the way. Compare a circle's description as a Logo procedure with a Cartesian equation of a circle

One can make the Cartesian circle into an ellipse by adding non-zero and non-identical coefficients to the X and Y terms. How does one make the Logo circle into an ellipse ? It is possible but not quite so easy.

The first point is that different descriptions are useful for solving different specific problems. A second point is that one better understands both descriptions of objects when the descriptions are brought into contrast. Multiple descriptions permit the relativizing of each one involved. This is equally true for simple descriptions and complex ones. Beginning with turtle geometry -- realized and embedded in hardware systems implemented in Cartesian reference frameworks, represented by the X-Y plane of reference for a video display, one is destined to learn about both descriptions of space and objects and therefore also about their interrelations. This is a "fated" learning outcome. This is what a microworld user will discover if there is interest enough and time.

There is a second answer to the question of what is ultimately to be discovered in a microworld: the interrelation of structure and function of its elements. Consider the fact that Logo is a procedurally-structurable, recursively-interpreted language. These are the primary features of its function. Notice also that its primary data structure is a list, each of whose elements are either symbols or other lists. Once when discussing Logo procedures as data-structures, a colleague remarked the insight noting, "Ah! It is no accident that Logo is both a list-structured and procedural language." Exactly so. Because Logo data-structures are recursively interpreted lists of lists, Logo procedures can be defined as structures permitting recursive invocation limited only at the time of interpretation.[2]

ILLUSION AND COMPREHENSIBILITY

Nabakov's observations on some objectives and techniques of writing appear prominently in Transparent Things, most notably his appreciation of the critical issue of the relationships between the surface and depth of descriptions. His first chapter focuses on the challenge of exposition under the constraint of maintaining an illusion of the seeming reality of the present time. [3] He says it this way:

"When we concentrate on a material object, whatever its situation, the very act of attention may lead to our involuntarily sinking into the history of that object. Novices must learn to skim over matter if they want matter to stay at the exact level of the moment. Transparent things, through which the past shines.

Man made objects, or natural ones, inert in themselves but much used by careless life... are particularly difficult to keep in surface focus; novices fall through the surface, humming happily to themselves, and are soon reveling with childish abandon in the story of this stone, of that heath. I shall explain. A thin veneer of immediate reality is spread over natural and artificial matter, and whoever wishes to remain in the now, with the now, on the now, should please not break its tension film. Otherwise the inexperienced miracle-worker will find himself no longer walking on water but descending upright among staring fish...."
from Transparent Things

If we cast our minds back to earlier days of the AI community, we will remember that the choice always needed to be made between the strategies of depth first or breadth first in searching of large information spaces. For the different effort we make in education, understanding and communicating information more than searching for it, our appropriate contrast should be between issues of depth and issues of the surface, where the maintenance of coherence and comprehensibility across the fringes of an arena of interest is our best analog of maintaining the illusion produced by the magicians of language. A flexible banded editor might help us do that.

References
Minsky, M. The Society of Mind , New York, Simon & Schuster, 1985.
Nabakov, V. Transparent Things, Greenwich, CT. Fawcett Books, 1972.
Tolkien, J. R. R. Tree and Leaf, London, Unwin Books, 1964.

Yazdani, M. "Computational Story Writing," in Computers and Writing, Williams and Holt (Eds.) Norwood, NJ. Ablex, 1989.

Publication notes:

• Written in 1992. Accepted as an abstract for a keynote talk to be presented at ICTE in1993.

Text notes:

1. Such has been done by Horacio Reggini using his three dimensional Logo.
   
2. It is only a small step from that observation to notice that Logo embodies in software the vision of a uniform memory where program and data are identical, differing only by whether they are accessed as program (memory-contents fetched for execution as an instruction) or data (memory-contents manipulated by an executed instruction).
   
3. Chapters two and three of this work show how the illusion is worked out -- through seemingly casual, descriptive details to be recalled and used later -- and how the process can go completely awry when one breaks "the tension film".
   
4. This suggestion derives from Minsky's Level-band principle in the Society of Mind (1985). It seems reasonable that his characterization of how the mind might work should help us understand how we might work better.