Home | Calendar | About | Getting Involved | Groups | Initiatives | Bryn Mawr Home | Serendip Home

2003-2004 Weekly Brown Bag Lunch Discussion
"What Counts? "

September 15, 2003

Anne Dalke (English and Gender Studies)
"What Is the Better Story?"
A Humanist Reflects on the Relation Between Numbers and Narrative

Summary
Prepared by Anne Dalke
Additions, revisions, extensions are encouraged in the Forum
Participants

Anne began by reading Mary Cornish's poem "Numbers," and with it sounded her theme, in 3 notes:

  • how does one decide what to count in the first place?
  • what's left over when one "counts"?
  • what to do w/ the "gift of the odd remainder"?

She invited the group to begin considering such questions with a parable taken from Yann Martel's novel The Life of Pi. The young narrator , Piscine Monitor Patel, escapes from shame of the narrative embedded in his nickname ("Pissing") by re-naming himself with the "irrational" number (3. 14) with which scientists attempt to "make sense" of the universe. His use of number to escape from narrative is only temporary, however. Marooned in lifeboat at sea for 227 days with a Bengal tiger (?!), Pi is very soon thrown much larger questions about the nature of the universe which he cannot answer. When, at the end of the novel, he gives the fantastical account of his adventures, his listeners say that they "do not like his story"; they "want to know what really happened." Pi replies that they want only "dry, yeastless factuality," a story which "has no surprises."

What, then, makes "the better story"? What is the relation between the goodness--or usefulness--of a story and its statistical probability, its surprisingness, its improbability? Is the relation between story and statistic, narrative and number, fixed? Last fall, the "science of culture" series turned repeatedly to the idea that humanists study the improbable, the unique, the "outliers," while scientists seek out the repeatable and the probable. Humanists, in other words, like stories that are not statistically probable, that "surprise," while scientists like stories that are both probable and have predictive value. In taking up, refusing and complicating that binary, Anne intended to suggest that a much more interesting interaction--a much more interactive interaction--between probable and improbable guides the research of all of us and the decisions we make about "what counts."

In doing so, she drew on two other (non-literary) texts. The first of these was Lisa Belkin's "Coincidence in an Age of Conspiracy," or "The Odds of That" (New York Times Magazine , August 11, 2002), which explores the unexpected connections, with no apparent causal relations, which "rattle and rivet" us, the surprising concurrences which we construct as meaningfully related. That we do so is no surprise, given the recent work of linguists. George Lakoff (who spoke, last fall, with the Language Group) argued in Philosophy in the Flesh that we are programmed to see patterns: we are pattern-seeking/pattern-making creatures who make smaller sets from large amounts of information and, conversely, infer larger structures from whatever limited information is available. This Thursday afternoon, the Language group will start up again for this semester by reading Ray Jackendoff's Patterns in the Mind, a Chomskian argument that we have an inherent impulse to find the simplest way to make sense of missing information. In the absence of information, we will generate some by filling in a pattern; in the absence of a story, we will make one up.

But when we make a pattern out of random events, when we turn a coincidence into a conspiracy, does our rational pattern-seeking become irrational? Belkin argues that, especially in age when paranoia runs rampant, we are discomforted by idea of random universe. Finding a reason or pattern where there is none makes it less frightening, because it makes it logical. In a "big enough" world, she suggests, one with a "large enough" denominator, patterns will emerge. Belkin makes the point that the Web has changed the scale of things, has given us the technical ability to gather bits and pieces of information. The Internet makes it easier to collect random noise--and then to find chance patterns in it; it has fed a generation of conspiracy theorists who see highly improbable patterns in large data sets.

The "better story" may very well be the one that accounts for most data (for example: evolution accounts for more of the details in world as we know it than does creationism; in psychoanalysis, one may well learn that the "better story" is one that takes into account one's family of origin, as well as current relationships). But in the era of the Internet, with enormous amounts of data available, how do we decide what is relevant, what random? How do we decide when the outlier is a key to a new story or pattern not yet seen?

In Once Upon a Number, Temple mathematician John Paulos says that the gap between stories/statistics is a synecdoche for the better-known gap between literary and scientific cultures which this series traced last year, and suggests that describing the world is a contest between the simplifiers (scientists, statisticians) and complicators (humanists, storytellers)--a.k.a. lumpers and splitters. In that simple divide, mathematicians are the radical thinkers, the ones getting to the root. But perhaps we are all up to the same thing, the same synecdochal act of substituting a part for the whole, a sample for whole population (why it was suggested last week that quantitative and qualitative science do not differ, in KIND, from one another). Scientists, social scientists, humanists and administrators are all looking for some way of "domesticating disorder," all searching for a good theory/short program that is capable of

  1. making sense of available data AND
  2. if not predicting, then generating some new knowledge.
Really, for all of us, the only phenomena of interest is that which is NOT random; all of us are negotiating the space between pattern and newness.

Paulos asks his readers to imagine prehistoric humans without any rudimentary idea of the typical--to imagine how, as situations reoccur, the notions of the average (mean, median) develop, along w/ their compliments: deviation, statistical variation, the far-out, the unconventional, the tail of a graph. . . . Like our ancestors, we are all statisticians, all make inferences based on tiny samples (first impressions of new colleagues, for instance). And then we all need details/background/back stories to evaluate the validity of stereotypes, statistics, the distillations of narratives that are first impressions. Along the way, we also notice what does not fit the pattern and learn that what is rare, unusual, exceptional, must still be attended to (the stripe in the grass may seldom be a tiger, but when it is: you have to move--and fast!) Experience generates--and so teaches us the notion of--probability, but just because some scenarios are more likely than others doesn't tell us which ones we should most attend to, which ones we should "count."

The week before, the intersection of narrative and number in the social sciences was described as enabling us to test the validity of an assertion; it was suggested that measurements (like polls) have predictive value. The suggestion this week is that, while storytelling involves ideas of alternative possibilities and open-endedness (and modernist aesthetics are far more inclusive than classical prescription were), our acceptance of extraneous, superfluous information has grown within bounds: there are still limits, Paulos claims, to how much irrelevancy we can tolerate. So far, for example, there is no very satisfying computer-generated fiction: hypertext novels "bore" because they "wander around," without linear progression. They are open-ended, "life-like," with too much digression, too much of what Paulos calls the "misleading richness of stories": they offer too many possibilities and not "enough" direction. The claim here, then, is that the best, the most precious, the most useful stories are those which BOTH

  1. reveal the unexpected, are counterintuitive, have a "surplus resonance"
  2. AND ALSO are predictive, have what (literary theorist) Mark Turner calls "parable-like correspondences."
Paulos observes that "novels aren't very novel" and that "models don't model"; more suggestively, perhaps, good novels have a model quality and good models have a novel quality. In all domains, we are trying to distill information, to find a pattern in it that has some meaning...and some space for play, for the unpredictable. Our hunger for stories is strong; our impulse is to make information MEAN something. But some perfect correlations mean nothing (consider, for instance, the pattern traced by Susan Sontag in Illness as Metaphor: various personality types have been associated, through history, with various diseases; those stories evaporate once a physical cause for the illness is identified.)

The presentation ended with a return to Life of Pi, an existential search for meaning which includes an impassioned defense of zoos for providing animals with relief from the compulsions and necessity of life in the wild. Pi believes firmly in the value of zoos, and in the value of religion, in making the world safer, in providing "the better story," one that incorporates "a trusting sense of presence and of ultimate purpose." For some of us, perhaps, the more satisfying story is one in which WE create the pattern with a purpose, believing that gives us more agency and more freedom. But the discussion opened, not with a consideration of whether, on the largest scale, meaning is given to us, or made, but rather by asking, at a slightly different level, how all participants understand the relation of statistics to stories in the work they do: is there an inverse, causal or no relation at all between number and narrative? How does each of us make meaning out of the information we gather?

There were a range of responses to these questions: scientists hunt for randomness in order to eliminate it, refining their techniques in order to identify the outlier that will open up new questions. Economists, who generate huge data sets, see every object as exerting a force on every other in a universe that is entirely deterministic; for them, nothing is random. Comparisons were made between disciplines in which data sets are given externally and and those in which change can be controlled; in every discipline, it is necessary to observe change in order to gauge our understanding of what has occurred. In the theater, there is a tension of opposites; a good story can actually change the data, rendering strange what we thought we knew. Is it our business to destroy old stories? Discussion ended with a description of the history of tetonics: a battle between those who wanted to believe that the earth was static and those who were looking for evidence of drift. There will be further discussion of "what counts" in geology next Monday @ noon, in a conversation led by Arlo Weil.

Return to Brown Bag Home Page


Home | Calendar | About | Getting Involved | Groups | Initiatives | Bryn Mawr Home | Serendip Home

Director: Liz McCormack -
emccorma@brynmawr.edu | Faculty Steering Committee | Secretary: Lisa Kolonay
© 1994- , by Center for Science in Society, Bryn Mawr College and Serendip

Last Modified: Tuesday, 07-Mar-2006 15:45:29 EST