Reflections

It's quite nice to be done with this story, for the time being anyway; it's been bouncing around my head for the past couple of years.

I've long been fascinated by the argument known as "Lady Lovelace's Objection." It claims, roughly speaking, that computers are not capable of original thought, but are instead bound by the instructions given them by people. This claim was given a title by Alan Turing, whose famous paper "Computing Machinery and Intelligence" gave rise to the Turing test while arguing against Lovelace's position. Alas, Countess Lovelace died nearly a century before Turing's paper, and never had the chance to respond to Turing's arguments.

Hence the premise of my story: suppose that Lovelace and Turing had been contemporaries, and had had a chance to carry on a full-fledged debate on the nature of digital intelligence. What then?

Originally I had hoped to make the whole thing rather didactic. My own view is that Turing was mistaken, and Lovelace essentially correct; indeed I would go further and argue that Lovelace is correct by definition. I had planned to write this story as a kind of dramatic exposition on this argument, and my own take on it.

I decided to unravel that intention a bit, I think in order to make the ideas a little more palatable (if you will.) I figured that a dry exposition on algorithms and intelligence would be just a tad boring. So algorithms became recipes, chefs became computers, and typewriters became... I'm not sure exactly. Universal Turing machines? It's not quite clear. Lovelace (whose birth name was Augusta) and Turing (whose middle name was Mathison) become relations - whether siblings or cousins, I'm not entirely sure - reflecting on their inheritance from a loving if somewhat quirky set of grandparents. The exchange of letters follows some of the parameters of the real-world debate between Lovelace and Turing, but loosely at best. To put the matter mildly, I think some of the original ideas get a little muddled.

There is something magical, in my eyes, about the epistolary format. It is so sparse that it forces the reader to fill in the blanks; and it has a lot of room for musings and melancholy, which I find somewhat relaxing. Moreover it opens up the possibility of hidden structure that has the whiff of believability about it: it's no accident that Mathi includes a recipe, of some kind or another, in every letter he sends to Augusta (except the last), for instance. At the same time, as the author, there is a pronounced danger of digression when writing these fictional letters. It's all too easy to bore the reader with extraneous exposition. Altogether, it's a good challenge.

Furthermore, I wanted to explore the techno-magical realism genre a little more. Hence we have a magical typewriter, capable of sending emails, composing recipes, and even sifting through boxes of old junk. That was fun, although it wound up being something of a smaller piece of the story than I had originally imagined. Tippy does loom large in what you might call a dramatic re-enactment of the Turing test, which wound up working out about the way I thought it would. The epistolary format is quite a challenge, and the genre gives it an interesting flavor - so I was happy to try it out.

Did I succeed? I suppose that is for the reader to decide! In my own view the story is less gripping than it might be, and whatever plot there is does get rather buried under the abstract exchange of ideas. It's not altogether believable that Augusta and Mathi would spend quite so much time writing about recipes and cookies, nor is the defense of Lady Lovelace's Objection all that clear or rousing. That entire idea gets rather convoluted and lost.

At the same time, I did succeed in clearing at least a low bar of sorts: I took the idea in my head and gave it concrete manifestation. At some point as I was writing this piece I decided to try and shoe-horn it into National Novel Writing Month, by attempting to write the whole first draft - with a target of five thousand words - inside the month of November. I did meet that goal, with a day or so to spare if I recall. All around I'd say I'm satisfied with hitting those marks.

There were some hidden surprises along the way. I had originally thought that there would be some form of adventure, perhaps even a compelling conflict, in the narrative - perhaps Mathi needs help keeping a restaurant afloat, something like that. Perhaps even Augusta loans him the magical Tippy for a time, and Mathi winds up trying to fool Augusta rather than the other way around. Maybe I might have tried something like that, but I could not find a way to fit it in to the story after the first few letters had been exchanged, which I liked too much to undo. One might say that I had chosen a stylistic beginning over an interesting plot.

What came as a genuine surprise was the literary allusion to the Wizard of Oz; that was nowhere in my original plan. But then I found myself comparing Tippy to the Tin Man and it occurred to me: well, why not? The themes of nostalgia and imitation are all right there for the plucking, and I've always loved that story a great deal. It was kind of a fun diversion. The fact that it brought an American sensibility to a British debate seemed somehow fitting.

Zooming out a little, it's interesting to me that surprises like this one should occur at all during the writing process. It's one of the most unrestrained productive processes I can think of: if a word appears on the page, I can delete it with ease. I can write something today and change it tomorrow, or indeed ten years from now. I have essentially perfect control over what I write. Or do I? After all, I had no plans for this particular motif to appear, and yet here it is. This question itself appears in Turing's critique of Lovelace's objection: the computer surprises the programmer from time to time (in fact, quite often), therefore it must be somewhat intelligent of its own accord. It seems to me that there is a key difference between the intelligence of intent (did I intend for the story to follow a certain path, and to use certain words?) and the intelligence of outcome (did the finished product align to my expectations? might I have changed it if it hadn't?) Turing conflated the two in his famous article, and as a result I think he missed some of the meaning behind Lovelace's original claim. If I had the wherewithal to rigorously respond to Turing's original paper, I might write that originating a piece of work requires both the intelligence of intent and the intelligence of outcome; a computer may or may not have the intelligence of intent, but it certainly lacks the intelligence of outcome.

Back to the story itself - I'm not entirely certain how this abstract exchange of ideas became a sentimental tale of nostalgia. I think it was always there in the premise of a shared inheritance, although it was not meant to take center stage. I have for a long time thought that home, as a literary motif, shares some really striking connections to Turing's notion of an "oracle machine" and its place in the arithmetic hierarchy. Exploring that in greater depth is a topic for another day, but I think the idea bled into this story; as you can see, I let it remain there.

The motif of reflection was a little late to the party, originally something of an afterthought. It blends in nicely with the larger question about how much, or how little, of a machine's intelligence belongs to its creator. And there are a handful of tie-ins to the Wizard of Oz in the bargain. Over and above that, it gave me the title for this appendix, which I think is a little amusing.

The conflict between Augusta and Mathi has changed quite a bit since the first draft, and I think for the better. Gone is the sharp hostility, replaced by a kind of genial ribbing. It's a bit open-ended as to whether Augusta is furious with Mathi at story's end, or whether she's merely teasing him, and I rather like that. To begin with it makes the story that much more enjoyable to read - despite its artistic qualities, I simply dislike conflict - but more than that I think it allows the reader to appreciate better the various ideas swirling around Mathi's cherished recipe, without the distraction of a family feud. More than that, it opens up some space to make the dramatized Turing test a little more interesting. I'm glad to have reworked the ending a bit.

Allow me to close with a word about the first sentence of the story. This particular sentence has been in my head for a long time, and I'm not sure why; only that I like it very much. Of course, it calls to mind the idea at the heart of the story, Lady Lovelace's Objection. It introduces the search for the recipe, and it introduces our two main characters and their relationship; functionally it seems like a reasonable piece of writing. More than that, it's a somewhat oblique reference to Blaise Pascal's famous apology, "I would have written a shorter letter, but I did not have the time." It turns out that the original line in French ("Je n’ai fait celle-ci plus longue que parce que je n’ai pas eu le loisir de la faire plus courte") is a good deal longer than the English translation, which is the kind of thing I find very funny. In truth there is no real thematic connection between this sentence, or the Provincial Letters in which it is embedded, and this story. But somehow I felt it imperative to draw a connection between this debate and Pascal's earlier writing.

I'm certainly not a fan of including explanatory texts like this one as an appendix to my writing - the actual work should stand on its own, in my view - so you may take or leave this reflection as you see fit. I view it as something of a patch around the holes and missing pieces in the original story. However that may be, I hope the whole thing is enjoyable, and perhaps even thought-provoking.

-Shai Sachs, November 2020