(My commenters wrote my first response to Bill's post for me. They nailed it so accurately posting what I'd originally written seems unnecessary. I'm neither kidding nor, it seems, necessary. So in the [likely] event of a [hilariously hi-jinxed] tragedy, Acephalous can [and should] live on.)
Let me start with a statement that will annoy everyone: if a close-reading reveals that a work flirts with the formal elements of its genre or genres—whatever they may be—that work should be canonized. Not that works that fail to engage the formal limitations of their genre are uncanonizable, mind you, but works that succeed both as an example and a critique of a given genre deserve canonization.
But canonization into what?
In an age of inexpensive and practically limitless storage, the question of canonization need not be hidebound to the idea of preservation. Within its first month of operation, Google digitized the 99 percent of the Western Canon, and even though some of those works are too recent to be viewed, they'll all eventually be released as copyright expiration rolls forward. When I began my Mark Twain chapter in late 2005, for example, only the 1894 edition of Pudd'nhead Wilson was available through Google Book Search; by the time I began revising the chapter in the summer of 2008, I could track revisions of the novel over the span of two decades. Because Twain is culturally significant and canonical, the saturation of Google Books with variant editions of his most important works was inevitable.
When I began working on The Youth of Washington (1904), I had to order it through interlibrary loan. It took three months to arrive. Henry Cabot Lodge's George Washington (1889), Paul Ford's The True George Washington (1896), Woodrow Wilson's George Washington (1896), Worthington Chauncey Ford's George Washington(1900) and Norman Hapgood's George Washington: A Biography (1901) trickled in. Had I held off on writing my chapter until I'd looked over all 140 of the novels of English Colonial or Revolutionary America published between 1895 and 1908, I'd still be waiting for interlibrary loan. Now all those Washington biographies are available, as are most of the historical novels I wanted to read for deep background.
Are those novels good? No. Do they deserve canonization? No. Is it significant that as tensions between Spain and America strained and Americans became uncomfortable with the imperial pretensions of their leader, an appetite for works relating to Revolutionary figures or set in the Revolutionary period become incredibly popular? Might that not have something important to say about what Americans thought it meant to be American at the time? Is that not a viable object of study? Do I not ask a shitload of rhetorical questions when I get polemical?
For a few of generations, English professors claimed that cultural knowledge was the provenance of the literary (what with perceptiveness being the core feature of literary sensibility). So when a scholar wanted to know how things stood between America and Europe at a given time, they would not turn to any of the countless travel narratives written by Americans in Europe and Europeans in America, but to the most acutely literary accounts of the current state affairs. To wit:
The problem with Lucy's account of the canon and cultural knowledge should be obvious: however you define the literary, it is not the same thing as cultural knowledge. An alternate canon, based on how a text registers and reflects the conventions of its time, is required; a subset of that canon would include texts that fought against their cultural constraints in order to articulate something convention had difficulty accommodating. No matter what you call those texts, they are the ones deserving of close-reading.
Note that I slipped from speaking of genres in the first paragraph to cultural conventions in the last. I did that on purpose. Genre focuses too intently on what a thing made from words looks like. ("My novella's a long short story in 10-point Times Roman and a short novel in 14-point New Courier," says the aspiring writer.) From the get-go, however, these things made from words contained stuff like this. Texts have always had visual components, and while those components are sometimes abstracts (as with the afore-marbled page), the introduction of visual representations of the world into the textual economy of a novel alerts readers to the presence of a realist ethos. But remember:
Verisimilitude is not an end. It is an always imperfect—because always filtered—means. What matters is not the presence of supplementary gestures of verisimilitude but the manner in which they interact with the text. For example, this comic sucks:
The image of Red-Eye (from Jack London's 1906 novel Before Adam) adds nothing to the description provided by the text; in fact, it looks like what a sketch artist would draw if provided the description in the text. Now compare that to this or this or this or this or this. Why do I have to justify studying the subject of all those thises but can write about Red-Eye with nary a care? Is it because Before Adam was written by London? Because he wrote it a century ago? Because a quinquagenarian with Ivy credentials stretching four generations back wouldn't feel mortified if caught reading it on the subway?