A Few Types of Literary Compression

“And I said to Mabel, I said, ‘computational aesthetics, super-short. Jürgen Schmidhuber’s Theory Jürgen Schmidhuber, an AI theorist and theoretical computer scientist, has proposed a computational account of aesthetic judgments. In his view, a stimulus is judged to be beautiful or attractive by a subject T to the extent that the stimulus is compressible for T. Schmidhuber’s notion of compressibility is taken from algorithmic information theory, but concerns actual rather than ideal compression: it refers to the actual # of bits in T’s mental representation of the stimulus, bounded and fallible as T may be. Beholden to the limitations of T’s computational resources, two kinds of stimuli should be the most compressible: stimuli with evident internal structure (e.g. fractals or a chessboard), and stimuli with noticeable similarities to stimuli already stored in T’s history (e.g. English words or a the sight of a friend’s face). Experimental psychology supports both a preference for stimuli with internal patterns and a preference for stimuli with a similarity to past stimuli.”

— Peli Grietzer, Amerikkkkka
(website)

This is a scratchpost outlining some of the primary mechanisms of literary compression. (Background for this concept is provided in “Art as Engineering.”) Compression in Schmidhuber’s formulation is achieved primarily through pattern matching, identification, and structure, but it can be more generally understood as the practice of “packing in” — of increasing a work’s ratio of information to bit, effect to resource, benefit to cost. 

Though to state it as such borders on tautology, understanding compressive acts as being, therefore, either additive or subtractive is essential. That is to say, compression must involve either an increase in information conveyed (at a proportionally lower cost in bits) or else a decrease in the number of bits (at a proportionally lower cost in information). The specific types of literary compression theorized below fall somewhere along the spectrum of additive and subtractive acts; they are described along with some acknowledgment of what is potentially lost through compression.

  1. Signification and symbols: Signifiers and symbols stand-in for complicated systems, ideas, and things; they are a sort of conceptual and mimetic shorthand. This category includes names (concept-handles, crystallized patterns), visual symbols, and motifs. All language can be understood as compression: 7*1027 atoms — so many elements in flesh and blood and abstract consciousness — become a “human.” When we iterate this compressive process, we increase the informational density of a text beyond the ratios of everyday speech. Of course, for information to be properly conveyed, audience and author must share a vocabulary, must have a general consensus about what any given word means. When this type of literary compression is executed poorly, or else exposed to an inappropriate audience, we call it “jargon.”
  2. Signal-to-noise: The elimination of details which contribute minimally either to message (communicative/informational intent) or to effect (emotional/intellectual cognitive stimulation). Mysterious trade-offs inevitably occur: we often prefer buried artistic messages to overt ones, and compositions which are too on-the-nose suffer in quality (in fact, are sometimes decried as not being art at all). This is perhaps because we are hardwired to appreciate puzzles, and because a general obscurity of message allows a range of conflicting interpretations, stimulating in turn a cultural-philosophical conversation.[1] 
  3. Double-duty/Multitasking: Using compositional elements which achieve multiple effects (or else convey many pieces of information) at once.[2] Often, a multitasking, “Jack-of-all-trades” element executes its individual tasks less well than would a devoted, single-purpose element. Nevertheless, its work’s overarching economy is improved.
  4. Capitalizing on shared knowledge bases: One step beyond signification and symbols. More than merely sharing the name of a thing, and relying on the audience to understand relationships between referent and reference, this compressive technique eliminates (vs. condensing) information. It operates off the assumption that certain information is implied by a work’s artistic, cultural, or historic context; a remark in a contemporary novel about the “melancholy of September 12, 2001” takes for granted that its reader is familiar with the World Trade Center attacks, and may refrain from mentioning them entirely. In technique #1 (signs and symbols), when an author writes that a character has been diagnosed with cancer, s/he trusts we are familiar with the medical phenomenon which “cancer” refers to. In technique #4, however, the author might describe said character vomiting in the bathroom, and assume we readers have the existing knowledge base necessary to identify chemotherapy as the culprit. All texts also operate off the shared knowledge base that is “information thus far gleaned by the reader from inside the text.”

Endnotes:

[1] The obscurity → multiplicity of interpretation mechanism is an entirely different one than the Iserian model of indeterminacy explored in “Generic Fit”:

We have seen that… the impressions that arise as a result of [the reading] process will vary from individual to individual, but only within the limits imposed by the written as opposed to the unwritten text. In the same way, two people gazing at the night sky may both be looking at the same collection of stars, but one will see the image of a plough, and the other will make out a dipper. The “stars” in a literary text are fixed; the lines that join them are variable.

[2] Whether “information” and “effect” are meaningfully different concepts is unclear.