The Buddha, peace be unto him, is famous for declaring there’s no self. Strictly speaking, he denied the existence of an abiding, permanent self, especially the metaphysical Atman of Brahmanical Hinduism. We are born, we grow into adulthood and then we pass away. Some think we restart that process in the next life. The Buddha says: one life or many, there’s no rock to tether the ship of existence.
The Buddha left out space in his calculations. Sure, there’s no single self over time, but what about having the same self in space? Are we the same person in every direction?
Every one of us experiences ourselves from the inside-out. We refer to ourselves as “I.” It’s commonly believed that we have unique access to that self, an experience of being me that no one else has, that there’s an inner door to a secret chamber that can only be opened by one key. Who else can tell me that I am in pain besides myself?
But there’s another self (or many selves) of which I am only partially aware. That’s the self others see and experience. Why do we assume these two selves to be the same? When my daughter asks me not to be upset with her, and I reply that I am not upset at all, is it possible that both are right? Is it possible there’s a MeMe that’s fully transparent to me and a YouMe that’s fully transparent to others and the two aren’t the same Me’s?
It’s much more likely that the two are somewhat consistent but far from being identical. Which poses a problem for any autobiographical effort because a recounting of MeMe can’t pass off as a recounting of Me in general. The rich and the powerful have always had alternatives — they can hire people to write about their YouMe or even better, if they are famous enough, others want to write about them of their own volition.
The rest of us have to try hard to get others to talk to us for a few minutes, let alone writing praises. But even the most avid biographer doesn’t have the access to my daily routine. In fact, I am too absorbed or distracted to fully grasp what I am doing. The wake of my passage is invisible to me. Fortunately, that data is being scooped up by our friendly neighborhood tech giant. If my data across various websites, social media properties and calendars is aggregated and made available to an automated story generation system such as Narrative Science, I might receive a half decent autobiography in the mail every morning.
“Rajesh left home early yesterday morning. He caught the first train to South Station where he waited for the Acela for a full thirty minutes during which he flipped between his kindle and his phone. On the train he worked on the Acme report for the third time in so many days, changing most of the ten pages that he had written the day before.”
More suspense than my real life for sure. I might even pay for that service. But why stick to the real world. Why not probe lives I have never lived and don’t plan on doing so? Technology comes to the rescue once again. After all, most of my online explorations are funded by personalized ads trying to sell a future different me. The same as every advertisement in the history of marketing but personalization brings new opportunities to the creative autobiographer.
Paths not taken
Who does Facebook think I am?
In an attempt to understand myself through the eyes of Skynet, I have decided to take a screenshot of the first ad that Facebook inserts into my newsfeed every time I log in.
Hypothesis: If I take a screenshot every day for a hundred days I will learn more about who I am than a hundred years of Vipassana.
Just kidding, but I bet I will learn something. Don’t ask me what though, I am only on day 2.
Day 1: Today’s ad wants me to read like a CEO. Which is to say, not read at all but to get my staff to summarize it for me. Hey, at least I am better than Trump who doesn’t even read his summaries.
Sadly, I am going to pass. No $7 a month summary of business books for me. But the exercise frees up the imagination. Who is this CEO Rajesh? I’m thinking he wears a black suit everyday. Except for Saturday when he changes into a silk kurta to celebrate his pride for Mother India.
Day 2: Life is a roller coaster. Having rejected the offer to have summaries of business successes sent to my inbox, I must have missed a major opportunity while my competitors were making detailed notes. End result: I have been fired and my wife has left me.
Not to worry: DreamBuilder is here to rescue me from the jaws of failure.
I have a firm leg in the doom and gloom camp. My friends alternate between sending images of the burning Amazon and pictures of Amazon — the company, not the lungs of the planet — replacing all jobs with robots. Not that I mind; all that misery gives my optimist brain something to push against.
So when someone asks me if startups can save the world, my first response is: of course. Five seconds later, I change my mind to: you must be kidding! Back and forth, here’s how I argue with myself:
Me: How can startups change the world? They are tiny and the world’s problems are huge!
Me: Yes, but when they become bigger, they are no different from the other Death Star corporations sucking the life out of the planet.
Myself: but we can invent a new type of startup that’s less Death Star and more Jedi Knight. Startups don’t have to be about profits any more than factories have to manufacture cloth and nothing else.
and so it goes.
There’s something about how startups harness psychological energy and navigate uncertainty that appeals to me, and increasingly, we have data that tells us what works when people come together for a common purpose. Why not use it to make the world a better place?
To paraphrase a man who first gave me hope and then disappointed me: yes we can.
It’s never been easier to go from idea to implementation to turn a profit. Everything from Y-Combinator to my neighborhood angel investor are waiting to turn blood, sweat and tears into 💵. Unfortunately, saving the world is mostly not a matter of 💵. It’s about putting people and planet before profits. Here are three of my favorite world-savers:
Not a single businessman in that panorama. They were all politicians. That’s cuz politics is the most important method through which we have saved the world in the last two hundred years — both liberators and dictators have been politicians.
Note: none of them is a woman. I apologize for that snub to half the earth’s population. Patriarchy inserts itself into the STW business.
While the US doesn’t encourage political startups, i.e., new political parties, it’s common in other parts of the world for parties to be formed. Especially when there are classes of people whose needs aren’t being met by any of their current choices. Sometimes those new parties win elections and become political corporations and even one party states. Political monopolies suck even more than business monopolies.
The good news is there are more forms of political entrepreneurship than we can imagine, so why only political parties, why not other political startups? What about international political associations around global topics such as climate change — do we really think such wicked problems can be solved by middle aged women and men sitting in the U.N General Assembly? If you believe so, I have a couple of bridges I want to sell you.
I will tell you why like startups. They are the most robust institution we have devised for collective action amidst uncertainty. Corporations and governments are good at delivering solutions that work, but only startups are good at finding out what works in the first place. How will Indian farmers handle shorter, more intense monsoons? I don’t know, but I bet there’s a startup somewhere that will come up with a good idea.
Please stop thinking a startup is only about money.
We should divorce the idea of the startup from its capitalist origins, just as factories arose in the capitalist world but spread to societies where the government ran all forms of production. I am not saying one’s better than the other, just that startups and factories are flexible institutions capable of doing any number of different things.
I am a 100% certain that the future of the human species is bursting with uncertainty — pun intended. We have to become adept at navigating chaos and for that “Startup thinking” is an essential quality. Only if it combines politics, engineering and design though.
I went to an engineering school but I didn’t study engineering. In fact, I have stayed away from engineering my entire life despite being a geek. Cooking, yes. Sports, yes. Maker spaces: of course. But not engineering.
That’s because engineering struck me — perhaps wrongly — as being focused on the small picture, of making this widget in front of me work without caring about its connection to the wider world and damn the consequences. It wasn’t a discipline that encouraged a philosophical bent of mind. Engineering has undoubtedly changed the world, but it has done so without taking responsibility for that change.
In contrast, politics always struck me as being closely tied to philosophy: or to paraphrase Marx: we don’t want to study the world, but to change it. And there’s no shortage of political writing as to why we should change the world this way rather than that way. Politics, however disagreeable, takes responsibility for changing the world, which is why the metaphorical levers of politics were better suited to my theory of change than the mechanical levers of engineering.
Of late, I have been feeling that the division between politics and engineering is disappearing. Both are technologies that create human artifacts in response to individual and collective needs. Until recently, it was easier to encode big-picture goals in political technology than in engineering technology. For example:
Do you think democracies should have a separation of powers those creating policy and those enacting it? Solution: create separate institutions for the two purposes.
Today, similar decisions are being made in engineering technology: if you want to create a platform in which the platform owner doesn’t have an undue advantage vis-a-vis other participants? Make sure their business development wing only has access to the same data as any other business using the site. API design can encode ethical features and value judgments in a manner unthinkable fifty years ago.
The reason politics and engineering are coming together is code — and I use that computing phrase in the broadest sense of the term. Political technology has always been based on text: constitutions, policy briefs, white papers and such. Engineering technology has been based on things:steam engines and marble tiles. Code functions both as text and as thing. That’s a huge transition in how we change the world. We are just scratching the surface of that revolution.
That realization got me thinking about problems that should be solved simultaneously as engineering products and political policy, with solutions exhibiting a combination of good design, good data and deep concern for social implications. Technology that pays attention to the forest and the trees. I am on the lookout for such “forestrees.”
Here’s the first.
Halfway through the first game of the season on Saturday, my daughter took a soccer ball to the face. She continued playing for a couple more minutes before her nose started bleeding when she had to leave the field and couldn’t play the rest of the game. This being the United States, a trainer checked her out and suggested she might have the mildest of concussions, which meant no more games that weekend. Fortunately, she was fine the next morning and ended up playing on Sunday.
I love my daughter more than anything in the universe but this essay isn’t about her. It’s about how she was assessed for a potential concussion. She was checked by trainers three times on Saturday and Sunday. I noticed that on all three occasions, the trainer whipped out her phone and used the phone camera to make an assessment.
There are a couple of concussion assessment apps in the iOS app store but none of them are fancy — they are just a list of protocols to follow, including making the player stand on one foot, move their arms in set patterns and so on. Looks quite crude if you ask me, though arguably optimized for assessment by a young person with little experience.
I asked myself if we have better signals for concussion.
What about eye movements or other neuromuscular signatures? A quick google search lead me to this paper which says that disconjugate eye movements (i.e., when the two eyes don’t move in synchrony) are present in more than 90% of concussion and blast victims. I am not sure if trainers have the medical training to detect disconjugate eyes, especially if lighting conditions aren’t good. Disconjucation detection (DCD for short) might be too hard for untrained human beings.
But we are forgetting that camera. It seems underutilized — I saw the trainers shining it into my daughter’s eyes one eye at a time. DCD needs to process the signal from both eyes at once for it to work — after all, we want to find out if they moving in unison.
Let’s eliminate what I think of as the easy case — in the case of severe concussion, the two eyes are more likely to be completely out of sync. A severe concussion is likely to be a result of a major collision either on a sports field or in an automobile. Those aren’t the obvious cases I am thinking about, since they will be referred to emergency care right away. Concussions from a minor incident on a children’s playground or from an elderly person falling in a bathroom are harder problems to solve, and the solution has to be be in your pocket.
Phone is all we got.
An optical problem has to be solved first, a robust method for detecting eye movements from both eyes. There has to be a way of sweeping a phone camera in front of someone’s eyes so that it picks up the eye movement signals from both eyes at once. It’s a technical challenge because the signal is masked by an enormous amount of noise: jitter because of shaky hands, changing reflection patterns because of blinking eyes and head movements, changes in light sources if clouds block the sun and so on.
Fortunately, we have a clean separation of movements:
There’s the relatively smooth movement of my arm as I scan the camera in front of your eyes. Assuming that the light source from my phone camera is the only light that’s changing in intensity — ambient light from the sun or artificial lighting being assumed constant — the light reflected back from your eyes is going to be a smooth function of my hand movements. Further, smartphones now have motion sensors so we can use hardware to detect and filter out movements initiated by the person holding the camera.
There’s the jerky movement of your eyes as they saccade and change focus, and every time that happens, there’s a sharp change in light intensity. There’s are also the jerky movements of my eyes blinking, but that happens at a slower rate than eye movements and is an up and down movement while saccades have two degrees of freedom.
I am betting on a relatively clean separation of signal (eye movements) from noise (the camera movement, her head movements, blinks etc). In short, while there are genuine technical difficulties, I am reasonably confident that the signal detection problem can be solved. But once we have the two channel signal — one channel for each eye — we are left with an inference problem: how do I know when a signal indicates concussion?
The simplest kind of processing that can be done on the two channel signal is a summary statistic, such as the correlation between the two channels. Disconjugate eyes will have lower correlation between the two channels than normal ones. If we are happy with a simple diagnostic, this is all we need to do: set a concussion threshold and slot anyone who meets that threshold for medical intervention.
That, by the way, is the nature of most medical interventions based on bodily indicators. If I go to my doctor’s office with a test result and if my blood pressure, blood sugar or cholesterol is above a certain threshold, they will likely talk to me about further testing. If the statistic is in a band that’s not too low or or too high, they will talk to me about my diet and exercise regime and suggest changes if necessary. Otherwise, the machine’s working as it should and I go home happy.
But we can do better than that today, can’t we?
There are several problems with the simple diagnostic. Let me mention three:
It’s not personalized: My body might disconjunct at a lower contact threshold than yours. Even if there’s momentary disconjunction, your body might recover more quickly from it than mine. If disconjunction is a transient signal, how do we know when it’s a reliable indicator of concussion?
More generally, signs of concussion might be hidden in higher-order statistics instead of simple correlations. If so, noise will prevent us from extracting those higher-order statistics from a single observation.
The alternative is to go for a top-down approach based on extensive data collection. If I collect my eye movement data over time, the system will learn the typical conjunction between my eyes and how that changes with exertion, time of the day and other variables. With a robust data set like that in the background, we can be much more confident about when a genuine concussion is the cause of disjunction. Instead of creating a simple summary statistic and basing our diagnosis on just that alone, we can create a Bayesian concussion detector that answers the question:
How likely is it that X has a concussion based on the record of her eye-movements?
Detection accuracy will obviously improve if the system has access to eye movement data of thousands of soccer playing children. Having that data in the background will also help diagnose whether my child’s post-concussion recovery is on track.
Where is her disconjunction one week post concussion relative to the population average?
Should we be looking at a more intensive check-up?
Every trip to the emergency room costs money and leads to higher insurance premiums. We want to base any decision to send a child to an emergency room on the most robust data we have on hand. Longitudinal data is better than sporadic data.
Not that you need convincing, but there’s no shortage of advances in health and wellbeing that need repositories of biological data, from eye-movements to cholesterol levels. But, and there’s a HUGE but: the possibility of exploitation, control and oppression is so much greater when data are collected and made available to corporations and governments. In order to avoid big brother, platform design should encode a “fair use” policy with respect to all the data hosted on their premises.
To put in crude terms, whose data is it?
First: who will create such a data set and if I create that set, do I own it? Let’s start with the latter — that the creator of the data set is the owner, which is the current default. Since data is supposedly the new oil, it’s no surprise there’s a rush to capture as many valuable data sets in your hands as possible, leading to all kinds of problems. Search monopolies are bad enough, but we certainly don’t want health data monopolies.
Let’s say Startup A raises a ton of VC money and creates a comprehensive eye-movement database whose API used by Startup B for concussion detection and Startup C for dyslexia monitoring. Two years down the road, Startup A enters its own concussion detection app, competing directly with Startup B. What’s B to do? How does an application company compete with a platform company?
There might be a programmatic solution — as I had mentioned in the introduction, we can design APIs that prevent the business development side of Startup A from having access to data that the users (Startups B and C in this case) have access to. But can API modularity be enforced without regulation?
I doubt it.
Also, platforms keep evolving. Imagine that Startup A discovers that while the market for concussion and dyslexia apps is individual parents and teams, hospitals and HMOs are an ideal market for the platform as a whole. What does A do? Make an offer to B and C that they can’t refuse? Enter into a complicated revenue sharing model?
Platform monopolies are even more entrenched than widget monopolies — the dominance of the FAANG platforms being a case in point. Despite the popular slogan, data isn’t oil; it’s not a resource that disappears after being used once. Instead, it gets more valuable with time and accretes more uses.
Which makes data prone to platform monopolies since platforms are designed for current as well as future uses — once you list all the books in the world, you can sell them yourself, offer space for others to sell, convert them into ebooks sold by your company or direct the customer to a competing book that has a higher rating on your own system of ratings. I can’t see a future in which privatized data is good for anyone besides the monopolist.
How will that work out in healthcare? If we want to avoid monopolization, we should keep the data open, say through the creation of a platform commons. That leads to another challenge: who is going to pay for such a platform? It’s not like creating an open database of cat videos — the regulatory demands of collecting and storing biological data will make their platforms prohibitively expensive for your typical non-profit.
Is the only financially and politically viable solution is to socialize the data? Which is to say, governments pay for the creation and maintenance of health data repositories and own the platform. Governments having ultimate ownership has its own challenges, especially in countries where citizens don’t have political control over what happens to their data. Which, to be honest, is the case in most liberal democracies let alone authoritarian regimes.
Plus, what do you think are the chances of a government creating a high-quality platform? Might be possible in a small and rich country like Sweden, but the U.S health care debacle suggests that creating universal health data systems in a large and diverse nation is an incredibly hard problem to solve.
What is to be done?
I have a utopian answer: data should be a universal resource like time. Clocks became important after the industrial revolution and transcontinental railroads, but there were thousands of time-measures in the early days of mechanical time-keeping. As this article says,
When the Union Pacific and the Central Pacific Railroad formed the Pacific Railroad, later called the transcontinental railroad, more than 8,000 towns were using their own local time and over 53,000 miles of track had been laid across the United States. Railroad managers and supervisors well understood the problems caused by so many discrepancies in time keeping.
There could have been many ways of solving the problem of time standardization:
Let railroad mergers dictate time mergers so that at the end of the process there are a few private time companies in the world that own my time and your time.
Let the government own the time — and tax you for owning a watch😏
Create an international standard for time that no one owns or even thinks of owning.
Aren’t we glad we chose the third option — can you imagine time being owned by Acme corp instead of being an international standard? Why can’t we do that with data? Why can’t we have universal, secure data systems that aren’t owned by anyone and enable any number of products?
One international data standard that stores and maintains all the important data in the world. Cue photos of smiling children and dogs playing in the sunshine.
You may not agree with my solution — feel free to leave yours in the comments, but I hope you’re convinced that:
It’s hard to design platforms that serve our needs today and in the future.
We need both engineering and political technologies to design such platforms.
Now for the philosophical climax of today’s program…
I can’t end an essay about creating data driven systems without a nod to data dystopias. There are the obvious dangers: hackers stealing your medical records and blackmailing you, insurers refusing service because of genetic predispositions, governments denying treatment to political dissidents and so on. While they are important worries — it would be a disaster if a hacker changed your baseline heart rate during a cardiac treatment — but in my view they are problems of the past, based on the model of the “all-seeing eye” or the panopticon.
There’s a difference between the world of meager data and the world of rich data.
A world of meager data is one where I don’t know what’s going on in your head. We are atomic individuals separated by infinite mental space. The all seeing eye assumes a uniform space occupied by atomic souls who are mostly like each other. In that world, the panopticon appears either as a blessing or as a nightmare — blessing if you’re the religious type and like the idea of a divinity knowing every thought that crosses your mind, and a nightmare because you’re the cynical type that doesn’t want god or the government having access to your desires.
Note how both the blessings and the curses arise from the act of being “truly seen.”
In response, we created liberal democracies where the government knows some things about you but not too much, where insurance companies write policies based on the normal individual and where you have to be in prison or a totalitarian state to be completely exposed to the authorities. To summarize: meager data, health insurance and liberal democracy are a package that pleases many people much of the time.
I believe the world of rich data will be substantially different. It’s prime worry (or blessing!) will not be whether I am being truly seen, but whether there’s an I at all. There’s no reason why the Snapchat self and the iHealth self and the iVote self are the same self or even feel like being the same person. What if the experienced of a unified self is an artifact of history?
What if the reason you feel that you vote, you work and you play is because you live in a time when you have privileged access to your internal states — as Descartes famously thought — while others have limited and indirect access to those states?
A useful way of thinking about technology is as an extension of our mental apparatus. Glasses are extensions of our visual system, hearing aids of the auditory system and equations of our conceptual capacities. But none of these do much computing. Imagine instead a third prosthetic arm that has as much computing as your peripheral nervous system — what do you think it will do your experience of the world? Or, for that matter, when data platforms are as good at predicting what you will do next as you can.
In that world we might feel less like human beings of the past and more like Octopi with eight arms, each one of whom has a mind of its own. Those arms usually act in unison but they don’t have to. Sometimes they clearly don’t. I don’t know what it feels like when arm 7 and arm 3 go to war against each other and I have no way of stopping the fight. I certainly don’t know what it’s like when arm 6 votes for Trump and arm 1 votes for Hillary.
Let me leave you with a final question: what will it be like to build a global society around a multiple selved creature?
I am not sure if there’s a startup creating the Octopus empire, but there should be one.
Man is the measure of all things. So sayeth Protagoras, ancient scientist. If you’re the religious kind you might condemn Protagoras for idolatry, for only God has the measure of all things. Or if you’re William Blake, you might condemn Isaac Newton for succeeding at the task.
Before we get to Newton and Blake, let me make an important distinction between two kinds of measurements:
Objective measurements. These are measurements of entities out there in the real world, where despite the possibility of error, there’s an underlying quantity being measured. My height is an example of an objective quantity; you will measure my height wrong if you have the wrong tape measure and I might add a couple of inches to it while creating a profile on a dating site, but we can all agree that there’s such a quantity as my height.
Measurements of Exchange. Money is the best example. Let’s say I want to appear taller than I am and I go out to buy a pair of platform shoes. How much should you charge me? Should a man 5’4’’ pay the same amount of money to add 2’’ to his height as a 5’6′ tall man? If now, who should you charge more? Height’s objective, the increase in height is objective, but the money you charge for it isn’t objective. The measurement of exchange value is variable by design.
The measurement of objective quantities is closely tied to precision calculations and mechanization. I better measure the distance between my landing gear and the ground if I want my spaceship to land gently on the Moon’s surface instead of crashing into it. The flip side of precision is a dismissal of quantities that can’t be measured accurately.
Perhaps they don’t even exist!
In contrast, the measurement of exchangeable commodities is tied to assessments of value. Why does gold cost more than iron? Objective explanations only go so far. Is it scarcity? Not really, because my childhood drawings are scarcer and I bet you won’t pay any money for them. Is it because gold is hyper malleable and a good conductor? I am sure that plays a role, but advertisements have beautiful women wearing gold necklaces rather than highlighting the conductance of gold wires.
Exchange value can never be reduced to objective quantity.
That’s my reading of Blake, i.e., that measurement leaves out what’s most important about us. Perhaps, but there’s good reason to try measuring the most obdurate phenomena.
Most people living in modern societies work for money. How to value their labor? It’s a real challenge, for strictly speaking, we are trying to compare apples and oranges. Material inputs and labor go in and widgets come out. Labor is nothing like the widgets it produces, but yet there must be a way to turn widget numbers into wages for labor, for without that conversion, we have no way of keeping the factory going.
If our factory is a cooperative, we might say:
we produced X widgets that are sold at Y dollars each;
it cost us Z dollars to buy the inputs and maintain the equipment and we need to carry another W dollars for future investments, insurance etc.
Since there are A of us, we will each get (XY-Z-W)/A.
That seems relatively easy. But what if there’s one owner and A employees. How much should the owner get and how much should the employees? Should they be paid a fixed salary and let all the profits go to the owner? If so, why?
What’s the value of labor? What’s a fair wage? We don’t have unique answers to these questions because the measurement of exchange can’t be reduced to the measurement of objective quantities. However, we have to live with an uneasy merger of facts and values because the alternative is even worse. To understand why, let’s explore that age old question:
Can money buy me love?
One of the universal myths of the modern world is that subjective qualities, emotions in particular, are immeasurable in both senses of that term, i.e.,
there’s no objective way of getting to my feelings and
there’s no price to be put on them.
In fact, so powerful is the myth that love’s immeasurable that it sparked one of the most successful pricing campaigns in the history of modern advertising. Some relationships and feelings are beyond the reach of the accountant, but for everything else there’s:
The immeasurability of love reveals itself in all three sectors of human relationships:
Why does A fall in love with B? The myth comes with an answer: chemistry, “love at first sight.” Of course there’s something special about catching the eye of a person across a room and feeling a knot in your stomach when they look back with doubled energy. But who is likely to evoke that zing in the first place?
If you take romance novels as your guide, the answer is pretty clear: love on first sight is a lot easier if the other person is a born aristocrat with charming manners and the flawless skin that comes from a worry-less life. Money may not be able to buy love directly, but it sure tilts the scales in favor of the rich. In that, love is a lot like “merit,” where entrance to Ivy League schools is theoretically open to the deserving of all races and classes but in practice favors the graduates of Phillips Andover.
The romance of familial relations is equally suspect. A mother’s love is supposed to be infinite and unquantifiable but in practice it means that women labor long hours to keep a family going without compensation. How can you charge for the immeasurable?
Even friendship isn’t immune to the pressures of the market, for we treat friends differently based on how much money they have. There’s a reason why Drona was deeply offended when Drupada treated him like a servant. It’s much easier to raise money for my next startup if I am rich and my friends are rich and they know even richer people.
My point is that the lack of measurement often leads to injustices of value. Every parent of multiple children has been told at some point or the other that he loves child A more than child B. But what does morelove mean exactly? If I say I love my children equally and you (i.e., one of my children) say that I love Jimmy more than I love you, how exactly can we resolve this problem? There isn’t a final answer to this question, but we can all agree it’s unfair if I will 80% of my wealth to Jimmy and only 20% to you for no other reason than I love him more.
All of this would be moot if love simply can’t be measured, but this is where abstract philosophical and scientific questions about the theory of measurement meet changing technological resources.
Until recently, emotion measurement was a rare affair. I knew how you were feeling only when I saw you or heard about you from a common friend. Aggregate data didn’t exist — there was no way even the richest advertiser could have gauged the feelings of her customers on a daily basis.
All of that has changed dramatically. We reveal our emotional states to platform companies and governments several times a day, perhaps several hundred times a day. As a consequence, they have excellent models of our emotional state and wellbeing. Instagram and Snapchat probably knows when my daughter is going to have a fight with one of her friends even before she does.
That degree of access to emotions is clearly worth money and it’s reflected in the valuation of Facebook and other corporations. In fact, whether money buys love or not, it’s been able to buy hate at scale — and the electoral fortunes of Trump, Bolsonaro and Modi are testament to that success. The only way to counter that wave of hatred is if the measurement of love expands at a faster rate than the measurement of anger and if emotions more generally are made into a public resource rather than the property of private corporations.
If you’re Indian you know this already and if you aren’t you may not care, but in this note, I want to pay attention to a new innovation in the annals of violence: the Whatsapp driven lynchings of muslims accused of being rustlers of cows, though like every other innovation, it’s spread to lynchings of men and women accused of stealing children. We are having our Salem Witch Trial via social media moment.
TLDR; Why lynchings now?
My take is that they are a cheap and targeted way of sowing terror in the Muslim community. In fact, they are better at achieving that goal than the previous innovation in the annals of violence, i.e., riots.
In comparison with riots, lynchings are less expected, more sharable and cause more helplessness in the minds of the victims.
In other words, truly shameful and disgusting and evil but with a malevolent logic that suits the times we are in.
Yes, it’s the 21st century and it’s an abomination that lynchings are even happening. Of course they have to end and of course everything about them is about terrorizing Muslims. Especially when you read that the police are doing absolutely nothing to save the victims.
Nevertheless, there’s a gruesome logic to lynchings; one might argue that they are better than riots, for the latter kill people wholesale while the former are retail murders. They might also be more effective in achieving the political goals behind them. For that reason, they are the perfect terrorist acts. In fact, they might be the quintessential acts of terror in the age of social media.
Let’s answer that by asking what an act of terror must accomplish:
It must strike fear into the enemy.
It should cost you little.
It should make maximal use of media.
Now consider this headline:
What does this tell an Indian muslim? It says that your life is utterly unimportant (as a vegan, I am not inclined to say “your life is less important than an animal’s”) that you can be assaulted out of the blue and when help arrives in the form of the police, they might do nothing or side with the oppressor.
At least you can prepare for riots. But if you can be picked out of the blue, assaulted and your beating is broadcast on a thousand Whatsapp channels, you are being told that all resistance is futile and that you have no power whatsoever.
Now to speculate on their political objective, we have known for a while that the emasculation of Indian muslims serves the current dispensation. The message that’s being broadcast is that you can vote as an individual but you can’t vote as a muslim, i.e., your community’s interests don’t matter and we will actively discourage you from voting as a community.
How might one accomplish the emasculation of an entire community? By facilitating a form of learned helplessness and what better way to accomplish that than to encourage small but targeted acts of extreme violence so that you lose all will to articulate and defend your political goals.
If riots constitute an industrial form of violence, lynchings are the violence of the information age. Quite similar to drone attacks, which too track their victims from far away and the bomb arrives out of the skies. It’s the marriage of video games, predictive analytics, social media spectacle and communal hatred.
TL;DR: Planetarity == Solidarity with all beings on this planet. If that piques your interest, read on….
On January 1st, I took on the minor ambition of reimagining our planetary condition as my new year’s resolution. It’s best seen as a counter-revolutionary manifesto. Wait, what? You didn’t think we were living under a revolutionary regime, did you? You would be right in thinking so if you only heard our great leaders, but don’t pay attention to what they say and ask what they (and we) do and have been doing for the last two hundred years. Industrial society is a revolutionary regime headed by the Carbon Liberation Front, aka the People’s Republic of Exxon, Aramco and Gazprom.
Trotsky wanted us to live in a permanent communist revolution, but the liberation of the proletariat is nothing compared to the liberation of carbon, which is the one liberation theology that unites socialists, communists, capitalists, fascists and every other istist. Unfortunately, the carbon revolution is running out of gas (and steam). What comes next?
We are entering an era of existential politics, where the current obsessions of government such as taxes are going to be replaced by the elemental obsessions of air, water, food & climate. That thought entered my head about a year ago. A few months later, when I read David Wallace-Wells’ spectacular piece of climate pornography and learned that it was the most shared article in the history of New York Magazine, it struck me that I am not the only person in the world thinking about existential politics. Every hurricane, every fire and every drought increases the ranks of my fellow travelers. It’s only a matter of time when instead of worrying about Russian spying we will start worrying about oceans eating away our lands — we should be doing so already, but soon we will be forced to do so.
We aren’t used to inviting oceans into the senate but the survival of the human species depends on a politics that embraces the non-human world.
Don’t we know that already? Isn’t the interrelation of all beings the oldest Indian insight that’s been tweeted by every new age guru in the world? Yes, but there’s a new twist: the interrelation of all beings has left the concert grounds of Woodstock and is on its way to the halls of power and money. We are at the cusp of organizing a planetary liberation struggle.
I don’t need to tell you that a politics in which oceans and glaciers get a vote will be radically different from our current one. Existential politics will completely transform our idea of society; in fact, I think we will need to rework the basic categories through which we experience the social — history, freedom, production and most importantly, the category human.
What happens when the human bubble bursts. In any case, shouldn’t we be poking it until it does?
I am of two minds here.
Yin says: if humans are a geological force as the anthropocenic scientists claim, the natural is being assimilated into the social.
Yang responds: if human survival is at stake because of climate collapse, the social is being assimilated into the natural.
So what’s the right image: is the earth a giant factory or is the revenge of Gaia upon us? My gut says both images are true and my head adds that we can’t answer such questions systematically. In fact, we don’t even have the right terms in which to formulate an answer.
I consider that lack of categories to be a challenge rather than a problem: it’s exciting to imagine a radical expansion of the political to include earthworms and sheep along with blue collar laborers in Michigan and subsistence farmers in Madhya Pradesh. While we continue to write history as if it were that of humans alone (and until recently, of certain human cultures alone), the actual story of our times has always been more than human. While Earth-huggers have been talking about the intrinsic value of the non-human world for decades, now the expansion of the political sphere can be motivated on hard-edged grounds as well (see halls of power above). If you don’t believe me, consider that a few centuries ago, only kings were considered sovereign but now, throughout the world, we take for granted (in principle, if not in practice) that people are sovereign.
Why did that happen?
The transition from kingdoms to democracies is certainly a sign of moral progress, but it’s also a necessity — you simply can’t run a modern society along feudal lines: the changes in production, trade and consumption necessitated a new kind of society with an altered distribution of power. Similarly, the dramatic shifts we are seeing now necessitate a transition.
A transition to what?
Answer: to a planetary politics based on solidarity across species boundaries, a planetarity. Don’t ask me what planetarity is, I am not going to give away the season in the pilot 🙂 Instead, let me introduce the ABCs, the three planetary themes that I will be tracking throughout:
A for Animals: we can’t talk about the politics of the planet without talking about how we treat our fellow planetarians, which is to say, horrendously. Our treatment of animals, especially in factory farms, is easily the greatest moral failure of human society. On the flip side, expanding political rights to the nonhuman world is a key marker of planetarity.
B for Brains: if physical machines and the factories that housed them marked the transition from a feudal to the industrial society, then planetary politics will be marked by intelligent machines and the networks that house them along with their biological counterparts.
C for Climate: many of my friends like to think of climate change as a moral crisis — civilization as we know it is about to end! what are we going to do about it! — but my view is that the moral crisis lies elsewhere, i.e., in our treatment of the nonhuman world. Instead, climate change is a human crisis that points out the limits of the complexity that can be handled by our current socio-technical systems.
There are plenty of people who think of each one of these themes separately; animal rights activists, roboticists and climate scientists come to mind, but my goal is to juxtapose them.
Well, for one, because they are actually related; to give just one thread connecting the three, note that whatever machines and automation have done to human labor, they have completely destroyed animal labor, so if we want to understand what machines might do post intelligent automation, we might want to look at what mechanization did to animals. And of course, it’s the exhaust from these machines (including the methane coming from mechanized factory farms) that’s the underlying cause of climate change.
Second, each one of these themes adds a lens that illuminates the other two; for example, what if we look at climate change primarily from the point of view of the non-human world, might it start looking like a good thing? I am sure the end of human civilization will be applauded by the billions of pigs and chickens who live out their lives in crates the size of their bodies before they are slaughtered. Why shouldn’t we be taking their side? Extending political and moral rights to the non-human world can be justified on hard-edged grounds, a strategy the animal rights world can learn from the climate action world.
Third, drawing out the connections between these three helps us understand the planetary system in the Anthropocene, for it has many moving parts and no single theme can hope to capture the complexities of the system. In fact, the political embrace of the planet is the greatest complex system challenge of all time and should be of interest to scholars and thinkers for that reason alone. Just as the genesis of the modern nation state went hand in hand with the collection of statistics (and spurred much of its development), the planetarity of the future will go hand in hand with the development of big data and associated machine learning techniques.
If there’s one place where the three themes come together in an orgy of evil, it’s the modern factory farm: animals engineered to be automatons, living a life of ubiquitous surveillance and unchecked violence with the flatulence from all that misery warming the planet as a whole.
If there’s one place that planetarity has to destroy, it’s the factory farm.
PS: By the way, I am not talking about planetarity as a holist — no forest and tree metaphors were harmed in the production of this article. These three themes — they aren’t the only ones of course — aren’t like three blind men and a planetary elephant. There’s no seamless transition from one to the other. However, there is productive insight to be obtained by focusing on each theme individually, seeing where they cohere and where they clash with the other two and finally in noticing what lies beyond all three.
Or so I think. More accurately, or so I imagine, for what follows is as much speculation as analysis; after all, I am trying to peer past a veil that hides a mutation. I might imagine sensing the world through an earthworm’s skin or giving those earthworms a vote (of some kind); there’s more than a little bit of fiction science to be found here.
I wrote this a couple of years ago, but the lessons are as valid today as when I wrote these words.
One advantage of living in the Boston area is the chance encounter with smart people in a neighborhood coffee shop or PTA meeting as the case may be.
I had coffee today with a new friend who I met at a PTA meeting a week ago. He works for a Boston based non-profit that protects MA residents from snooping. Having spent the last seven years outside the US, I wasn’t sure why one needs a local organization for this purpose — isn’t the NSA a federal agency? Isn’t electronic snooping primarily a national and international affair?
It turns out, local agencies do a lot of warrant-less snooping in the name of counter-terrorism. For one, there are 80 “fusion” centers spread across the country where data from different agencies is aggregated and compiled. These centers are funded partly by the federal government, but most of the funding is from the state — about 70% if my memory serves me right.
If we know one thing in the post 9/11 era, no politician can refuse money that’s demanded for the cause of countering terror. If anything, you might get more money than you asked for. No politician can afford to question anti-terrorism funding in public.
Now consider this scenario: there are 80 fusion centers across the country and between them they have to chase a handful of serious terrorism cases. If I were to hazard a guess, I would say no more than a dozen serious plots a year. Just to be sure, let’s multiply that number by a hundred. Even then, a fusion center will handle 15 cases a year on average. In other words, exceptionally well funded organizations with lots of staff on payroll have an annual case load comparable to what a local police station handles every week.
What does an underemployed snoop do to justify his paycheck? How do they spend the millions? Simple: expand the notion of terror. The term “terrorism” is fuzzy, it’s not like assault or battery. When a term has a poorly defined legal basis to start with, it’s easy to include anyone and everyone you want to control. That includes the various Occupy X-ers , peace activists, hackers, people who sympathize with hackers, people who’re related to hackers. Anyone whose last name is Assange.
That’s how big brother works in democracies — not as conspiracy or paranoia but as justification for a line-item in the budget. You start with a small but terrifying threat, add a couple of boundary cases and before you know it the lovely old lady down the street who once brandished a placard in front of a federal courthouse is tracked 24/7 in the name of fusion.
Data is power. Data is also control. Anyone who conducts scientific experiments knows there’s a positive feedback loop between collecting data, controlling experimental variables in order to collect better data, and expanding the experiments to include new situations. The snoops are being good scientists. I believe someone once called the creeping rationalization of depravity the “banality of evil.” I prefer to call it the Murphy’s law of Villainy.
What was that old saw: in God we trust, everyone else bring data? Data and information are the bedrock of modern society. Money, numbers, bits; however you count the beads, it’s data everywhere.
Yet, there’s no real understanding of data among scientists and scholars, let alone the general public. Even the experts view information from within their specialization — let’s say machine learning or information visualization — than an understanding of the science as a whole. Imagine a world in which people learned numerical simulations for space travel without learning classical mechanics. Physics is a great science because it’s basic concepts — not it’s foundations, but the concepts that all physicists need to know in order to apply their methods to problems in the world — are drilled into physicists from mechanics 101 onward.
There are two sciences of information: computer science and statistics; both are backed by mathematical theory, but go well beyond mathematics in their real world applicability. Still, there’s a tendency to identify these subjects with their (current) mathematical foundations, i.e., the theory of computation and probability theory. A physicist would find that strange; physics is mathematical, but no physicist would confuse the foundations of physics with the foundations of mathematics.
Until our understanding of information makes that transition, we won’t have a robust science of form. I believe that transition will require a deeper unification of computing and statistics than is on offer today and in order to do so, we will have to look at the two disciplines from a bird’s eye view first and then narrow down on important questions for unification. It’s a topic that’s beginning to concern me more and more, so I am going to use these newsletters to talk about my ideas every so often. Bear with me if you think I am going all technical.
Let’s first note that computing and statistics bite different chunks of the information universe. Computing helps us engineer information systems — desktop, laptop and mobile computers and computer networks being the most important. Computing (and once again, let me emphasize that I care more about computer engineering than computer science) integrates information vertically, i.e., it’s about engineering information systems from logic gates all the way to iPhone apps.
Statistics on the other hand helps us with experimentation, getting data from the world. The integration is horizontal; statisticians care about experimental designs and survey techniques; as the data is brought in for analysis, statisticians also care about techniques for crunching and visualizing the numbers.
Computing and statistics have stayed away from each other for most of their history, starting with training and ending with their typical applications. Statisticians learn continuous mathematics and most of the important applications of statistics have been in unsexy fields such as agricultural genetics and psychology. Computer scientists learn discrete mathematics and from the beginning the science and engineering has been very sexy — from it’s involvement in code breaking to the foundations of mathematics.
The proliferation of data is the main reason the two fields are beginning to come together. In particular, we need the vertical engineering of computing systems to be driven by the horizontal flow of data. Incidentally, this is exactly what my PhD supervisor, Whitman Richards, was advocating several decades ago. He got the germ of that idea from David Marr’s work on Vision. The marriage of the vertical and the horizontal is not only interesting as engineering, it’s arguably the best way to understanding the relationship between the mind and the brain as well. Machine learning is at the forefront of the marriage of vertical information and horizontal information. I believe that merger will expand to more and more fields in the future. To be continued.
Everything you can do, you can do better with data
That’s the new nerdism. We think our obsession with data is brand new, but data predates big data by a huge margin. In fact, data and bureaucracy go together. Ever since someone notched a stick or chiseled a stone, we have been collecting data. I might even speculate that writing has been as much about inscribing the written numeral as it has been about inscribing the written word.
Representing data is easier in some respects than representing natural language. For every idiot who scribbles a curse word in the mens room, there’s another idiot who has marked the wall with counting sticks. Data is as primal as language. For the same reason, it’s no less a human creation than language. We have been brainwashed into thinking that data is an accurate representation of the world while the written word is soft and subjective. I disagree: both are human representations of human realities. It’s true that the written word is associated with storytelling while data is associated with science, but they are both symbolic systems. There’s no such thing as pure data.
Data is one of three ways of inscribing the world — the other two being the written word and money — that converts flows into stocks. In fact, that’s the primary function of inscription — it helps us keep track and takes pressure off fallible memories. Inscription also runs the risk of fundamentalism. We are all aware of fundamentalist religion, which one might call the fundamentalism of the written word. We are increasingly aware of market fundamentalism, which is the fundamentalism of money. We aren’t all that aware of data fundamentalism.
There’s a minor version of data fundamentalism that’s been around for a while. It goes by the name of scientism or materialism, depending on whom you ask, but I say it’s minor because it hasn’t affected political or policy realities much. My scientist friends complain about their lack of influence in political circles and the lack of rational thinking in said circles. Social engineering has had a bad rap for the most part.
That’s about to change, if it hasn’t already. Data driven policy is the new rage. From Obama (and Modi?) winning elections with better data to nudges that make us more virtuous citizens, we are entering an era of data driven societies. The new data fundamentalism threatens to have much more influence on policy makers. As the India Against Corruption protests, the various Occupy protests, the Climate Change protests and the Ferguson protests have shown, there’s a huge gap between the future people desire and the heavy handed response to those who complain. In such a situation, there are two ways of using data:
1. Minority report style nipping dissent before it happens and military style state violence after it does. This is what's happening (or so the promise goes) in fusion centers. 2. Uncovering people's genuine needs at a fine-grained level and creating customized policies for addressing those needs.
The first is the natural instinct of both the law and order conservative and the best and brightest technocrat. It’s a low trust attitude that feeds back into a low trust response from citizens. The second is what we really need. It’s not a purely technical solution; it’s a political one. Politics can and should work hand in hand with technology. Two hundred years ago, mathematicians such as Condorcet created the voting systems that we now take for granted; so much so that we don’t think of them as technology at all. Why can’t we do that again, except with real time data instead of a vote every five years? I think there’s a great future for data driven approaches to government, but it has to start with data driven citizenship, not data driven services. The government has to be of, for and by the people. Data should only serve to strengthen that claim.
This Week’s Links
Ian Hacking’s book on chance. The best explanation I know of the intimate relationship between data and modernity.
I think it was Nehru who used the term “unity in diversity” in his book Discovery of Indiato describe India and it’s culture. Then, as now, India was an incredibly diverse country. No single language, religion, ethnicity or political persuasion unites the country as a whole; and yet it moves. It’s the opposite of the European nation state built around a common language and religion. India’s unity amidst diversity has always been contested — there are people who believe that India should be Hindu (though it’s not clear how that would solve the problem of diversity) or India should be a Hindi speaking country. Some people will always try to impose their monolithic vision on to India, but I believe it’s impossible to succeed. No centrifugal force triumphs in India for too long. I am gladdened by the irreducible diversity of India. It’s a fluid existence of identities that have multiple loyalties. Loyalties that refuse to be suborned to an overarching identity.
I have been thinking about the implications of fluid, mostly peaceful (and sometimes incredibly violent) coexistence for domains such as philosophy that are far removed from the passions of nationhood. I believe that philosophy, design and the arts should come closer than they’re now. Epistemology should be replaced by epistry.
Technology is rapidly changing the landscape of knowledge. We no longer hit the library for information. We might soon be going to the web for education as well — not that I expect physical learning spaces to disappear, but their role will change and new blended learning environments will emerge. Some of those environments will be dystopic, but others will help us imagine a better future.
This Week’s Links
Fluidity is a central design challenge: how can we create knowledge environments that let data and information stream through as we adapt to new situations. Fluidity doesn’t live in the abstract, it needs context and circumstance. This week’s links have something to do with fluidity of the phenomenon they seek to describe.
The New York Times has a blog on contemporary philosophical matters called The Stone. Gary Gutting has a wonderful interview with Jonardon Ganeri, a well known Indian Philosopher (i.e., a philosopher who works on Indian philosophical texts rather than a philosopher of Indian origin) where Jonardon artfully parries questions about Hinduism. It’s a great read.
Moving from philosophy to technology, I am sure many of you read this news report about MIT researchers recovering audio information from visual information, of figuring out what something sounds like from knowing what it looks like. You can imagine the possibilities for surveillance. Visual surveillance is relatively easy if you’re out in the open. On a clear day, you can be recorded from a satellite in orbit. Audio surveillance is hard. You need to be pretty close to the target. If speech is recoverable from sight, video becomes audio and you have no privacy in public. Your word really would be your bond, for someone could mine a public video of you making a promise and hold you to your word. I find that cool as well as troubling, but my real fascination isn’t with the technology. It’s with the idea that information itself is distributed across sensory domains. Think about this way: it feels very different to hear something than to see it. Red is a color, not a sound. Yet, we know that objects have shape and make a particular sound. We still don’t know what binds sight to sound. Perhaps information is the glue. That will be a huge advance in our understanding of mind and consciousness.
While some people believe that we’ll be uploading our minds to google’s computers soon, other’s are not so sure about these reductive approaches to the mind. In what I call “the reduction of all reductions,” scientists are measuring brain activity during meditation and trying to arrive at a science of enlightenment.Read this interview with Evan Thompson to understand why that’s such a bad idea.
Mindfulness is one of our modern mantras. Stressed? Meditate. Want a better job? Meditate even more. Meditation is fast becoming another commodity. There’s even an app for it. So is big data. Where an earlier generation fought for our rights, we seem to believe that social challenges can be solved, like a calculus 101 problem. I am of two minds about this. I do think data will help us address social challenges. At the same time, data isn’t a substitute for democracy.
Finally, while some of us binging on twitter and facebook, others are OD’ing on MOOCs. Why take six courses a semester when you can take sixty? Is that really a productive way to learn? Would you hire a person who took a degree’s worth of MOOCs in one year and aced all those courses?
That’s it for this week. As usual, your comments are welcome. If you have any suggestions for great links, let me know.
If you like this post and want to receive others like this in your email, sign up for the weekly newsletter
I take your privacy very seriously. Your email will remain private
Surveillance is ubiquitous these days; in fact, a Soviet KGB handler would be shocked how easy it is to find out intimate details of a mark’s life. There’s no need to threaten or coerce either, though that doeshappen sometimes.
Information is central to control. We have known that for ever. Spies are the world’s second oldest profession. Much bureaucracy is about information. All those land and birth records, tax forms and license permits; what are they besides data?
The logic of data is not that different from that of the PC revolution. The totalitarian state is much like a mainframe: huge, expensive, hard to maintain and always running into one bug or the other. Further, the mainframe provider isn’t interested in improving the technology. Instead, they use the coercive power of the state to keep competition under control. A totalitarian system is an information monopoly.
Unfortunately, that monopoly doesn’t like competition, so it doesn’t allow other purveyors of information to set up shop. No google in China. It also means that frivolous collectors of information don’t find fertile ground — matchmaking sites, hook-up apps, stock prediction markets, none of these can exist as autonomous entities. Of course the actual reality of China is different, but that’s because the state has a fine grained view of information that it cares about and information that it doesn’t. Maoist China wouldn’t have allowed that subtlety.
In comparison, contemporary data collection is like the PC. It’s fast, cheap (and almost certainly) out of control. It’s not big brother and it’s all the more efficient and powerful for it. Once data collection becomes fast and cheap, we can use it for the every frivolous or creepy purpose that comes to mind. Want to know how much you’re eating: there’s an app for that. Want to know where your children are tonight: there’s an app for that too.
Perhaps even more importantly, it makes surveillance a universal reflex. Are you worried that your child is safe? Install a camera at home and inspect the baby-sitter. Afraid that terrorists are coming in? Put a camera on every inch of every border. Worried that your workers are talking too much at the water cooler: put electronic tags on them and measure what they do.
Cheap data makes it possible to universalize information gathering to every sphere of our lives, from our dreams to daily workouts to defense contracts. That universalization, strangely, makes surveillance less ominous. You’ll probably freak out if you knew someone was tracking whom you meet, where you eat and where you sleep. But if that someone was also tracking every yawn and every burp and every little tic in your left eye, you might think it’s too banal to worry about. Arendt might have something to say about our current situation.
Decentralization is Key
Every transaction once conducted on faith is now being replaced by surveillance, and we are all doing it. This new form of spying is worse in the long term than the older kind because it’s more likely to pass off as the natural, decent thing to do. It hits all the right registers: it’s decentralized, it’s voluntary and open. It also changes us from the inside, so that we don’t need to be told to snoop on our neighbors. The logical outcome is an atomized society glued together by data rather than trust. We are like the frogs in the data well, smiling as the surveillance meter rises; but at some point it’s going to boil and we will be cooked.