The Anatomy of a Digital Object

1
The Anatomy of a Digital Object

• Matthew G. Kirschenbaum
• University of Maryland
• e(X)literature
• UCSB, April 2003

2
Lets talk about bibliography.

• Bibliography? Isnt that just making lists and
  stuff?
• Audience yawns and reaches for coffee. After
  all, its 900 AM. Plus, hes using PowerPoint.

3
Wrong.

• Actually, bibliography is the oldest and most
  sophisticated form of media studies we have (says
  Kirschenbaum).

4
Media studies ca. 1933 . . .

• Bibliography is the study of books as tangible
  objects. It examines the materials of which they
  are made and the manner in which those materials
  are put together. It traces their place and mode
  of origin, and the subsequent adventures that
  have befallen them. It is not concerned with
  their contents in a literary sense, but it is
  certainly concerned with the signs and symbols
  they contain (apart from their significance) for
  the manner in which these marks are written or
  impressed is a very relevant bibliographical
  fact. W. W. Greg

5
Properly speaking, there are three kinds of
bibliography

• Enumerative Bibliography
• Analytical or Physical Bibliography
• Descriptive Bibliography

6
Enumerative Bibliography

• This is the making lists branch of
  bibliography.
• The first enumerative bibliographer may have been
  Callimachus of Cyrene (ca. 310-240 B.C.), who
  compiled a 120-volume catalog of the library at
  Alexandria.

7
Analytical or Physical Bibliography

• Forensic examination of book-object for material
  evidence of its own production.
• Foundation of textual editing and literary
  criticism.
• For example, decoding the five different
  compositors who set the type for the first
  edition of Hawthornes novel Fanshawe.

8
Descriptive Bibliography

• Outgrowth of analytical bibliography.
• Rigorous, standardized, many would even want to
  say scientific description of books as physical
  objects.
• Fredson Bowers, Principles of Bibliographical
  Description (Princeton UP, 1949).

9
Quoth Bowers

• Many a literary critic has investigated the
  past ownership and mechanical condition of his
  second-hand automobile, or the pedigree and
  training of his dog, more thoroughly than he has
  looked into the qualifications of the text on
  which his critical theories rest.

10
OK, Im awake now, but . . .

• Whats all this got to do with electronic
  literature, much less e(X)literature?

11
Well, electronic literatures done a good job
with enumerative bibliography . . .
12
. . . but our use of analytical and descriptive
bibliography is not very far along at all.

• Kirschenbaum thinks this has something to do
  with the fact that until fairly recently, the
  widespread prejudice was that electronic objects
  were simply immaterial (presumably because you
  cant reach out and touch them). Kirschenbaum
  calls this the haptic fallacy.

13
Anyway, electronic objects exist in space and
time . . .

14
And they are to a certain extent self-documenting
. . .
15

• . . . but in the electronic literature community
  we lack a sophisticated vocabulary for expressing
  the synchronic and diachronic changes that inform
  the material make-up of digital objects.

16
Will the real afternoon please stand up?

• First edition of afternoon (1987, Mac) and third
  edition (1992, PC).

17
And bibliography can change all that?

• Yes, I think it can. And PAD is just the group
  to do it. So for the rest of my time, Im going
  to concentrate on descriptive bibliography, and
  work towards some principles of computational
  description.

18
In Bowers words, descriptive bibliography seeks
to . . .

• Furnish a detailed, analytical record of the
  physical characteristics of a book which would
  simultaneously serve as a trustworthy source of
  identification and as a medium to bring an absent
  book before a readers eyes (vii).
• Hmm, kind of sounds like metadata, doesnt it?

19
Put another way (still Bowers)

• The concern of the descriptive bibliographer is
  to examine every available copy of an edition of
  a book in order to describe in bibliographical
  terms the characteristics of an ideal copy of
  this edition, to distinguish between issues and
  variants of the edition, to explain and describe
  the printing and textual history of the edition,
  and finally to arrange it in a correct and
  logical relationship to other editions (6).

20
But bear in mind (with Bowers)

• The description and analysis contained in a
  bibliography is not intended to exist on a
  factual basis as an end in itself. Bibliography
  would be a limited science indeed if collection
  of external facts were its sole reason for
  existence. True bibliography is the bridge to
  textual, which is to say literary, criticism. The
  history of an authors book is an intimate part
  of the literary history of that author (8-9).

21
Bowers terminology for hand-printed books

• Edition
• Impression
• Issue
• State
• Ideal Copy

22
My very provisional terminology for describing
digital objects

• Release
• Version
• Layer
• Object
• State
• Instance
• Copy

23
Consider as an example David Blairs hypermedia
WAXWEB . . .

• Extruded from his 1991 film, Wax, or the
  Discovery of Television Among the Bees (90 min).
• In 1993, Wax became the first feature-length film
  to be distributed over the Internets
  high-bandwidth M-Bone.

24
WAXWEB (1994-2003) http//www.waxweb.org

• Storyspace
• MOO
• 2000 HTML nodes and 25,000 links
• 4800 JPEG image files
• 560 MPEG video files
• 2200 WAV audio files
• 300 VRML renderings
• Final ? CD-ROM

25
Layer

• A layer comprises all elements of an electronic
  work that are both computationally compatible and
  functionally integrated. When one or more
  elements are ported such that the whole is no
  longer computationally compatible or functionally
  integrated, then a new layer is created.

26
Version

• Layers can be distinguished by assigning them
  versions.
• David Blair versions WAXWEB as follows 1.0 was
  the first web version, with the html constructed
  by the MOO, using text, picture, movie 2.0 added
  the old vrml there are two versions up now... the
  output of the MOO, which I started to work with
  in 95 before finally just tossing it out,
  putatively called 3.0 alpha. The finished version
  is just Waxweb.

27
Release

• Different releases of the work are all
  computationally compatible with one another, but
  they are not functionally integrated.

28
Object

• A generic identifier for some discrete digital
  entity, such as a file (and a file can itself be
  defined as a named collection of data that
  persists over time).

29
State

• A state refers to the computational composition
  of an object in some particular data format.

30
Instance

• An instance refers to a particular object in a
  particular state as presented in a particular
  software environment.

31
Copy

• A copy refers to one precise and particular
  instance of one particular state of an object.
  (If I download an image in JPEG format from the
  Web, the copy my browser caches is ontologically
  distinct (even though it is computationally
  identical) to the copy on the server.)

32
Putting it all together . . .

• In the slide that follows, one single digital
  object (this frame of video from the original Wax
  film) is presented in two separate states
  (RealVideo and JPEG) and four separate instances,
  all distributed across three layers and versions
  of the work. (Note that this display does not
  necessarily include all instances or states of
  the object.)

33
(No Transcript)
34
Very nice, but isnt the library science already
there? METS and all that?

• No, I dont think so. As I understand them, METS
  (and other things like it) do not help with the
  problem of expressing relationships between
  multiple versions and states of an individual
  work. This is whats at the heart of the
  bibliographical enterprise, and its a
  perspective thats essential for PAD.
• Oh, and about that word work . . .

35
Exactly what is the work here? Either one? Both
together?

36
The other CVS . . .

• Which is why I think we can also learn a lot
  from developers tools, such as the open source
  Concurrent Versions System (CVS)
• A version control system maintains an organized
  set of all the versions of files that are made
  over time. Version control systems allow people
  to go back to previous revisions of individual
  files, and to compare any two revisions to view
  the changes between them. In this way, version
  control keeps a historically accurate and
  retrievable log of a file's revisions.
• http//www.cvshome.org

37
Like Howard, Bowers understood the importance of
standards . . .

• It is necessary in the extreme for
  bibliographers (like chemists or mathematicians)
  to agree on a standard and uniform system of
  notation, to be followed in all its details and
  to be maintained rigidly in their works. Personal
  prejudice against one or other device should not
  be permitted to interfere with the adoption of
  all details of a standard system (24).

38
Recommendations for PAD

• Recognize the value of the critical knowledge
  analytical and descriptive bibliography afford
  for traditional literary studies
• Help foster analytical and descriptive
  bibliographical practices for the electronic
  literature community by developing and
  disseminating standardized principles of
  computational description (not necessarily mine
  here!) for the digital objects that matter to us
• Tightly couple any such principles of
  computational description to our technical
  metadata and encoding practices.

39
Shameless self-promotion

• Kirschenbaum, Editing the Interface Textual
  Studies and First-Generation Electronic Objects.
  TEXT 14 (2002) 15-51.
• Kirschenbaum, Mechanisms New Media and the New
  Textuality (forthcoming, MIT Press).