Marginal Literary Revolution

Average Poetry, Marginal Great Books

Tyler Cowen’s The Marginal Revolution: Rise and Decline, and the Pending AI Revolution (2026) may be the most important book he has written, or at least the most important for my purposes. Tyler invited readers to see it as “a non-fiction novella of sorts,” which opened the door to applying his work to questions rarely asked by literary scholars: what do most readers want to read? And what does that preference have to do with which books survive? I studied statistics and microeconomics before I studied literature, so I am primed to apply economic thinking to canons and syllabi and anthologies — groups of texts that “represent” by being taught and remain in print. The landscape of works in print seems like the entire corpus rather than representative samples. In fact all the works still in print are already only a very small selected sample of what once existed.¹

For literary readers, here’s the short version of Tyler’s argument. Two things happened as statistics developed in the nineteenth century: first, the rise of marginalism, the idea that economic value comes from the effect of one additional unit under fixed conditions (one more loaf of bread, one more hour of labor). Second was the rise of averageism, the idea that aggregates are best described through averages, with innovations such as Adolphe Quetelet’s “average man” and William Stanley Jevons’s calculation of index numbers and average consumption across populations. Tyler argues that marginalism and averageism are a kind of package deal, advancing together as mutually reinforcing methods. Averages became central to measurement, while marginal reasoning helped interpret those measurements. The marginal unit is only legible against a distribution, and the distribution is only interesting because marginal changes can be inferred from it. The coming AI revolution, as Tyler sees it, brings averageism back to the center.

The literary analogue of the economic margin is one more text added to a syllabus, anthology, or canon under otherwise fixed conditions. Literature scholars don’t use this sort of language even when anthologizing or putting together a syllabus or arguing about the canon. We don’t ask what the difference is between an “average” Victorian novel or an “average Victorian novel” or a representative Victorian novel, especially when only a few dozen (out of the many thousands that were written) are taught regularly. Is a particular Dickens novel average? Marginal? How can Tyler help us here?

Tyler might recharacterize the “average Victorian novel” as an averageist object: a construct built from many novels. The “average novel” is only a concept in relation to the corpus (let’s say the average novel was a certain number of pages or has a certain plot structure or cost a certain amount) not at the level of the individual classroom text.

A novel on a syllabus becomes “marginal” when you move from “what is typical of the whole corpus?” to “what does one more novel add?” This second question can even be “what does one more novel bring to the reader who has read all the others?” Tyler’s marginalism is about the effect of an additional unit under fixed conditions, and he keeps insisting that what we observe are average magnitudes while the real analytic task is to infer the marginal effect from them. A syllabus is exactly that inference problem in pedagogical form. If I add this novel, how much does the student’s map of Victorian fiction change relative to what some other addition would have done?

On that basis, canonical texts are better understood as organizing exemplars than as averages. A syllabus is a systematization, a small set of texts chosen to make the field’s sorting rules visible to a reader who has not yet internalized them. It picks a few texts that let students organize the field into genres, problems, forms, and historical tensions. Those texts may be canonical because they are especially useful for structuring thought, not because they sit at the statistical center of the archive.

Anthologies

All literature professors should try their hand at anthologizing. An anthologist assembles averageist objects, though we don’t use this term in literary studies. An anthology is a corpus-reducing device. A representative anthology approximates an averageist object; a “best of” anthology performs a different task, closer to canon formation. Both are legitimate scholarly work: one needs to know the whole field well in order to curate a “best of” or representative sample.

I am currently assembling an anthology of African American sonnets, collected in the course of writing my book on the African American sonnet tradition. My new anthology is a corpus-level construction. My point is to show what a form does across a tradition, which means it has to include sonnets that are competent rather than singular, sonnets that establish the defaults against which the great ones are obviously great. Each editorial decision about inclusion can be described as a marginalist decision. It doesn’t matter whether a particular sonnet is good in isolation. The question is what it adds to the reader’s map once the other sonnets are already there. The peaks in the tradition (Claude McKay’s “If We Must Die,” Robert Hayden’s “Frederick Douglass,” any Gwendolyn Brooks sonnet, Terrance Hayes’s breathtaking recent work) stand out. The eye and ear are trained by “typical” or “average” works (which is not to say they aren’t all very very good) to understand why the peaks are peaks. You need the whole corpus to understand the best of the corpus and the best of the corpus explains the corpus.

My averageist claim about literature departs from (rather than extends) Tyler’s thinking. He treats peaks such as William Shakespeare (“the greatest author of all time, to this day”) and Johann Sebastian Bach (“perhaps…the greatest and most musically complex composer of all time?”) as evidence that pre-marginalist civilizations produced greatness, but he does not treat the peaks as emerging from a distribution. I do. I’ve spent countless hours reading “average” literature in order to understand how the best is best.²

The “who decides which books are Great Books” question is a bugbear because Great Books are usually chosen as organizing exemplars or representative proxies, not as averages of an archive. Books count as “marginal” only when you ask how much adding or dropping it changes the reader’s understanding relative to other possible additions. The “great books” are fluid of course. I taught in the Great Books program at Johns Hopkins for years, in which faculty chose their own Great Book; I chose Frederick Douglass’s Narrative (1845), which had high marginal value because of how it sat against the rest of the syllabus, not because it escaped the average. It opened a new genre, form, and civic position, “adding” more than another statistically typical “great book.”

The great syllabus stagnation

I’ve written about the problem of everyone sharing the same syllabi across U.S. universities. Tyler would suggest that on the one hand, the homogenization revealed by the Open Syllabus Project shows that various fields’ classificatory schemes and teaching frameworks are working and appropriately “scaled.” English literature, for example, is almost uniformly organized by century, country, and genre. Teaching the “same” American literature course across institutions lowers coordination costs, makes survey teaching possible, and gives faculty a small set of proxy texts that can stand in for a much larger archive. Students may only get one or two literature classes, making these choices zero-sum.

But on the other hand, homogenization can turn into staleness. Tyler might see this as decline; I see it as a kind of attenuation, which means that things persist in a weakened form. As method and canon rise, they become background, receding from the frontier while remaining embedded in ordinary practice. Those of us deep in a field see how some texts persist for no particular reason. University libraries may still have many copies. Recorded lectures on the text have the lowest marginal delivery cost. The organizing principle still coordinating teaching is no longer carrying the field’s most active intellectual energy. The scheme still sorts the syllabus, yet fewer people can say why this scheme should dominate over alternatives.

What would Tyler say about the margins? Open Syllabus says the disciplinary gravity of English is real, though fragile, and that thematic or problem-centered curricula can pull titles out of the old clusters. His diagnostic question is about the border zones that reveal the formation of a new organizing principle. If nothing strong pulls against the same small center, the field looks stale. But if new formations start emerging, the field could be reorganizing itself. Homogenization and a common organizing system can be evidence of maturity or evidence of inertia. It depends how you are looking at the margins and you need to look at the whole corpus to see the margins.

A new border zone

This fall I’m scheduled to teach a 20^th century American literature course. Most professors will pick representative samples: some Edith Wharton, some F. Scott Fitzgerald, maybe some Robert Frost, Flannery O’Connor, Richard Wright, Philip Roth, Toni Morrison, depending on the professor’s preference or the course’s “theme.” The idea is each text “stands in for” a type of text and does a certain kind of work. Tyler would call this a representative or proxy-building move. Everyone reading this has taken such a course, I’ll wager. Every single one of you.

I decided I would organize the syllabus around works of American literature that Chinese professors assign to teach their students about America and Americans. Tyler would first say I have changed the course’s organizing question, or rather I am offering a new organizing category, grouping texts by reception and use-value inside a new interpretive community. Tyler suggests that fields move when someone supplies a workable scheme for organizing knowledge. I am supplying a new organizing principle. If “the humanities” are dead, perhaps staleness is why. More on this below.

Google Gemini’s version of the inside of the Great Library of Alexandria...

But first, a question: can my course replicate across universities, like Linnaeus’s “arbitrary” rule was useful across researchers? Perhaps. I would need to explain why my organizing principle should replicate. A syllabus built from what Chinese scholars read to learn about America is highly valuable for a specific problem, which is knowledge formation in China. They read us, while we do not read them. They have books and curricula about how to read American literature in their universities. We absolutely do not mirror this, at our peril, I might add.

I take from Tyler’s book the idea that a new organizing structure can force a “see around the corner” shift. In chapter 3 (“Why Did It Take So Long for the Science of Economics to Develop?”), his Linnaeus example shows how a new series of organizing categories can create a common framework for knowledge, make new kinds of comparison possible, and open the way to further inquiry. Important advances come from learning to see a field through a new conceptual scheme. A course built around how the Chinese read us reveals which texts carry explanatory weight abroad. The marginal pedagogical value of one added text could be very high once the course goal is defined that way. My syllabus is not meant as a universal map of twentieth-century American literature. It is a special-purpose taxonomy. But I do hope to make students think. I haven’t completed the syllabus but Mario Puzo’s The Godfather (1969) will be on it, as will be some Robert Frost.³

My course will overlap substantially with a standard American literature survey because Chinese syllabi often draw from the same canonical pool, but it reorganizes that pool for a different question. It is both a standard survey and a diagnostic instrument.

My sonnet anthology is doing the same double duty from the other direction, as an averageist construction of a tradition that has never been assembled as a corpus, which means the act of building the average is itself the marginal contribution to the field. The sonnet form long outlived its original Petrarchan and Renaissance rationales, and the African American sonnet tradition is partly the story of writers picking an old scheme and putting new intellectual energy into it. Sometimes a scheme survives its original rationale because it turns out to be a good container for arguments its inventors never anticipated, and the people who notice this are usually writers rather than methodologists. Tyler’s arc runs from rise to decline to supersession by a new research program. The sonnet case suggests things may not always be so linear. Old forms can be reactivated and new energy put into schemes the methodologists have moved past.

Many people seem alarmed by the studies showing that people prefer LLM prose. Most readers do not want great literature. They want average literature. This is not a recent development and it is not caused by LLMs. Most bestsellers will turn out to be books nobody reads anymore. The peaks that survive are not what most readers were buying at the time. Food critics do not tend to write about the places most people eat most of the time. Tyler treats averageism as a research program, a way of measuring what people do. Averageism is also a fact about what people want. Higher education ought to be about what people need.

Tyler ends his excellent book by suggesting that AI reveals economic intuition to have been a small corner of understanding all along. The literary version of his point is that AI shows how the distinction between great and average writing has never been as important as literary critics thought, because most readers were never operating on that distinction in the first place. People were primed to like LLM prose by their prior preference for averageist prose. LLMs did not create the taste they satisfy.

1. The Library of Congress has archived Marginal Revolution as part of its historic collection of economics blogs. The averageist machinery of preference will move on, as it always does, but somewhere a librarian decided this one was worth keeping.

2. I rarely look at “lists of books” put out by ex-presidents, CEOs, famous VCs because such lists tend to be performative preferences, not averageist or marginal texts.

3. The Chinese love Robert Frost and many sources cite Xi Jinping’s admiration for The Godfather film.