Liars Figure, But Figures Are Dumb

 New York Times Pitchbot:

Our Bayesian model shows only a 10% chance that Shakespeare was really a great writer. And Nate tells you which prisons have the best bread and water. All that and more on the next 538 podcast.

Yeah, this is a thing.

If Google is any guide, Bayesian statistics are usually applied to Shakespeare for things like this:

Bayesian probability theory can be helpful in organizing the multiple evaluations required in analyzing complex problems that involve the comparison of several hypotheses on the basis of several datasets. The problem of deciding the authorship of the Shakespeare literary material falls under this heading. We here discuss just one aspect of this major problem: whether or not the available evidence indicates that “William Shakspere,” of Stratford-upon-Avon, was a writer. We consider 24 known writers who lived in England at the same time as Shakspere. For each of these writers, and for Shakspere, we follow Price in considering whether or not there exists evidence in each of 10 categories relevant to the literary profession. We find that there is evidence conforming to at least 3 categories for each comparison author, but none for Shakspere. We evaluate the probability, based on this information, that Shakspere was a writer similar to the 24 comparison writers. According to this analysis of Price’s data, we find that there is only one chance in 100,000 that Shakspere was a writer. These considerations support the heretical view that Shakspere was not the author of the Shakespeare material. I have questions; this does not provide answers.  But if you want to chase down that rabbit trail, this is a way to look "scientific" while doing it.

This, is not:

I could go on and on about the failings of Shakespeare and the constitution and Stradivarius violins, and at the bottom of this post I do*, but really I shouldn't need to: the Bayesian priors are pretty damning. About half of the people born since 1600 have been born in the past 100 years, but it gets much worse than that. When Shakespeare wrote almost all of Europeans were busy farming, and very few people attended university; few people were even literate--probably as low as about ten million people. By contrast there are now upwards of a billion literate people in the Western sphere. What are the odds that the greatest writer would have been born in 1564? The Bayesian priors aren't very favorable. 

Ben Jonson said of Shakespeare, his contemporary, that he had "little Latin and less Greek."  In 17th century terms, this meant he was not well-educated.  Jonson was, and yet we remember Shakespeare, not Jonson, generally.  Alexander Pope was the son of merchants, yet he spent his adult life in the rarified atmosphere of the English aristocracy, and championed the idea of the "Great Chain of Being," and proved himself Jonson's equal in education.  And yet he was no Shakespeare.

What are the odds that Bayesian priors have nothing to do with artistic ability?

Shakespeare came of age during the English Renaissance, at a time when England itself was catching up with Europe; and yet Shakespeare stands head and shoulders above the greatest playwrights and writers of Europe itself.  The article from which I drew that SBF quote argues this has to do with Shakespeare, basically, getting there first.  Which is quite a slam on Marlow (who, had he not died young, might well have challenged Will as England's greatest playwright), not to mention Sophocles, Euripides, and several Roman writers of plays.  Not to mention the great playwrights of the 18th and 20th centuries; just to pluck at a few.

Whoever wrote Shakespeare (again, if you want to chase that smoke disappearing in the fog), there can be no doubt of the genius of the work.  Unless you are SBF, apparently.  Whose claim to understand Bayesian "priors" does not put him on the same level as the great poets and critics of English literature, who uniformly acknowledge the premiere status of Shakespeare's works. The blog post from which I draw the SBF quote is as illiterate and uninformed on the subject as SBF is.  Shakespeare is not pre-eminent because he was the first playwright in English, or because he invented the tragedy or the comedy.  He did neither.  He did invent a sonnet form far easier to use in English than the Petrarchan (a/k/a Italian) original; but most poets still use the original.  Shakespeare's version, peerless as he was in it, never caught on.

Shakespeare came along at a time when English itself was shifting, from Middle to Early Modern.  He created neologisms as lightly and freely as Thelonious Monk or Ornette Coleman invented new forms of jazz (and new ways to hear music).  Shakespeare's felicity with metaphor was so great we use many of them today, without realizing where they came from.  Most of his invented words are still in use, too.

What are the Bayesian priors for that?  Or for the work of Monk or Coleman, for that matter?

Shakespeare took the forms of Greek plays (tragedy and comedy), mostly as identified by Aristotle in his Poetics, and over his career invented the five act play.  His great genius was not originality in plots, but in creating characters, and in creating in words on a blank stage, whole worlds.

There is a line (forgive the diversion), in one of my favorite "X-Files" episodes, where a writer tells Dana Scully how amazing hypnotism is, because you can hypnotize people by using "just words."  Shakespeare's plays were presented on a bald stage:  props were swords and costumes, which replaced the masks of Greek dramaturgy.  Otherwise, the stage changed little from Sophocles' time to Shakespeare's.  Shakespeare translates rather poorly to the screen because we can put Hamlet on the parapets of a castle in the opening act of the play, or Lear in a hurricane during his mad scene.  Shakespeare had to describe it to the audience, and the only way he could do that was through dialogue.  A poor writer makes such exposition tedious; in the hands of a genius, it is part of the flow of the story, revelatory of character, rich with dramatic tension, necessary to move the story forward, often thick with dramatic irony (Iago's speeches to the audience, and to other characters in "Othello," is probably the most brilliant use of such irony in any play ever written).  I saw a production of "The Tempest" in London when I was 21, and the dialogue wasn't heavy and cumbersome; it was the play itself!  I saw, years later, a high school production of "A Midsummer Night's Dream," and even my then very young daughter laughed at all the right places and had no trouble following the action (or the comedy) on a stage little more decorated than Shakespeare would have known.  That play usually uses masks, to make Bottom an ass (as he is already!). Which takes us straight back to Sophocles.

Just words.  But Shakespeare uses them to paint scenes like the woods of Dunsinane coming to Macbeth; or the hurricane that roars around Lear; or the monstrous countenance of Caliban.  He uses words to break hearts, make us laugh, make us despair.  Words.  All he has is words.

What are the Bayesian priors for that?  Except all of English literature before Shakespeare.  There is actually quite a lot of it, though I doubt SBF is familiar with any of it.  He thinks of literature as monkeys pounding on typewriters: inevitably, the old canard goes, they would produce the works of Shakespeare.  But why?  Because the universe is infinite and time is infinite and at some point all the stars line up and a word of poetry is produced?  Doesn't that violate a law of thermodynamics, or something?  Are we all just random collections of random particles bouncing randomly around an infinite space in infinite time, and occasionally actually forming a human being with what it thinks is a human consciousness, who further thinks it is moved, motivated, illuminated, entertained, informed, engaged, by "just words"?

Because that's the bottom line of SBF's "analysis." If you can call it that. I don't know what his argument for Shakespeare's failings is.  I don't want to know.  I suspect it is ignorance on his part that is the failing; not Shakespeare's works.  I used to get the same "analysis" from people who claimed they'd "read the Bible," as if that was all that was necessary.  I've read James Joyce, and have never finished Ulysses, nor do I ever intend to.  Life's too short, IMHO.  Does that make me right, and all the admirers of Joyce wrong?  No.  It leaves me out in the cold, I'm quite sure.  But Joyce simply doesn't appeal to me, somehow.  OTOH, I consider his story "The Dead" one of the finest in the language.

What are the Bayesian priors for that, I wonder?  Except no, I really don't. It's an irrelevant and stupid consideration. Sort of like setting out the statistical probabilities that I would meet the love of my life over 50 years ago, and still be in love with her to this good day.  I know, I know, the available pool is confined to people I met in my hometown in my high school years.  Except my brother met his wife in college.  And some people never find a lifelong love.  So what are the Bayesian priors for that?  And how do you prove it, without referring (ultimately) back to the concept of Bayesian statistics? 

I mean, it's pretty much a closed system, isn't it?  Bayesian analysis that says odds are against Shakespeare being the author of the works Ben Jonson said he was the author of?  How do you prove that analysis valid, except by appeal to the concepts underlying statistics and Bayesian analysis?  And does that account for why I love my wife?

I suppose in a large enough generality, it does.  But how large?  Large enough to say that if you close one eye and lay on your right (not left!) side and hold the numbers upside down you can see that some political polls accurately predicted a trend that supports a supposition that proves an inference that the present was somehow foreshadowed in the past?  I mean, if you insist the math has to be correct, and all....