Chapter 5: Those Portrait Games Born of Love, Enmity and Ignorance
"You can have my donation if you stop preaching science to me." He looks ahead expressionlessly when their paths of commute converge outside the subway station the next morning.
"Great! That sounds like a Bayesian game. Should I accept one guaranteed payoff and move on or should I hang around to fish for an unlimited number of uncertain but not unpredictable rewards?" She walks briskly alongside him, tilting her sharp chin forward in determined hauteur.
"You can have another donation if you respect my right to be happy being unhappy." His black leather suitcase and her purple and white striped duffel bag swing like pendulums in the suited crowd.
"Uh-huh. Word games are a walk in a park with or without Tolstoy. Empirical testing will illuminate whether anyone's conception of happiness is right in all senses of the word." His black leather shoes and her ruby red lurex boots click down the pavement in perfect synchrony.
"Above all, whether you have my donation or not, you are not to initiate lip contact with me without my permission under any circumstances." He turns sharply towards her. "Has it never occurred to you that I may have a wife, a fiancée or a sweetheart? Have you never considered what that potential reality calls for? This is how pathetic general humanities education has become in this country. Even someone deemed to possess the intellect to qualify for a doctorate degree does not know how to examine issues from different angles, how to spot non-obvious areas of ambiguity, and how to truly tolerate and navigate uncertainty."
"Wait, wait. What kind of attached guy would have spent a weekend reading a book beside a lake all by himself?" She looks at him with a miffed expression.
"That's another assumption to question. Many kinds of romantic relationships are imaginable. It's not an absolute truth that everyone can be—or even wants to be—physically together with the person he holds dear on his day off. Moreover, even if you happen to be right, there's the problem of insulting a man's modesty. Want of intimate bodily contact only with those one is comfortable with and only in occasions of one's choice is not the preserve of women. Men are not cheap animals you have the right to exploit for physical pleasure at your whim," he retorts.
She is now clearly incredulous.
"Wow, just ... wow! This is the first time in my life someone's insinuating that I'm a pervert. What kind of a prudish background do you come from? And physical pleasure?" she fans herself and casts her eyes elsewhere. "Don't flatter yourself."
"Assumption again. 'Prudish' is a value-laden adjective," he glares at her. "It takes for granted that there are deliberate skin-to-skin interactions no reasonable person would find fault with. It treats individuality, personal sensitivity and religious diversity as non-existent. Maybe it's true that nothing anywhere close to carnal satisfaction was ever on your agenda, but any warm feeling your expression of gratitude was supposed to generate ironically served to gratify only yourself at the expense of the recipient. Don't impose your inelastic lifestyle beliefs, autocratic standards and unconsulted vision of reciprocity on others."
Their eyes meet again. Nearby, cars horn at surrounding traffic.
She thinks about the matter for a while. "Fine. I admit I was wrong. I apologize. Sincerely."
He stares at her silently with an inscrutable countenance.
"And now that you have my unreserved apology," she ventures. "Perhaps it's time you examine your own bias as well. You seem to be of the view that humanities alone can cultivate various cognitive skills. How do you expect me to stop harping about science to you when you're parading unfair judgments on the many professionals and thinkers who have put so much thought and effort in the field?"
Smiling brightly again, she walks on and continues, "Let's concentrate for now on handling uncertainty. STEM research, which many college seniors receive a taste of and which PhDs are all about, entails a thicket of unknowns. Scientists have no idea when and whether they will solve the puzzles on their hands. Ninety-nine out of a hundred experiments, it's lamented, do not work. Those that do at one time may flop at another. Each time one flops unexpectedly, we may well end up troubleshooting the entire setup relentlessly for a teeny clue on why. Yet in both success and failure, the real reasons are beyond us. How things work, learned researchers know, is ultimately a matter of interpretations, conjectures, more hypothesis testing and more interpretations. We are aware that so-called scientific truths are merely what best explain phenomena and work in applications at any moment. These theories are grounded in uncertainty and forever subject to possible revision."
"That otherwise well-read people think we think otherwise," she says with theatrical melancholy and a wink in her eyes, "reflects the sad state of general science education in the various countries from where pedagogical ideas and traditions have cross-pollinated to shape STEM curricula worldwide."
At the sandwich bar during lunch hour, her argument runs on as the waitress places corn tortillas and tuna salad pinwheels onto their stained wood dining table.
"Why, even if you doubt that the philosophy of science amounts to anything more than a blip on the radar of the average lab rat, you can't deny that warnings about the elusiveness of a perfectly representative sample, random variations in experimental conditions, and idealized presuppositions like the treatment of atoms as perfect spheres in molecular mechanics modeling have been drilled into every STEM practitioner since their neophyte days. In each piece of data analysis, we dutifully spell out margins of errors. In each journal manuscript, we lampoon our own work by openly sifting out issues like methodological imperfections and limited applicability of the experimental findings."
"And why, EVEN then, the game is not over! A torch and pitchfork crowd, you can imagine, potentially awaits us," she twists a four-tine steel fork menacingly between her fingers above the tortillas before swiftly stabbing into the heart of one piece. "whenever we submit the manuscript for editorial and peer review and present our work in seminars and conferences, all of which frequently expose us to hawk-eyed, snake-brained skeptics in the field with a keen need to out-survive us."
"In a nutshell, we stare into the deep abyss that is the vulnerability of human knowledge and capability all the time," she casts a dramatic faraway look at the dark water underneath the bridge on their evening walk.
He watches her impassively as she carries on expounding her argument the next morning while walking backwards before him along a row of office towers, "But as a whole, we triumph in the end, because the doggedly questioning system turns those human insecurities into fuel with the potential to bring mankind closer and closer to the truths of the universe, societal transformations and longer lives."
"Cognitive psychology buffs may claim that the Homo sapiens brain frequently finds it inordinately tough to apply skills developed in one domain to another, so adeptness at identifying and reacting to scientific uncertainty does not necessarily translate to any sort of proficiency at handling ambiguity in areas of human thought and activities like social affairs, politics, business and economics, all of which everyone partakes in in some way. You may even cite yesterday's me as evidence," she rues over a black-and-white floral-patterned cup of cappuccino at lunch. "But that's just one line of thought out of a million of mine, and I'm just one scientist out of a zillion on this globe. And guess what? We actually have this elegant technology that can factor in anyone's belief about the likelihood of a phenomenon and, as it acquires real-world data, revise the prediction. At the same time, the software can use such continuously updated probabilities in automated decision-making. In this manner, we get a form of artificial intelligence that, unlike deep-learning AI, can function on limited data and show us how it arrives at its decisions [1]. It all boils down to the same centuries-old theorem that powers analyses of Bayesian games."
Rising from her seat, she rolls up a piece of unused tissue napkin and uses it like blackboard chalk to scribble an equation onto the sandwich bar's glass window:
P(X|Y) = P(Y|X) x P(X) / P(Y)
"The vertical stroke means that what in front of the stroke is true given what goes behind, and of course P means probability of stuff inside the brackets. Remember that the probability of two independent events taking place is the mathematical product of the probability of each event? This theorem comes about because the probability that X and Y are both true when they depend on each other is the probability that one is true when the other is true multiplied by the probability that the other is true. We use an arch to represent a conjunction of two events."
She adds more lines to the window [2]:
P(X∩Y) = P(X|Y) x P(Y)
P(X∩Y) = P(Y|X) x P(X)
So P(X|Y) x P(Y) = P(Y|X) x P(X)
Hence, P(X|Y) = P(Y|X) x P(X) / P(Y)
"In our case, we have ..." she writes another equation:
P(Phenomenon|Data) = P(Data|Phenomenon) x P(Phenomenon) / P(Data)
Slender lines and graceful curls characterize her handwriting. Without a scent, it brings to mind the scent of Stargazer lilies. He almost does not notice the increased buzz some distance away from their table.
"P(Phenomenon) is what we call a 'prior probability' we assign according to our intuition, hypothesis or prior knowledge, which would mean, as more updating goes on, previously revised probability. P(Phenomenon|Data) is a 'posterior probability', the latest updated probability based on collected data. P(Data|Phenomenon) and P(Data) can be estimated by, say, measuring how often you get the data under tailored conditions or in selected cases where the phenomenon is true for sure and how often you get the data regardless of the phenomenon. Nevertheless, if you only want to compare the probabilities of different phenomena given the same data, you can ignore P(Data) and just see how different the size of the numerator P(Data|Phenomenon) x P(Phenomenon) is for the various phenomena. Still with me? Those are the bare basics."
"Imagine designing an online 'Do you know me?' quiz and inviting netizens to submit simple self-introductions from which respondents' judgments are made before they answer questions for themselves and for others. We can then dispatch Bayesian artificial intelligence—right, the name of this technology!—to measure how relatively perceptive humanities and STEM graduates are overall and when it comes to this or that trickier category of respondents that demands broader imagination to land a correct guess. In any case, quality humanities articles will be initially recommended to low STEM scorers and quality STEM articles will be initially recommended to low humanities scorers. The AI system will take note of clicks, track people's performance over time, experiment with different articles as necessary and learn for itself what kinds of articles work best at raising scores. Of course, when crunching the numbers, we'll have to control for covariables in test subjects' background like innate intelligence and family upbringing. Still, that's a fledgling step towards settling the question of humanities versus STEM for one type of ambiguity. You've got to at least give mathematicians and computer scientists credit for laying the foundation for this possibility!"
"Maybe as we run the test, we can brainstorm ways to mitigate sabotaging and other pitfalls and expand the approach to many other types of ambiguity. Any particular quiz you'd like to order?" she chats on merrily as she keys in her number into his phone and sends his to hers on the way back to his office. He looks straight ahead without a sound.
"Say, I've tremendously enjoyed exploring, dissecting and reconciling our divergent views! It's exhilarating to encounter an opinion opposite to mine, that I never imagined was possible," she returns his phone to him and stretches her arms towards the cloudy sky overhead. "Quite a pity you don't speak much. Let's have fierier, headier and messier debates in the long days ahead!"
"Oh, here's the venue of my seminar. See you next time!" She gives him a vivacious hand wave and rushes towards the place.
He stares in the chilly street as her loose hair bun and pastel blue coat disappear behind a revolving door. His office building is two streets and three traffic junctions away. He stares as she enters a transparent elevator at the exterior of the building, spots him and gives him another cheery wave as the elevator starts going up. A tire shop owner said that he will be visiting the company right after lunch to sign a fire insurance contract.
He dials the number. She looks down to see his incoming call, to her puzzlement.
"This may be all a game to you but—" Stricken wisps of lines break through his facade as he speaks into the phone and looks straight at her. "I am serious, heart and all."
Mouth slightly open in astonishment, she stares wide-eyed at his fast-vanishing figure below.
References:
1. Metz, C., 2017. AI is about to learn more like humans—with a little uncertainty, Wired, viewed 19 February 2018, https://www.wired.com/2017/02/ai-learn-like-humans-little-uncertainty/
2. Bessiere, P., Mazer, E., Ahuactzin, J.M. and Mekhnacha, K., 2013. Bayesian programming. CRC press. URL: https://books.google.com/books?id=4XtcAgAAQBAJ&pg=PA20&dq="the+conjunction+postulate+bayes+theorem"
