The Turing Test, first proposed concretely in 1950
by Alan Turing, the father of modern computing, supposedly tests whether a computer can think like a human being. But it doesn't measure that at all, but rather whether a computer can fool a human into believing it's another human. And the test in one form or another has been fooling unsophisticated observers ever since 1961 's ELIZA. ELIZA was basically a rudimentary chatbot, and it wasn't intelligent at all. It was basically a parlor trick, a shuffling of conversational cards, using symbol manipulation sans understanding, sans consciousness, never directly answering, rephrasing and obfuscating any questions which might reveal that it was not, in fact, human. AI still has no consciousness, 60 years later.
Through the years both the test and the testers have become more sophisticated; the goal remains elusive. But it struck me recently that AI's problems with the Turing test can be analogized to the test a historical novelist must face: they must, like magicians, force a card on their unsuspecting subjects, presenting what they know while hiding what they don't know, convincing the audience they are human; or in the novelists' case, convincing them that they are humans of another time period. To manage this sleight of hand we are partially dependent upon the audience's complicity of ignorance--what the audience doesn't know, and therefore has to take on faith. And god forgive you if you've chosen an 18th century Gascon woodworker as your murder victim and your book has just been read by a Ph.D. in French literature who wrote her thesis on carpentry techniques in the south of France 1820--1870. You're busted. There's always someone who knows some detail (however picayune) about the era which has eluded you.
Because nobody writing today lived then--obviously (unless your novel takes place within living memory). Nobody, not even aforesaid Ph. D., can know what it was like like to live and breathe those wood shavings from three hundred years ago. The only trick we have available is not to duplicate the time period exactly, but to capture enough significant detail that the reader will be lulled, will fill in the lacunae with their imagination. And for that task we turn to a number of available resources.
For instance, in my first book, I needed to delay a train from London to Birmingham in the year 1912. I found the answer online--an RAF plane crash that year near Oxfordshire that could be slotted in nicely. For my second book I wanted to know as much about 19th century art forgery as possible. I came across a fascinating book called The Deceivers by Aviva Breifel which painted a picture of female student copyists in the Louvre in 1890s, setting
the scene perfectly. For my third book, I wanted Dr. Watson to develop claustrophobia from attempting to explore the Great Pyramid at Giza. This was a rather easy cheat--I had personally developed claustrophobia exploring the Great Pyramid at Giza. We can call these examples significant detail or local color--in AI they're known as training sets. And it is incredible drudge work for humans to train AI. To train it to recognize a dog, for instance, it must be fed thousands of pictures of dogs--labeled as dogs, by humans. AI can't envision a dog, not yet. We can.Because there's no such thing as a "dog." There are malamutes and mutts, chihuahuas and chows, Great Danes and English bulldogs. The "dog" exists only in our minds, abstracted and analogized from individual experience. It may be a specific breed of dog, or an amalgam of different dogs, whether a concrete creation or an ever-shifting eidolon. The same is true of a tree. There are oaks and pines and sequoias and aspens and birches. No "tree." But when you say the word, an image comes immediately to mind. I suspect this is the phenomenon which gave rise to Plato's forms.
Plato's forms are the eternal essences, not only things, but qualities: Truth, Beauty, Good.The Platonic ideal may not exist in space, but it does exist within our minds. I would extend this idea to say that there are Platonic ideals of historical eras. Each unit of description, action, or dialogue in a historical novel must be checked against that ideal, that reader's understanding, just as we check both chihuahuas and rottweillers against the platonic dog in our minds--and for each person that dog is a little bit different.This may seem a handicap for the writer, but it's actually a gift. The writer doesn't have to spend many dreary pages setting each scene. There is a tacit agreement between writer and reader that unless historical differences are specifically detailed , things do not change between time periods. A door is a door, a street is a street, unless we add a brass knocker to the door, or cobblestones to the street. Fashion may change, technology may change but:
You must remember this A kiss is just a kiss, a sigh is just a sigh.
The fundamental things apply
As time goes by.
Some things don't seem to change, like human emotions, hearts full of passion, jealousy and hate, death and taxes. Or if they do, we turn a blind eye to them like footprints in the snow, soon to be wiped out by a new snowfall.
This extends especially to habits of human thought. I remember reading Michener's The Source in college and thinking it was utter bosh because it portrayed prehistoric characters as using exactly the same processes as modern man. My doubts sprang from the fact that I had just read Julian Jaynes's The Origin of Consciousness in the Breakdown of the Bicameral Mind (a fascinating book), which posited that early man had a completely different consciousness. (I won't go into detail, which would require a much longer piece to do it justice) It was a very convincing argument, though I don't know that it's true. Perhaps our minds have evolved over the centuries, even over the decades. How could Michener have presented the difference, even if he could grasp it? It would be a job of simultaneous translation from an unknown language, fumbling in the dark for a new Rosetta Stone.
We cannot know the language of the past. We can only fish for a few clues, relying on the good will and cooperation of the reader, to admire our catch. Luckily, we're not subject to a Turing test. No fooling.
