(I think I posted the first part of this 2 years ago as a quiz, then deleted it and never posted the rest because you guys were too smart to fall for the “P(B|A) ~ P(A|B)” fallacy.)

Half of undergraduates at ivy league schools scored at the 98th percentile or above on their SATs.  (This is the true value, accurate to about 2 decimal places. I computed it myself from the 2011 and 2012 data for all 8 ivy league schools.)

If someone in an American undergraduate college scored in the 98th percentile or above on their SATs, what are the chances that they’re in an ivy league college?

P(X|Y) means the probability that X is true given that Y is true.  So we can restate this question as:  Given P( 98th_percentile(A) | ivy(A) ) = 0.50, what is P( ivy(A) | 98th_percentile(A) ) ?

A) 10%

B) 50%

C) 90%

The answer, which I also computed myself: P( ivy(A) | 98th(A) ) = 0.0876.

Just because P(X|Y) is high doesn’t mean P(Y|X) is high.

[page_break]

I've been reading stories from the New Yorker, Atlantic Monthly, Tin House, Ploughshares, and the Kenyon Review, trying to figure out what they want. Instead of a wild ecosystem of stories of all kind, they feel to me like most of them were written by the same author. The only ironclad rule literary magazines have, as far as I can tell, is that they will not publish anything that has a beginning, a middle, and an end that work together to tell a story. Anything with a clear interpretation is, presumably, not subtle enough.

I don’t like these kinds of stories. The reader comments on All the Pretty Pony Princesses” (which is a story like that) made me think there can be some merit to an incomplete story like that. The different possible interpretations in the comments showed that the story made people think and let them participate in the story telling. I usually call that explanation “pseudo-intellectual wankery”, but I guess… maybe there’s some use in reading a story like that, every now and then.  But I don’t think every last goddamn story in every literary magazine in America should have to be like that.

Literature deals with things that can't be reduced to logic and examined in an essay. Instead of clearly defining your terms and delineating your argument, you work with a web of associations and connotations. You grab onto it, pull, and see what comes up.  Cleanth Brooks wrote a famous essay about this called “The Heresy of Paraphrase”, which is chapter 11 in his famous 1947 book The Well Wrought Urn describing his theory of poetry.  To put it more geekily:  You can't summarize a poem, because a poem is its own minimum-length description.  A logical summary of the poem's meaning, using well-defined terms, would be much longer than the poem.  Poetic language communicates more efficiently, and possibly with less error, because it uses roughly the same fuzzy categories and connotations that our brains do.  The "inaccuracy" of language is, I think, a form of data compression.

So great literature is usually subtle, indirect, and ambiguous. It fades when you stare at it directly.

But P(A|B) <> P(B|A).  P( subtle(X), indirect(X), ambiguous(X) | great(X) ) > .8  does not imply P( great(X) | subtle(X), indirect(X), ambiguous(X) ) > .8.  Great literature may usually be subtle, indirect, and ambiguous, but that doesn’t mean that subtlety, indirectness, and ambiguity make great literature.

Stories in literary magazines are, I think, deliberately opaque. They take vague themes and approach them obliquely to create something difficult to comprehend.  This became fashionable in the 1920s with “The Love Song of J. Alfred Prufrock” and Ulysses, which were deliberately crafted to be subtle and ambiguous.

Chapter 1 of Cleanth Brooks’ The Well Wrought Urn says that the basis of poetry is “paradox”:  the expression of two apparently contradictory truths.  Brooks seems to believe that poetry is artfully-constructed tension between opposing attitudes.  This suggests it is the difficulty itself that we enjoy.

I believe, however, that there are no true paradoxes, so manufacturing them for our titillation hardly seems a worthy goal for art.  I prefer to say that good literature confronts apparent paradoxes, not because paradoxes are good, but because resolving them is good.  Literature wants to bring us closer to that resolution, or at least make us more aware of its difficulties.  A poem that honestly showed the tension between the values of Democrats and Republicans would be good.  A poem that successfully resolved them would be even better.  It’s just not feasible.  A poem is unlikely to resolve things we’ve argued over for a century.  Anything that can be resolved in a poem isn't worth writing a poem about.  We've probably already resolved it.

Truly great literature isn’t difficult because difficulty is good.  It’s difficult because it’s trying to do something difficult.  It’s trying to say things as clearly as it can. There is still some paradox, ambiguity, and confusion unavoidably left after the author has tried to make everything as focused and clear as possible, because literature deals with difficult and ambiguous issues.  Lack of clarity is an inevitable side-effects.  It is not therapeutic.

Suppose you want to be a weightlifter.  You go to the Olympics and watch the competitors grunt and strain, their faces contorted in pain, as they lift.  Some of them strain a muscle or tear a ligament.

You conclude that the secret to being a great weightlifter is to lift the weight in a way that causes you pain.  You go home and practice different ways of lifting, and devise these principles of weightlifting:

- Avoid stretching or cooling down.

- Arch your back as much as possible.

- Use machines to isolate weak points such as knees and rotator cuffs.

- Jerk the weights up, and let them fall down using gravity.

- Lift with your back, not with your legs.

- etc.

You're doing it wrong.  In order to lift as much as he can, a weightlifter wants to lift with a form that minimizes pain and stress.  When you see him lift, he is at the limits of his endurance.  That's because every time he learns how to lift a weight with less stress, he adds more weight.

Take a great not-really-modernist work like A Passage to India, The Member of the Wedding, Tess, or anything by Dostoyevsky or Turgenev. You can tell exactly what the author is writing about; you can see the different arguments and counterarguments being made. Yet you're left with doubt and debate, not because the author deliberately hid things, but because the subject matter is difficult.  There is nothing difficult about Dostoyevsky's Grand Inquisitor except that he is human.

But academics don't like to study literature that's written so well that it's easy to understand what it's about. They want works that require a priesthood to divine their purpose and meaning. Like the philosophers in The Hitchhiker's Guide, they demand rigidly defined areas of doubt and uncertainty. They want puzzles, not literature. That's why they prefer Hamlet to Julius Caesar, and William Faulkner to Pat Conroy. And that's why so few people want to read what they publish.