Ulysses, Wilde, and a Theory of Literary Compression

“He looked at the cattle, blurred in silver heat. Silvered powdered olive trees. Quiet long days: pruning ripening. Olives are packed in jars, eh? I have a few left from Andrews…

A cloud began to cover the sun wholly slowly wholly. Grey. Far. No, not like that. A barren land, bare waste. Vulcanic lake, the dead sea… Brimstone they called it raining down: the cities of the plain.”

Ulysses, 4.200-221

An introduction to this text can be found here.

The mobilization of Ulysses and Earnest is purposefully audacious and inevitably missteps. The overarching tone, and parts of the analysis, I would characterize as understandably wrong.


Ideas of optimization and economy are often seen as incompatible with art, a discipline historically resistant to attempts at quantification.[1] Critic Haley Thurston, citing the work of Jürgen Schmidhuber, argues otherwise: that art is frequently successful on the basis of its ability to compress reality in an economic fashion. If reality is incomprehensibly complex, and the human brain a “hard drive with limited amounts of space,” then art is a technology which condenses reality, eliminates noise, and orders chaos via pattern identification. We appreciate compressive art because it eases our storage burden, and because it is through art’s patterns — among others — that we can best interpret, navigate,[2] and learn about the world around us.

There is precedent for this model of compressive learning. Schmidhuber writes:

A long time ago, [Jean] Piaget already explained the explorative learning behavior of children through his concepts of assimilation (new inputs are embedded in old
schemas—this may be viewed as a type of compression) and accommodation (adapting an old schema to a new input—this may be viewed as a type of compression improvement)…

And a natural extension of this theory follows that, if art is effective at shaping and updating our worldview, then compression is the primary mechanism by which it occurs:

Good observer-dependent art deepens the observer’s insights about this world or possible worlds, unveiling previously unknown regularities in compressible data, connecting previously disconnected patterns in an initially surprising way…  (Driven by Compression Progress, 9)

Patterns, however, are only one type of artistic compression, acting in the service of of a single scarce resource (cognitive storage). It is equally true that words are not just read but generated and recorded; in the case of oral tradition they require memorization. Literary optimization would likely take place for all these processes and scarce resources; our cultural assessment of it should take, in theory, these optimizations unconsciously and vestigially into account.  We can apply Thurston and Schmidhuber’s compressive concept to literature but then take it further still. Are not the pages upon which a book is printed, and the time which readers spend absorbing them, also scarce resources which successful texts would theoretically optimize for?

Approaching literary value in terms of utility or benefit can initially appear coldblooded, but it holds up if we stretch the definition of “utility” and “benefit” to their broadest boundaries — the utility of interestingness and intellectual engagement; the benefits of emotional rendering, personal recognition, sensory awe, or moral examination. A thing is valuable in that it adds value human life, and there can be no doubt that literature’s essence is in just that. A work’s degree of value, then, is the degree of economy between benefit and cost, between utility gained and scarce resource spent. A novel which does many beneficial things at once, which has a desirable ratio of cognitive stimulus (interestingness, emotional engagement, visceral or sensory prompting, etc) to materials used in its creation, or time spent in its consumption by the reader, has a naturally high value on the basis of its economy.

If economy is an ends, then compression is its means. This is a compression not just in the service of “capturing reality” — we can move beyond this definition, and reclassify it as subcategory — but as a general compacting of many cognitive stimuli and bits of information so as to maximize the economy between utility gained and resource spent. When it is done well, there is an unmistakable intensity which surges through the prose.

Let us for the moment focus only the cost of a text’s reproduction — the expense per word and page in man hours and material — as a primary denominator for judging the relative value of a literary work:


Such a system of aesthetic judgment makes sense given literature’s oral, scribe, and early-printing histories, when textual creation and preservation required either Herculean efforts of memorization or else costly and labor-intensive processes of production. Language of artistic economy is most common today in film, where costs of production are so high, and we can envision a much more affordable film tradition centuries hence still retaining vestigial structures of critical assessment and perceived value. Perhaps likewise should we perceive literature.

Moving away from a concept of compression as reality condensation is especially important here, in responding to Wilde’s Earnest and Joyce’s Ulysses, since both works focus primarily on creating “effect” versus “representation”; they are each “first and foremost a structure for eliciting responses and thereby engaging its readers.” We can refer to “effect,” from here on out, as cognitive stimulation in all its forms, be it emotional, intellectual, moral, sensory, or philosophical.


We would do well here to clarify what an economical, compressed work of literature looks like or means. It is easily confused as synonymous to economical storytelling, but the latter is only a subset of the former. Well-compressed literature at its most rudimentary is merely doing many somethings simultaneously or within the confines of minimal material. We might theorize that there are, broadly, story-driven and idea-driven novels; Ulysses is the latter. It uses narrative as canvas, landscape, and bedrock — almost as medium or material itself — upon which to compile ideas and explorations be they stylistic, psychological, linguistic, or moral etc. Narrative is the very terrain upon which Joyce builds vertically as if constructing a cityscape — he is not attempting to pack as much landscape, or narrative event, into Ulysses’ pages as possible, but to build upwards, densely, to maximize cubic footage per narrative square foot. This analogy is vital since it resolves the intuitive dissonance between Ulysses’ monumental size and its description as a compressed work.

As in building a city, some terrain will inevitably be better adapted, more able to handle the load of compressed, compact, compiling, than other terrain: Manhattan schist is the sturdy bedrock which supports the borough’s upward development, but where the schist disappears underground briefly, circa Washington Square, suddenly the buildings cannot be packed as tall; the landscape will simply not support it.[3]

To illustrate the distinction between economical storytelling and compressed literature, it’s worth comparing the opening pages of “Calypso” to a summary from Stuart Gilbert’s James Joyce’s Ulysses:

With this episode Mr. Bloom’s day begins: June 16, 1904; this date is esteemed, I am told, by certain advocates of a reformed calendar, a holyday styled Bloomsday. It is 8 a.m. Within the residence of Mr Bloom, 7 Eccles Street, there is still cool twilight but, outside, the streets are already warming up, and there is a hint of thunder in the air. As Mr Bloom moves softly about the basement kitchen… kidneys are “in his mind,” for he eats with relish “the inner organs of beasts and fowls.” The cat requests and receives milk on a saucer. The cat and Mr Bloom are on excellent terms.

This example does not quite replicate economical storytelling, since its compression only prioritizes narrative event and setting, lacking all the layered stimuli for anticipation, suspense, empathy, etc which are to be found in good storytelling. Nevertheless, we can ignore these features in order to see what is lost in the process Gilbert’s narrative-prioritizing compression, for information and stimulus alike are missing indeed. To say the cat “requests and receives” milk from Mr. Bloom misses the way the ambiguous female pronouns, which in the original text alternate referent between Molly Bloom and the cat, create the sensation in the reader of a general feminine presence to which Leopold Bloom is subservient (reinforced by the parallels between the mkgnao-ing cat and the commands of Bloom’s wife). The cat’s relationship with Leopold is described as being on “excellent terms,” and yet, the relationship which is illustrated in the chapter opening goes beyond, is more ambiguous and complex, than can be summarized by “excellent.” If art achieves the general through the instantiation of the particular, then Joyce is here capturing through illustration a complex psychology between pet and human (for “master” is not quite the right word here either given Bloom’s character), a psychology arguably ineffable through any other form. It is in part this dynamic — as well as the characterization of Bloom through his interaction with his pet — that Joyce is prioritizing, is compressing for — not the narrative plotting and advancement of events as the synopsisist Gilbert is.


At the most broad and general level, we might say that there are two types of compression: pruning and packing; subtractive or eliminative versus additive compression. Though density and economy are, by some definitions, largely synonymous, the connotations of the two are very different. The former implies condensation into heaviness, is thus representative of an additive sort of compression, in which layers are compacted by enormous pressure. The latter, inversely, implies a certain lightness, a lack of burden, a result of being freed from extraneous and bulky information which might otherwise weigh it down.

Typically the compressive process works with both types of compression and in strict order: first, by clearing out and eliminating information through signal-to-noise [SNR] optimization, through symbol, signification, or metaphor (naming and analogy), and through pattern recognition. Then, these many signals are compacted together through additive to create density, and in combination create economical literature (“economical” here in the denoted sense of efficiency and optimization as opposed to the connoted sense of lightness).

A type of eliminative compression, symbolic compression, exists at the very core of language and thus literature; the novel in any conceivable permutation cannot exist without it. Complex objects and subjects, composed of billions or trillions of particles, are grouped together and given names, reduced by symbols which stand in for their indescribably complex entireties and thus allow us to discuss them. Next comes patterns, actions, and interactions; the process is similar to the deflate algorithm’s compression of a .zip file. Wilde’s “Bunburying” is an advanced example of deflate-style symbolic compression. First the concept of Bunburying must be described, but from then on out, once the reader understands the concept behind the referent, it can merely be mentioned in passing — an entire phenomenon, behavior, and way of life has been compressed economically into a single word. At the most basic level of The Importance of Being Earnest even being able to achieve “artistry” or “literary value,” the reader must be spared the indescribably dull (unstimulating) experience of having the phenomenon described in an abstract entirety each time it is brought up — since such a practice would dilute the ratio of cognitive stimulation per minute of play (or in the case of a text, per word/sentence/paragraph/page of material). This essay engages in symbolic compression — not just in the obvious sense of using language but by describing phenomena and patterns, naming them, and then using the given names as stand-ins for the phenomena.[4]

In depictions of the real world, eliminative compression might also look like the leaving out by the author of details which do not achieve some desired effect, or which do not achieve the effect as well as another detail might.

Other, more abstract forms of eliminative compression exist, in which information is eliminated which the reader would self-generate through context anyway. This includes taking advantage of shared knowledge, cultural connotation, and reader (default) assumption. Overlap between the information a book attempts to present, and the information already understood by the reader, is a type of redundancy which is ripe for eliminative compression:


At left, the actual text and information presented on the page. At right, the reader’s existing knowledge base and the results of his/her cultural/personal contexts. Where there is predictable overlap, information can be eliminated or “pruned.” Since personal contexts vary, the author can only capitalize on cultural contexts in his/her compression.

An audience’s impression, culturally established, of how reality operates will affect his default assumptions about the fictional world if it at all parallels the actual world. At the most basic level, this involves simple assumptions about reality — noting that Buck Mulligan is “stately” and “plump” is essential since the reader will not assume it as a default. Noting that Mulligan has two legs is redundant since it is assumed (implicitly conveyed) unless information to the contrary is presented (that Mulligan walks around the Martello tower further conveys, rather than contradicts, this default assumption). At a somewhat more complex level, Joyce mentioning that Bloom carries a potato as he leaves his house is information which the reader will not assume; carrying a key is, however, default behavior which the reader will self-generate and which needs not be mentioned; itcan be accordingly pruned. That Joyce does, in fact, draw attention (spend material resources of type and paper) to Bloom checking for his key — and failing to find it — is a deviation from the typical model of compression we would assume from an author of Joyce’s ability. Via this deviation, we can evaluate why this detail is included; we might even note it in our minds; and when later in the novel, Bloom, in a parallel scene, wonders where his hat went, and hypothesizes that perhaps he “hung it up on the floor,” we notice a pattern of behavior which is working to characterize him. It is not the having (or not having) of a key which is important here — it is that Bloom has nearly forgotten his key, having left it in a trouser pocket, and that this says something about him as a human being.

The presence of predictable reader connotation in a given audience also allows for eliminative compression. Consider the opening lines to The Importance of Being Earnest:

Algernon:  Did you hear what I was playing, Lane?

Lane:  I didn’t think it polite to listen, sir.

Algernon:  I’m sorry for that, for your sake. I don’t play accurately—anyone can play accurately—but I play with wonderful expression. As far as the piano is concerned, sentiment is my forte.  I keep science for Life.

There is no narratorial intrusion to comment on Algernon’s statement (partly inherent to the play’s format), nor does Lane or any other character offer anything in the way of contradiction or support. The lines stand on their own at the opening of the play, and therefore are only effective at characterization — effective at all as an opener, really — because of their implicit asceticist connotations which would have been perceived implicitly by the play’s contemporaneous audiences. The characterization of Algernon — what it means about him that he is saying these lines — is generated from the audience’s cultural and personal contexts applied to evaluate the character’s statement; adverbs, engaged interlocutors, or intrusive narratorial comment are all unnecessary, superfluous, in creating this effect; they are inefficient uses of the page; they lack artful grace in that they eliminate subtlety and come off as too on-the-nose.

Once signals have been cleared out of extraneous data through curation and symbol, they can be packed together — additively compressed — so that single words or sentences accomplish many things, and transmit multiple, complex pieces of information, at once (this involves, essentially, compacting many eliminative compressions together). To say “joggerfry” instead of “geography” in “Calypso” locates Bloom temporally and, yes, geographically in Ireland of the era, where the former was established gradeschool slang for the latter. It characterizes Bloom, as a man who would use such slang, and as still somewhat childish, or at least a nostalgic, for using the language of a young Dublin student. For an Irish reader, it creates a solidarity between reader and character; for the non-Irish reader, it gives the story an appealing exoticism. And, perhaps chief among the author’s priorities, it allows Joyce to further his experiment (or play) with the nuances of language and linguistic representation.

Ambiguity is a product or type of additive compression, in that it refers to the coexistence of two or more diverging narratives or meanings in a single text. Its mere presence certainly creates a valuable or useful effect: it pushes readers and critics into debates about reality, about psychology, about philosophy, which can all stem from what on the surface appears to be the simplest of sentences; essays, dissertations, and five-hundred page analytical texts can all be spawned from a half-dozen words if dense enough, where density refers not just to how much information has been packed by the author into a text (intentionality) but by the potential the text holds to spawn new, unintended information (in a way similar to theories of reader-completion in reader-response theory). Nowhere is this truer than the enormously dense and ambiguous Ulysses.

Here subtlety and the subterranean — so often seen as markers of successful art — are, moreover, byproducts of skillfully layered, additive compression. A lack of subtlety will inevitably correlate with one-dimensional text, inefficient and uneconomical for the space it takes up on a page. When this layering form itself — its hidden, subterranean quality — mirrors content — say, the suppressed emotion or sexual impulse — and here again Ulysses comes to mind — then the compression itself becomes symbolic, adding an additional layer at the meta level.

Additive framing is, as its name implies, a sort of compression. Joyce’s titular decision with Ulysses creates two sets of possible interpretations for essentially every event, character, and interaction which takes place in its pages — the literal event itself and the event in relation to the Odyssey, either as symbolic of, or modified by, the epic. Gilbert alone dedicates half-a-dozen pages to the various intersections and parallelisms between the two works in “Calypso” alone. We are given an easy, though imperfect, analogy for additive framing in “Calypso”’s pages: as Bloom walks across town after getting his kidney, he has a romantic vision of Israel — until a cloud covers the sun, and he sees it anew as a wasteland. Whereas these two interpretations are temporally separate, additive framing allows for simultaneous, divergent or supplementary readings, which, packed together, respectively enrichen the text through either the number of significations or the depth and fullness of the individual signification.

Credibility itself can be potentially seen a type of framing compression (which in turn is a subset of multivalence-as-compression) — it shifts the cipher of how a reader will read and understand the text. But whereas the effect of framing compression is that both meanings or significances are visible, coexisting and thus increasing the text’s density, a reader does not read a text as if it was both by a noncredible author and by a credible one. Instead, I think it more accurate to understand credibility as a tool by which compression can be maximally unpacked, a catalyst or incentive for the process. Unpacking itself often takes sizeable cognitive effort, and the reader will not exert that necessary effort unless he has faith that the unpacking will yield a useful, beneficial result (that the use of his time will be economical, valuable). This is perhaps why some texts are only revealed as dense and compressed, and celebrated critically accordingly, decades after their publication, when a leap of faith is finally taken and the necessary time taken to decompress the work. New Critics would likely argue that the text should argue its credibility on the surface in order to encourage subterranean exploration, but in practice many of the great texts benefit from established authorial reputations; they receive the benefit of the doubt not in the judgments they receive but in whether their audiences and critics will put the work in to explore the text at multiple, complex levels. Often, this effect is unconscious. Close reading obviously renders a dense text more effective since the reader is more open to the effects of its stimuli, to the reception of its ideas and the entertaining of its explorations. Credibility challenges reader to apply a number of different decompression algorithms, thereby allowing for the text’s actualization from flatness into multidimensionality.


Compression, however, while a hallmark of effective and successful art, has its inevitable sinkholes. This is especially true of eliminative compression, since it so often relies on a distinct audience’s cultural in order to avoid informational redundancy. As an audience’s context and knowledge shifts, key information is lost, and texts become increasingly impenetrable. This decay (in meaning, effect, understanding) happens both temporally and subculturally — in any scenario in which a reader is not a member of the group which constituted the book’s original intended audience; a reader, therefore, for whom the compression algorithm was not optimized.

Some decayed compressions are less problematic than others. There is a difference, for example, between Joyce’s “joggerfry” compression and the opening lines to Wilde’s The Importance of Being Earnest. The former is a known unknown: while it may interrupt the transmission of effect and information between text and reader, it is still accessible through annotation or research by the reader. Moreover, it will not likely misread; merely skipped-over or clarified externally. It still retains value as a literary puzzle in the sense outlined by Sarah Perry in her seminal “Puzzle Theory” — and judging by Joyce’s remarks on the subject, a literary puzzle is not far from how he intended Ulysses to work. “Got a short knock,” or the book’s many Dublin-specific allusions, works similarly.

Algernon’s comments on expressive piano playing, however, are an unknown unknown to the contemporary reader. The language and references are all familiar; even the character that Algernon is being presented as is familiar — we recognize this trope. We have seen it in the teenaged troubadour who, at a house party, playing an acoustic in front of his objet du désir, gaffes publicly and in order to save face, expounds a personally philosophy of expressiveness over ability. To the contemporary reader the declaration is adolescent, it lacks self-awareness. And yet to the contemporaneous reader — a historical peer to Wilde, familiar with the general philosophy of asceticism and perhaps Wilde’s particular identification with the outlook — this exchange would come off an entirely different light. Again, we think of Bloom’s meditation on Israel and the changing cloud, with the readership of the contemporary and contemporaneous era only able to see Algernon’s dialogue in one cast of light — the former, Algernon to adolescent gaffe as Jaffa to abandoned wasteland; the latter, Algernon to high ascetic sensibility. Neither aware of the other and therefore unconsciously extracting entirely different bits of information, noticing entirely different patterns in the text, experiencing entirely different effects — the former, revulsion.


The critic, then, in the absence of cultural continuity, is perhaps the best chance of preserving the old, complicated, highly compressed idea-texts for contemporary relevance — and for ensuring that future texts, which will be similarly compressed for specific audiences and eras (perhaps even more compressed, as cultural fragmentation increases and cultural change accelerates as a result of technological advancement and cultural liberalism); since this compression is necessary to creating a stimulating work with “literary value” of economy, the only way texts will retain their accessibility in the future is through translation. This need not necessarily involve issues of interpretation since the critic need only provide the necessary context — the connotative judgments, the default assumptions, the average knowledge bases — of the work’s contemporaneous readers in order to simulate or reenact the conditions in which the work was written and published. Ulysses has, and should continue to, retain some (relative) accessibility through this process.

One wonders, however, if there is a more serious threat to Ulysses’s longevity.


As critical ideas about a novel’s reading have shifted from author and text models to reader-completion, one wonders if, given the low material cost of contemporary texts, reader time and energy will emerge as a more dominant denominator, a far scarcer resource in need of consideration. The compression techniques and strategies which maximize information and effect per material diverge meaningfully from those which maximize information and effect in a given minute of a reader’s attention. Obviously, the ability of the individual has a bearing on absorption over time, but one can see how a single dense text, requiring laborious unpacking and decompression through meticulous readings and re-readings might hold less utility and value to the reader than multiple, less dense texts which are paragons of clarity and which complement one another. The trade-offs between the two ends are represented in the diagrams above; thus are the perils of multitasking.

In this model, informational density in proportion to material expenditure might merely be a vestigial concern; clarity and ease of comprehension, meanwhile, saves previous reader time. The additive layering of Ulysses, while compressive in the sense of information/effect per book and line, is far from optimized for the resource expenditures of reader effort; almost all its value relies on costly, lengthy extraction. Indeed, high culture has long been synonymous with the complex, difficult, and demanding. Mass culture, in contrast, bends over backwards for legibility, for ease of comprehension; it works so that its consumer does not have to.

If generous, one thinks of the reader-optimized model of literature as akin to an assembly line — the texts work in conjunction, and are valued for how they contribute efficiently and skillfully certain valuable effects or transmissions of information. Mass paperbacks need not be bad; they need only the attention of the skilled authors and of skilled audiences. We are already seeing a marked shift away from the postmodern doorstops of the twentieth century, which themselves were arguably attempts to reclaim this vague, ineffable “seriousness” of art through difficulty — an attempt which began with Joyce, Ulysses, and his high-modern peers. Whereas material-optimized literature is about richness and density, reader-optimized literature might prioritize clarity for quick transmission — eliminative compression in the service of one clear signal, rather than the additive layerings of many signals on top one another. Non-academic, non-”serious,” so-called middlebrow or lowerbrow readers have always chosen their literature for ease of comprehension; mass culture has always presented itself in accord.[5]

But we might also argue that this difference between high and mass-culture — its disparity in difficulty, so to speak — is not arbitrary but tied inextricably to differences in function. In explanation, let us mobilize Shaw and turn a turn of phrase. If, The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man., so too can we understand art. Only literature which imposes itself on the reader can effect transformation either personal or cultural, and thus those who look to the arts for more than leisure and pastime must seek out only unreasonable works. It is readily apparent why this would be true in matters of content and subject: works which are perfectly agreeable ethically and philosophically, which do not confront and therefore require the reconciliation of the reader, cannot transform — can only reaffirm consensus. It less clear, however, whether difficulty in comprehension similarly hosts transformative powers or whether it merely acts as a proxy for ethical/philosophical provocation. We can, however, hypothesize a scenario in which a reader, forced to grapple with a difficult text at length, better integrates and stores its informational content. Linguistically (and ethically, philosophically) difficult texts featuring both additive and eliminative compression are in this model optimized for their transformative power over the reader (and in aggregate, over the culture). Highly compressed texts, that is, are optimized for optimization.


[1] (though developments in the digital humanities appear promising)

[2] To identify a pattern is to be able to predict what comes next in a sequence; reliable prediction of, and preparation for, the future carries obvious evolutionary benefits.

[3] It would do good to clarify that both story-driven and idea-driven novels are working towards cognitive stimulation; it is merely that the type of stimulus is different. In the former, narrative is a direct means of engaging the interest of the reader; his attention piqued by teleological significance, suspense, and anticipation; he is invested in outcome due to character development in their lives. These are the many simultaneous effects which economical storytelling achieves. In idea-driven literature, narrative is an indirect means, which facilitates the development of other, varied cognitive stimuli separate from the narrative drive.

[4]  Books themselves — complex, seemingly irreducible works — can be compressed into symbols. Consider the structural and stylistic features leading to what we refer to as a “Joycean” work, or the pattern of mood crystallized into the label of the “Kafkaesque.”

[5] One popular explanation is economic: the so-called masses have less leisure time, and less energy in their leisure time, than cultural elites (this is the hypothesis favored by Macdonald in his paradigmatic “Masscult and Midcult”). And yet increases in leisure time since Macdonald have not witnessed corresponding increases in demand for dense, difficult art.

1 Comment

  1. Fascinating. Although, ironically, I’d be interested in hearing a more straightforward account of the model and how to apply it – but I’m a STEM type…

    It was interesting that you touched on a concern I’ve been having myself lately, namely that heavy compression and multilayered denseness requires so much shared vocabulary and common ideas between writer and reader that it’s becoming more difficult to pull off in a modern, fragmented cultural landscape. It’s become much harder to know what readers will or won’t understand, and any works of high density is almost necessarily confined to narrow subcultures because of this. But maybe it was always like that, and “high culture” has always been a subculture pretending to be something more.

    Liked by 1 person

Comments are closed.