Subscribe

RSS Feeds

Archives

Segmentation and personalization for philosophers and scientists

The techniques used by today's marketing professionals, such as "customer segmentation" and "web-page personalization," would appear to be emblems of instrumental, exploitative communication. Today, we are so saturated with instrumental communication that Orwell's 1984 sounds benign in retrospect ("doublethink" feels quaint compared to the "multithink" we have now). At the point where nearly the entire public sphere is occupied by instrumentally deceptive signals, I am beginning to wonder if the tools of high-tech mass deception might not be amenable to a philosophical refactoring. If I'm wrong, the risk seems quite low, given that one can hardly make the status quo much worse in this regard. So the question is this: If certain techniques can systematically deceive so many sub-populations in purposeful ways, is there any good reason why these techniques cannot be used to undeceive sub-populations? Can we not run the machine in reverse? If the goal of philosophers and scientists is to discover and transmit the truth, but people respond differently to the same statements, it is rather odd that none of our great authors have yet thought to write multiple versions of one book to optimize its transmission among multiple audience segments. My proposal cuts across the grain of many humanistic intuitions about the nature of intellectual communication and authorship, but perhaps this explains why today the truth appears to be lightyears behind the false.

Segmentation and personalization

Customer segmentation refers to grouping potential customers according to the key dimensions that condition their decision-making and buying behavior. Key customer segments are usually based on variables related to demography, interests, geography, class, and personality traits including IQ. After segments have been identified, different communication strategies are deployed for different segments, to maximize the the probability of purchases from each segment. The segmentation and subsequent conditioning of communication on customer differences increases sales better than one blanket set of communications. Customer segmentation is arguably an ancient practice (all of the history here is just from Wikipedia, but it's pretty good); some marketing historians find Bronze Age traders engaging in geographic segmentation. The first use of customer segmentation based on systematically collected data appears to have been in the first decade of the twentieth century. The technique gets developed with greater sophistication from there, until the 1980s usher in what marketing historians call "hyper-segmentation." For the whole history of customer segmentation, it was always conducted at the level of groups. It is only then that technology presents the remarkable prospect of segmenting an audience at the individual level: marketing one thing to one individual, while marketing differently to a different individual, and so on. Today, in the digital context, hyper-segmentation can be done across millions of customers quickly and programmatically. Today, digital marketers even segment within individuals, by sending different messages at different times of day, or different parts of the year, etc. For the largest and most digitally sophisticated corporations, we must assume that AI systems are already deployed to not only identify any number of optimal hyper-segments, but also to update dynamically (with some lag) based on customers' changing attitudes and behaviors. Try to escape the model and the model will learn from your escape decisions.

Web-page personalization is one particular technique for the application of segment-conditioned communications. Personalization refers simply to the practice of delivering different web pages to visitors from different segments. This process leverages the data collected about users in their web browsers, to deliver website experiences that maximize whatever the website owner wants to maximize (typically sales, but not necessarily).

The ethics of communication

Such techniques are generally and correctly seen as sinister because they tend to be harmful deceptions: an agent promotes two different pictures of the world, for selfish and ulterior motives, using opaque methods. Through this contradictory presentation of the world, individuals are misled into two different and somewhat mutually exclusive maps of the world around them. The goal generally is to increase the income of the manipulative agent. Rigorous philosophical or scientific commitments, on the other hand, are devoted to seeking, and telling, the truth.

However, Western culture in 2018 is a crash space. Different subcultures now use the same words in radically different ways that appear, so far, irreversible and irreconcilable. Because our basic cognitive capacities — such as moral intuitions — evolved in low-tech contexts where the background environment was relatively constant, they're now overheated by a background environment in which unfathomable quantities of information churn at an accelerating rate. We cannot not live according to our evolved intuitions, but they are constantly being cued in contradictory and nonsensical ways. Our cognitive and behavioral circuits are hi-jacked by a super-intelligent system— namely the price system of a globally integrated marketplace — and there is no way not to think and do whatever one is cued to do by agents with more data and more intelligence than you. Doing otherwise would make no sense, for some supplier will always know what we truly want, even better than we do. Techniques such as segmentation and personalization are only unfoldings of this collective superintelligence. This is perhaps why marketing historian Wendell Smith called segmentation — in an ominous flight of abstraction rare for marketing historians — a "natural force" that would "not be denied."

But what if there were agents with good data and machine intelligence who sought not to maximize sales, but the effective transmission of true messages? They would accept the empirical reality of segmented human cognition and behavior. But then they would reverse-engineer the principles of historical, mass instrumental deception to produce a cypher capable of translating any one true statement into multiple different versions, each of which will be interpreted as true according to the subculturally segmented linguistic conventions.

I don't think I've ever heard any serious philosopher or scientist propose this idea, but I don't see why it couldn't work, and I don't see why it might be objectionable from an intellectual or ethical perspective. If my proposal sounds unsavory — "Serious intellectuals cannot employ the tools of vulgar digital marketers!" — perhaps this may explain why marketing professionals have such extraordinary influence over the intellectual and affective content of so many lives, while professional intellectuals appear to have less and less of it every passing year.

To drive home my quite abstract idea, I should give a concrete example. Perhaps you will guess, correctly, that I am interested in this prospect for personal reasons, having some experience with being misunderstood on the Internet. Fortunately, I have ample material for some thought experiments.

One of my motivations for going on YouTube is that I wanted to escape the confines of provincial paranoid leftism. But another motivation is that I would quite like to counteract the nastier dimensions of reactionary politics one finds on Youtube, for instance, white nationalism. So let's say I wanted to write a blog post explaining this rationale, why I think this is valuable and necessary work. As far as I know, this is truly a key part of my rationale, and I would like for as many people as possible to understand this, on the left and right. Perhaps because I personally believe it'd be good if others chose to do the same. In this sense, I am seeking to effectuate behaviors just like a marketer seeks to effectuate behaviors, but the crucial distinction is that my motivations are one with the explicit content of my message. My statement and the behavior I seek to effect are essentially the same thing, whereas commercial marketing is based on generating symbols that say one thing, for the purpose of doing something very different (make money), which is nowhere stated or implied in the outputted symbols. It's widely and correctly understood that people write messages in public because they want that message to be understood by others, to increase the probability of consequences that are themselves implied by the content of that message. The other distinction is that I only have one signal I want to be received by multiple people. You could say that marketers only have one true signal (the purchase), but the problem is that their different messages leave individuals with pictures or experiences that push them into different worlds. A philosopher or scientist would seek to increase the similarity and consistency of the receivers' different pictures of the world, aligned with what they believe is the true one.

Now, I could write a blog post entitled "I Am Going on YouTube to Escape Leftist Political Correctness and Mitigate the Fascist Right—And You Should, Too!" The problem is that this is a message for nobody. It is likely to go nowhere because the part that's critical toward the Right is defined by right-wingers as SJWism, and the part that's critical toward the Left is defined by the leftist individuals as racist dog-whistling. It's a true and pretty straightforward statement of my overarching rationale, and its chances of reproduction in the memetic ecology — in short, it's chances of living beyond day zero, or what it means for a message to even be communicated — is effectively nil. I wouldn't even click that, and I totally agree with it. It's so affectively empty that I could not muster the energy to even pretend that I "like" or "support" or "agree with" such a stupid, lame, obvious writer!

Now, imagine that you are in a physical room filled with leftists. Wouldn't it be perfectly normal, reasonable, and appropriate to use a different set of words to describe this mission, than if one was in a physical room filled with right-wingers? Of course it would. This conditionality of language is actually the essence of genuine communication; it is, must be, and should be as context-contingent as possible, in order to be true. If you think about whatever cases of speech that, in your opinion, are the greatest examples of truth revelation, I think you'll find they possess a kind of unique and mysterious element, a je ne sais quoi. And the reason you can't quite pin down the general feature that defines them is that they so effectively nailed the multidimensional context problem, that they were perhaps the best possible words you can personally imagine for communicating that message in that singular, contingent moment. Because the moment is singular and it's the nailing of that context that impresses you so forcefully, we experience its uniquely effective truthiness as an ineffable, non-generalizable feature. All of this is simply to point out that the truth value of any statement is actually a function of how well one communicates a particular signal in the form of contingently and instrumentally-selected, context-conditional symbols. Here the instrumental optimization is with respect to the objective of signal fidelity and noise minimization. That's a strategic, instrumental sub-goal to the final goal of being radically truthful and honest (a non-instrumental or substantive value or goal.)

If I say the same exact thing to these two different rooms of people, when the meaning of my words is fundamentally different to those two groups; that's not some kind of radical authenticity or commitment to the one whole truth. It's idiocy in the technical sense, devotion to a private language. In the words of Wittgenstein, forget about it. It's the apotheosis of delusional narcissism. And one of the reasons why so many people are feeling so insane right now is that smart people with a fairly balanced and independent view of the world are precisely those who are becoming less and less able to express themselves; these are the people who feel more than others that suddenly everything is escaping the grasp of human cognition.

Can customer segmentation and personalization techniques really offer a rigorous protocol for making objective truths equally sensible and transmittable to various pockets of social reality (what I have elsewhere called hard forks of reality)? Well, let's play out the example, and we can see how plausible it sounds. For the example I've been using, I could write a blog post making the one same argument, except the web page titles and the first page headers would be served differently depending on whether my free Google Tag Manager infers that the visitor is a left-winger or a right-winger (perhaps from some combination of other measured factors; maybe female millenials who recently visited the Democratic Party website get tagged as leftists, while white males in their twenties coming from Youtube get segmented as right-wing— whatever, this can be improved over time by testing the results). If a visitor is segmented as a leftist, the post might be entitled "Youtube Is a Nazi and I Am Punching It in the Face," which translates my mission into the exotic dialect that leftist opinion managers speak, allowing my breath to become living speech among leftists. If a visitor is segmented as a right-winger, the post might be entitled "Biggest Red Pill Ever (How to Trigger Every Snowflake)." I might use Google Optimize (also free), not only to trivially create my two different web page experiences, but to also give me a direct measure of the effects of the experiment. Through constant iteration, I will converge toward the two, true, optimally aligned translations.

Now mind you, in this example, the content of each blog post would be exactly the same, other than the titles. But in the future we might manipulate every single word, when we have such a sufficiently precise model and the necessary data to conduct accurate and systematic subcultural/ideological translations at such a high resolution.

A problem arises regarding whether this does not become instrumental manipulation for the ulterior motive of my own personal power. So far I've stipulated that, by the definition of the thought experiment, I'm maximizing a certain conceptualization of truth value. But in practice, especially if I am selling things on the side of my truth-maximization goal, bias seems doomed to creep in. This is an empirical problem that turns on having a defensible measure of truth value, that is not simply a proxy for "how many books I sell, because by definition my book is the truth." This is not a trivial problem, but the main reason we don't have such a measure and an easily implemented tool for it yet (Google Truth?) is simply that the non-instrumental communication of truths generally does not pay. In fact, it tends to do whatever is the opposite of pay. It's expensive to produce and it makes most people dislike you. Maximizing your own income helps you live, it wins you friends, and makes you happy (up to a point). Maximizing truth value makes it harder to live, it loses you friends, and it makes you tired with nothing real to show for it (most of the time). Is it really any wonder that we don't have fancy and free Google tools for customer de-segmentation, when segmentation is what makes money? It's almost evolutionarily impossible to imagine under contemporary capitalism. Although perhaps, as information processing power becomes so strong and so free and so available, then maybe, just maybe, like a few days before the singularity takeoff, a few hackers will find it easy enough to code up this kind of system.

One thing you could do is somehow measure each segment's picture of the world after reading the blog post, and see if they moved closer together. Maybe you could measure this with facial responses using their web cam or something, or ask them in exit surveys, or look at behavior later on. The degree to which they moved closer together would represent the translation consistency, at least. If the initial truth is actually an error, then you're screwed, and optimizing for this measure won't help matters. But this measure would be a start, for an objective criterion to maximize, separable from selfishness-biased variables such sales or click-through rates. If initial truth statements were somehow vetted, perhaps with reference to some larger objective database or something, then optimizing for the translation consistency would be a pretty good "performance indicator" for a philosopher or scientist blogger. I want to say this would represent instrumentally optimised, substantive (non-instrumental, i.e. honest) communication.

The two different blog posts may accent or emphasise different components of the one truth, but any particular communication item is always going to over- or under-emphasise partial aspects. This is simple, textbook, random error in any particular communication. In some sense, you could argue that segmenting and personalising the framing of a communication in this way, should increase the average accuracy of what one says overall, in the same way that increasing the sample size of a well-conducted survey will tend to push your sample stats closer to population values.

I suspect many humanists, philosophers, and social scientists may be discomfited by my thesis, but this is partially because most of them don't know how to use customer segmentation and personalisation techniques. It would be found simply ridiculous if the future of intellectual transmission might rely on tools that incumbent intellectuals find at once too vulgar and too difficult. For me, this indicates only one more exciting opportunity ripe for the taking by the next generation of truth-maximizing enterprises.

By the way, how did you like the title of this blog post?

Hard Forking Reality (Part 2): Communication and Complexity

There was once a time, even within living memory, in which interpersonal conflicts among strangers in liberal societies were sometimes solved by rational communication. By “rational,” I only mean deliberate attempts to arrive at some conscious, stable modus vivendi; purposeful communicative effort to tame the potentially explosive tendencies of incommensurate worldviews, using communal technologies such as the conciliatory handshake or the long talk over a drink, and other modern descendants of the ancestral campfire. Whenever the extreme environmental complexities of modern society can be reduced sufficiently, through the expensive and difficult work of genuine communication (and its behavioral conventions, e.g., good faith, charitable interpretations, the right to define words, the agreement to bracket secondary issues, etc.), it is possible for even modern strangers to maintain one shared source code over vast distances. If Benedict Anderson is correct, modern nationalism is a function of print technology; in our language, print technology expanded the potential geographical range for a vast number of people to operate on one shared code repository.

Let’s consider more carefully the equation of variables that make this kind of system possible. To simplify, let’s say the ability to solve a random conflict between two strangers is equal to their shared store of social capital (trust and already shared reference points) divided by the contextual complexity of their situation. The more trust and shared reference points you can presume to exist between you, the cheaper and easier it is to arrive at a negotiated, rational solution to any interpersonal problem. But the facilitating effect of these variables is relative to the number and intensity of the various uncertainties relevant to the context of the situation. If you and I know each other really well, and have a store of trust and shared worldview, we might be able to deal with nearly any conflict over a good one-hour talk (alcohol might be necessary). If we don’t have that social capital, maybe it would take 6 hours and 4 beers, for the exact same conflict situation. Given that the more pressing demands of life generally max-out our capacities, we might just never have 6 hours to spare for this purpose. In which case, we would simply part ways as vague enemies (exit instead of voice). Or, consider a case where we do have that social capital, but now we observe an increase in the numerator (complexity); to give only a few examples representative of postwar social change, perhaps the company I worked for my entire life just announced a series of layoffs, because some hardly comprehensible start-up is rapidly undermining the very premises of my once invincible corporation; or a bunch of new people just moved into the neighborhood, or I just bought a new machine that lets my peers observe what I say and do. All of these represent exogenous shocks of environmental complexity. What exactly are the pros and cons of saying or doing anything, who exactly is worth my time and who is not — these simple questions suddenly exceed our computational resources (although they will overheat some CPUs before other CPUs, an important point we return to below.) This complexity is a tax on the capacity for human beings to solve social problems through old-fashioned interpersonal communication (i.e. at all, without overt violence or the sublimated violence of manipulation, exploitation, etc.).

Notice also that old-fashioned rational dialogue is recursive in the sense that one dose increases the probability of another dose, which means small groups are able to bootstrap themselves into relative stability quite quickly (with a lot of talking). But it also means that when breakdown occurs, even great stores of social capital built over decades might very well collapse to zero in a few years. If something decreases the probability of direct interpersonal problem-solving by 10% at time t1, at time t2 the same exogenous shock might decrease that probability by 15%, cutting loose runaway dynamics of social disintegration.

It is possible that liberal modernity was a short-lived sweetspot in the rise of human technological power. In some times and places, increasing technological proficiency may enable rationally productive dialogue relative to a previous baseline of regular warfare and conflict. But at a certain threshold, all of these individually desirable institutional achievements enabled by rational dialogue constitute a catastrophically complex background environment. At a certain threshold, this complexity makes it strictly impossible for what we call Reality (implicitly shared and unified) to continue. For the overwhelming majority of 1-1 dialogues possible over the global or even national social graph, the soft-forking dynamics implicit in the maintenance of one shared source code become impossibly costly. Hard forks of reality are comparatively much cheaper, with extraordinary upside for early adopters, and they have never been so easy to maintain against exogenous shocks from the outside. Of course, the notion of hard-forking reality assumes a great human ability to engineer functional systems in the face of great global complexity — an assumption warranted only rarely in the human species, unfortunately.

Part 3 will explore in greater detail the cognitive conditionality of reality-forking dynamics.

On British Reservedness and American Boisterousness

The British are known be to reserved, and Americans boisterous, but I don’t think Americans communicate more in their higher volume of noises and gesticulations. If one could somehow measure the information content of interpersonal micro-gestures — all the nods, grunts, spoken comments people use to lubricate interactions with others in public spaces, I think on average British people would be found to communicate more. What is called their reservedness refers primarily to a lower volume of noise, but because of this a greater proportion of their emissions are received as signals. In an American cafe, if you accidentally cut someone in line, say, you might apologize with a hammed-up smile, to signal that it was a genuine mistake and you mean well toward the other; seeing your smile the other might wonder if that means you’re playing some kind of joke, and their uncertainty and insecurity triggers in them perhaps a vaguely cold glare before they correct it, with an equally vague smile at the end. You, in turn, are left wondering whether they got your signal, or if they nonetheless take you for an inconsiderate aggressor. Both parties leave the situation less clear about who exactly they just interacted with, and what exactly just happened. A British person making this kind of faux pas might mumble an awkward “sorry” while nervously looking at their shoes, and the other British person might say nothing at all, or grunt inaudibly so as to dismiss the situation as a non-event. The British situation might look like poorer or weaker communication, but really it’s more effective communication, and more proportional to the situation: with an expenditure of nearly zero effort, both parties walk away quite confident this meaningless misunderstanding meant nothing at all and that the other thinks nothing of it. The Americans did not generate more light, but much more heat.

Stay up to date on all my projects around the web. No spam, don't worry.

This site participates in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites. The Privacy Policy can be found here. The content of this website is licensed under a CREATIVE COMMONS ATTRIBUTION 4.0 INTERNATIONAL LICENSE.

rss-square
linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram