do my reading home work

6

{ Narrative and Database Spatial History and the Limits of Symbiosis }

Ab, the power of metaphors-especially when a metaphor

propagates with viral intensity through a discursive realm.

At issue is Lev Manovich's characterization of narrative and

database in The Language of New Media (2002) as "natural en­

emies" (228), a phrase Ed Folsom rehearses in his discussion

of the Walt Whitman Archive (2007). The metaphor resonates

throughout Folsom's essay in phrases such as "the attack of

database on narrative" (1574), culminating in his figure of

database's spread as a viral pandemic that "threatens to dis­

place narrative, to infect and deconstruct narrative endlessly,

to make it retreat behind the database or dissolve back into

it" (1577). In this imagined combat between narrative and

database, database plays the role of the Ebola virus, whose

voracious spread narrative is helpless to resist. The inevitable

triumph of database over narrative had already been forecast

in Manovich's observation that "databases occupy a signifi­

cant, if not the largest, territory of the new media landscape"

(2002:228). Indeed , so powerful and pervasive are databases

175

176 Chapter 6

in Manovich's view that he finds it "surprising" narratives continue to eX I\!

at all in new media (228). In Manovich's view, the most likely explanation, ., narrative's persistence is the tendency in new media to want to tell a St(Jl ,

a regressive tendency he identifies with cinema. Even this, he suggests, is III

the process of being eradicated by experimental filmmakers such as Pet'·, Greenaway (237-39).

Rather than being natural enemies, narrative and database are more ell

propriately seen as natural symbionts. Symbionts are organisms of differclIl species that have a mutually beneficial relation. For example, a bird pic k

off bugs that torment a water buffalo, making the beast's existence mOl .

comfortable; the water buffalo provides the bird with tasty meals. BecaU M

database can construct relational juxtapOSitions but is helpless to interprl'1

or explain them, it needs narrative to make its results meaningful. Narrativ( '

for its part, needs database in the computationally intensive culture of t il l

new millennium to enhance its cultural authori ty and test the generality 01

its insights. If narrative often dissolves into database, as Folsom suggest ~, database catalyzes and indeed demands narrative's reappearance as soon ,I '

meaning and interpretation are required (as the discussion at the conch I

sion of chapter 5 illustrates). The dance (or as I prefer to call it, the complex

ecology) of narrative and database has its origins in the different ontologie~, purposes, and histories of these two cultural forms. To understand more pre

cisely the interactions betwee n these two cultural forms, let us turn now to consider their characteristics.

The Different Worldviews of Narrative and Database

As Manovich (2002) observes, database parses the world from the viewpoint

of large-scale data collection and management. For the late twentieth and

early twenty-first centuries, thi s means seeing the world in terms that the

computer can understand. By far the most pervasive form of database is the

relational , which has almost entirel y replaced the older hierarchical, tree,

and network models and continues to hold sway against the newer object­

oriented model s. In a relational database, data are par sed into tables consist­

ing of rows and columns, where the column heading, or attribute, indicates

some aspect of the table's topic . Ideally, each table contains data pertaining

to only one "theme" or central data concept. One table, for example, might

contain data about authors, where the attributes might be last name, first

name, birth date, death date, book titles, etc.; another might have publish­

ers' data, also parsed according to attributes; another, books. Relations are

Narrati ve and Database 177

constructed among data elements in the tables according to set-theoretic

ope rations, such as "insert," "delete," "select," and especially "join ," the com­

mand that allows data from different tables to be combined. Common ele­

ments allow correlations to be made between tables ; for example, Whitman

would appear in the authors table as an author and in the books table corre­

lated with the titles he published; the publishers table would correlate with

the books table through common elements, and through these elements

back to the authors table. Working through these kinds of correlations, set­

theoretic operations allow new tables to be constructed from existing ones.

Different interfaces can be designed according to the particular needs of

users. Behind the interface, whatever its form, is a database-management

system that employs set-theoretic notation to query the database and manip­

ulate the response through SQL (SQL is commonly expanded as structured

query language) .

The great strength of database, of course, is the ability to order vast

arrays of data and make them available for different kinds of queries. Two

fundamental aspects typically characterize relational databases. One, indi­

cated above, is their construction of relations between different attributes

and tables. The other is a well-constructed database's self-containment, or,

as the technical literature calls it, self-desc ription . A database is said to be

self-describing because its user does not need to go outside the database to

see what it contains. As David Kroenke and David Auer put it in Database

Concepts (2007), the "structure of the database is contained within the da­

tabase itself" (13), so that the database's contents can be determined just by

looking inside it. Its self-describing nature is apparent in SQL commands.

For the database mentioned above containing information about authors,

books, and publishers, for example, a typical SQL command might take

the generalized form "S ELECT AUTHOR .AuthorName, BOOK. BookTitle,

BOOK.BookDate, BOOK. Publisher, PUBLISHER. Location ," where the table

names are given in all capitals (as are SQL commands) and the data record s

are categorized according to the attributes, with a period separating table

name from attribute . The database's self-description is crucial to being able

to query it with set-theoretic operations, which require a formally closed

logical sys tem upon which to operate. This is also why databases fit so we ll

with computers. Like databases, computers work according to the opera­

tions of formal logiC as defined by the logiC gates that underlie all executable

commands.

The self-describing nature of database provides a strong contrast with

narrati ve, which always contains more than indicated by a table of contents

178 Chapter 6

or a list of chapter con tents . Databases can, of course, also extend out\\ "

when they are linked and queried as a network-for example, in data 11 11'

ing and text mining techniques-but they do not lose the formal proP(' 11 j.

of closure that make them self-describing artifacts. Nevertheless, the 11'11

nologies of linking databases have proven to be remarkably powerful. oil

the relations revealed by set-theoretic operations on networks of Iill ~

databases can have stunning implications. For exa mple, data- and I n l

mining techniques allowed epidemiologist re searchers Don Swansoll .11 '

N. R. Smalheiser (1994, 1997) to hypothesize causes for rare diseases II. hitherto had re sisted analysis because they occurred infrequently at wid d

separated locales. Even in this case, however, the meaning of the relaliOlI

posited by the database remains outside the realm of data techniques. WI L'

it means that Whitman, say, used a certain word 298 times in Leaves ofG rn while using another wo rd only 3 times requires interpretation-and illto l

pretation , almost inevitably, invokes narrative to achieve dramatic imp,II I

and significance. Many data analysts and statisticians are keenly awar(' 1.1

this symbiosis between narrative and data. John W. Tukey, in his classic te xl book Exploratory Data Analysis (1997), for example, explains that the data 01 11 alyst "has to learn ... how to expose himself to what his data are willing-ol

even anxious-to tell him" (21), following up the lesson by later asking t ill

student/reader "what story did each [dataset] tell " (101).

Database and narrative, their interdependence notwithstanding, rem,li ll

different species, like bird and water buffalo. Databases must parse infol

mation according to the logical categories that order and list the differe lll data elements. Indeterminate data-data that are not known or otherwi~(

elude the boundaries of the preestablished categories-must either be reI'

resented through a null value or not be repre se nted at all. Even though SOIlIl'

relational databases allow for the entry of null values, such values work ill

set-theoretic operations as a contaminant, since any operation containing .1

null value wi ll give a null value as its result, as multiplying a number by zero

yields zero. Null values can thus quickly sp read through a database , render

ing everything they touch indeterminate. Moreover, database operations say

nothing about how data are to be collected or whi ch data should qualify for

collection, nor does it indicate how the data should be parsed and catego ­

rized. Constructing a database always involves assumptions about how to

set up the relevant categories, which in turn may have ideological implica­

tions, as Geoffrey C. Bowker and Susan Leigh Star have observed (2000) .

In addition, such decisions greatly influence the viability, usefulness, and

operational integrity of databases. Thomas Connolly and Carolyn Begg in

179Narrative and Database

/)atabase Systems (2002) estimate that for corporate database software de­

velopment projects, 80 to 90 percent do not meet their performance goals,

80 percent are delivered late and over budget, and 40 percent fail or are

.d,andoned (270). Anticipating such problems, database textbooks routinely

ldvise students to keep the actual design of the database a closely guarded

~ l: cret, confining discussions with the paying client to what the interface

·; hould look like and how it should work. In addition to not "confusing" the

I lient, such a strategy obscures suboptimal performance. The indeterminacy that databases find difficult to tolerate marks another

way in which narrati ve differs from database. Narratives gesture toward the

Inexplicable, the un speakable, the ineffable, whereas databases rely on enu­

Ineration, requiring explicit articulation of attributes and data values. I While

I he concatenation of relations might be suggestive, as the epidemiology ex­

' llnple illustrates , databases in themselves can only speak that which can

explicitly be spoken. Narratives, by co ntrast, invite in the unknown, taking

lIS to the brink signified by Henry James's figure in the carpet, Kurtz's "The

horror, the horror;' Gatsby's green light at pier's end, Kerouac's beatitude,

Pynchon's crying of Lot 49. Alan Liu, discuss ing the possibilities for this kind

of gesture in a postindustrial information-intensive era, connects it with "the

ethos of the unknown" and finds it expressed in selected artworks as a "data

pour," an overflowing uncontainable excess that he links with transcendence

(2008b:81). Whereas database reflects the computer's ontology and operates with op­

timum effiCiency in set-theoretic operations based on formal logic, narrative

is an ancient linguistic technology almost as old as the human species. As

such, narrative modes are deeply influenced by the evolutionary needs of hu­

mans negotiating unpredi ctable three-dimensional environments populated

by dive rse autonomous agents. As Mark Turner has argued in The Literary

Mind: Th e Origins of Thought and Language (1998) , stories are central in the development of human cognition. Whereas database allows large amounts

of information to be sorted, cataloged, and queried, narrative models how

minds think and how the world works, projects in which temporality and

inference play rich and complex roles . Extending Paul Ricoeur's (1990) work

on temporality and Gerard Genette's (1983) on narrative modalities , Mieke

Bal (1998) analyzes narrative as requiring, at a minimum, an actor and nar­

rator and consisting of three distinct leve ls , text, story, and fabula, each with

its own chronology (6). To this we can add Brian Richardson's emphasis in

Unlikely Stories: Causality and the Nature of Modern Narrative (1997) on cau­

sality and inference in narrative. 2

181 180 Chapter 6

Why should narrative emphasize these aspects rather than others? B Olli

to the linear sequentiality of language, narrative complicates it through I,·"

poral enfoldings of story (or, as Genette prefers to call it, discourse) ,II

fabula , reflecting the complexities of acting when knowledge is incom plr"

and the true situation may be revealed in an order different from thl' 01

logical reconstruction requires. Narrator ·and actor inscribe the situatioll " I

subject constantly negotiating with agents who have their own agendas ,110

desires, while causality and inference represent the reasoning required I

suture different temporal trajectories, motives, and actions into an expl <1I1

tory frame. These structures imply that the primary purpose of narrati vi' I

to search for meaning, making narrative an essential technology for hum ,1I1

who can arguably be defined as meaning-seeking animals.

Bound to the linear order of language through syntax, narrative is a t l'lII

poral technology, as the complex syncopations betwee n story and fab ul.

demonstrate. The order in which events are narrated is crucial, and tern!"

ral considerations are centrally important in narratology, as Ricoeur's Wil l

(1990), among others, illustrates. Data sets and databases , by contrast, leu r! themselves readily to spatial displays, from the two-dimensional tables typ, cal of relational databases to the more complex n-dimensional arrays alld

spatial forms that statisticians and data analysts use to understand the S ill ries that data tell.

Manovich touches on this contrast when he argues that for narrative, ti ll

syntagmatic order of linear unfolding is actually present on the page, wIll I!

the paradigmatic possibilities of alternative word choices are only virtuall ~

present. For databases , the reverse is true: the paradigmatic possibilities ;'\1,

actually present in the columns and rows, while the syntagmatic progress (11 choices concatenated into linear sequences by SQL commands is only virt.u

ally present (Manovich 2002 :228) This influential formulation , despite il

popularity, is seriously flawed, as Allen Bye Riddell points out (pers. COIJI

munication April 7, 2008). Recall that in semiotics, the alternative choic~' ~ of the paradigm interact with the inscribed word precisely because they an'

absent from the page, although active in the reader's imagination as a set nl

related terms . Contrary to Manovich's claim, databases are not paradigmatit

in this sense. In a relational database configured as columns and rows, tht.:

data values of a row constitute the attributes of a given record, while tht'

columns represent the kinds of attribute itemized for many different records.

In neither the rows nor columns does a logic of substitution obtain; the terms

are not synonyms or sets of alternative terms but different data values.

Narrati ve and Database

These objections notwithstanding, Manovich's formulation contains a

!rnel of truth. Search queries, as we have seen, allow different kinds of

,lttributes and data values to be concatenated . The concatenated values, al­

tl lough not syntagmatic in the usual sense, can be seen as a (virtual) tempo­

f ,II process in which it is not syntax but the SQL command that determines

Ibe concatenation's order. Thus, although Manovich's formulation is not

l~c hnically correct, it captures the overall sense that the temporal ordering

n ucial for narrative is only Virtually present in the database, whereas spatial

lisplay is explicit. I would add to this the observation that time and space,

I he qualiti es Kant identified as intrinsic to human sensory-cognitive facul­

t ies, inevitably coexist. While one may momentarily be dominant in a given

~ i luation, the other is always implicit, the natural symbiont whose existence

IS inextricably entwined with its partner. It should be no surprise, then, that

Ila rrative and database align themselve s with these partners, or that they too

:xist in symbiosis with one another.

Given this entwinement, is it plausible to imagine, as Manovich and

Folsom imply at various points, that database will replace narrative to the

;xtent that narrative fades from the scene? A wealth of evidence points in

I he other direction: narrative is essential to the human Iifeworld. Jerome

J) runer, in his book significantly entitled Acts of Meaning : Four Lectures on

Mind and Culture (1992), cites studies indicating that mothers tell their chil­ dren some form of narrative several times each hour to guide their actions

,md explain how the world works (81-84). We take narrative in with moth­ l!r's milk and practice it many times every day of our lives-and not only

in high culture forms such as print novels . Newspapers, gossip, math story

problems, television dramas, radio talk shows, and a host of other communi­

cations are permeated by narrative. Wherever one looks, narratives surface,

as ubiquitous in everyday culture as dust mites.

What has changed in the informative-intensive milieu of the twenty-first

century is the position narrative occupies in the culture. Whereas in the clas­

sical Greek and Roman era narrative was accepted as an adequate explanation

for large-scale events-the creation of the world, the dynamics of wind and

fire , earth and water-global explanations are now typically rooted in data

analysis. If we want to understand the effects of global warming or whether the economy is headed for a recovery, we likely would not be content with

anecdotes about buttercups appearing earlier than usual in the backyard or

Aunt Agnes's son finally finding a job. Data, along with the databases that

collect, parse, and store them and the database-management systems that

182 Chapter 6

concatenate and query them are essential for understanding large-scal e ph,

nomena. At the global level , databases are essential. Even there, howl''',

narrative enters in the interpretation of the relations revealed by datah.,

queries. When Ben Bernanke testifies before Congress, he typically does . It

recount data alone. Rather, he tells a story, and it is the story, rather t h.II the data by themselves, that propagates through the news media, bec;)l1 ·

it encapsulates in easily comprehensible form the meaning exposed by d.ll

collection and analysis.

In contrast to global dynamics, narrative at the local leve l remains pen ,I

sive, albeit more and more infused by data . In the face of the overwhelmil l

quantities of data that database-management systems now put at our fingl"l

tips, no one narrative is likely to establish dominance as the explanation, tn, the interpretive possibilities proliferate as databases increase. In this respc( I

the advent of the Internet, espeCially the World Wide Web, has been decisi vl ·

N ever before in the history of the human species has so much information bel "I

so easily available to so many. The constant expansion of new data aCCOUlIl

for an important advantage that relational databases have over narratives, fOl new data elements can be added to existing databases without disrupting th <.:ll

order. Unlike older computer database models in which memory pointci

were attached directly to data elements, relational databases allow the ordl'l

of the rows and columns to vary without affecting the system's ability to local.·

the proper elements in memory. This flexibility allows databases to expand

without limitation (subject, of course, to the amount of memory storage al

located to the database). Narrative in thi s respect operates quite differently.

Sensitively dependent on the order in which information is revealed, narra

tive cannot in general accommodate the addition of new elements without.

in effect, telling a different story. Databases tend toward inclusivity, narratives

toward selectivity. Harry Mathews explores this property of narrative in Th('

Journalist: A Novel (1994), where the unnamed protagonist, intent on making a list of everything that happens in his life, thinks of more and more items, with

the predictable result that the list quickly tends toward chaos as the interpola­

tions proliferate. The story of this character's life cannot stabilize, because the

information that co nstitutes it continues to grow exponentially, until both list

and subject collapse.

That novels like Th e Journalist should be written in the late twentieth cen­

tury speaks to the challenges that database pose s to narrative in the age of

information. No doubt phenomena like this explain why Manovich would

characterize database and narrative as "natural enemies" and why thought­

ful scholars would propagate the metaphor. Nevertheless, the same dynamic

183Narrative and Database

li so explains why the expansion of database is a powerful force constantly

pawning new narratives. The flip side of narrative's inability to tell the

lory is the proliferation of narratives as they transform to accommodate

pe w data and mutate to probe what lie s beyond the exponentially expanding

w fosphere. No longer Singular, narratives remain the necessary others to

rl dla base's ontology, the perspectives that inves t the formal logic of database

II pe rations with human meanings and gesture toward the unknown hover­

Ing beyond the brink of what can be classified and enumerated.

Spatial History: A Field in Transition

Among the disciplines that routinely rely on narrative is qualitative history.

Traditionally seen as centering on "change through time," history has re­

cently been going through a "spatial turn," in which databases, GIS, and GPS

t I.!chnologies have provided an array of tools for rethinking and re-representing

I he problematics of history in spatial terms. Accordingly, historian s have

moved into alliance with geographers in new ways. This movement has its

own tensions, howeve r, for geographers often rely on discursive methods of

{; xplanation, whereas historians are turning more and more to databases .

The tension can be seen in the work of Doreen Massey (2005) compared

to such database history projects as those of the Stanford Spatial History

r'roject. Massey, an important contemporary geographer, has a dream: to

replace the idea of space as an inert container w ith a conceptualization of it

:\s an emergent property constructed through interrelations and containing

diverse simultaneous trajectories. In general, she advocates a view of space

as "lively" rather than "dead." Above all, she does not wa nt space conceptual­

ized as a pre-give n Cartesian manifold that extends in every direction, infi­

nitely subdividable and homogeneous through multiple scale values. Yet this

seems to be preCisely w hat Google Earth, GIS, and GPS digital technologies

offer, with the illusion of present-tense rendering, zoom functions through

multiple scale levels , and seamless transitions between map, satellite, and

street views. Given the power, pervasiveness, and allure of such visualiza­

tions, can Massey's dream gain traction in contemporary representations?

More generally, does the "spatial turn" in digital history projects imply that

the traditional focus on time for historians has now been transformed into

spatializations that take the Cartesian grid as the basis for hi storical repre­

se ntations? Answering these questions requires a deeper understanding of

what Massey's dream implies, along with its roots in prior wo rk by geogra­

phers and an analysis of how GIS and GPS technologies are actually being