I originally wrote this article on LinkedIn. While there is a host of articles that focus on mechanics and methods in Data Science and Data Analytics, I believe the foundations of information and its meaning are equally as important, if not more so. Philosophy has its place in the role of any science. This article touches on what I believe are important themes.
The notion of attachment has been understood and practiced for thousands of years, even before the first emergence of art in the age of Paleolithic man, and certainly long before smartphones, Facebook, Twitter and other variants of Social Media. Attachment to others is what has allowed human society to survive, prosper and transform through time, but not so much evolve. I know, quite a statement. That is my opinion.
I mention art because it is an important expression of the human experience, for it required the engineering of tools. With these tools to create art, the whole endeavor is an expression of the reality we experience but also can act as an expression of the ideal we wish to create. Throughout epochs of time, art has been used as a means to communicate via pictures how human societies should be ordered and structured for its survival and dominance. This requires a visual program that dictates how people should cooperate with one another to vanquish chaos, uncertainty and cope with the reality of death. In the modern world, we call this advertising.
In this age of digital communications, we too are still creating “art,” but it is fragmented across multiple mediums. It is now more difficult than ever to integrate them into a coherent picture of what it means to go through time, especially with others who share this mortal coil. The human experience requires interactions with others, and our internal measure of attachment dictates the choices we make, whether they are emotional, rational, economic, political or, dare I say, primal.
How do we know what a person is feeling or thinking via digital interactions without the “benefit” of explicit and discrete bits of data, encoded in all its forms (texts, posts, photographs, videos, transaction histories, user ratings, usage logs, surveys, etc.) that traverse through the Internet? Do we make inferences? Do we simply guess? What about intuition? How do we know these things without data in this realm? Certainly this set of questions seems rhetorical and runs the risk of the appearance of an idiot musing to one’s self. However, as I am typing this, I am not thinking of my audience as a set of algorithms that will perform sentiment analysis!
Even before the proliferation of the internet, the use of mathematics to model the reality of the world has been tremendously useful. Mathematics has been the Swiss army knife of the intellect, allowing us as a species to not only construct, but to refine models as to the nature of reality itself (and, I would argue, math is both a tool and an art). The Swiss army knife not only has improved, but the skill in which it has been used has improved as well. As this pertains to the realm of human emotion, behavior, and psychology, these tools have sought to quantify what was previously unquantifiable. All dimensions of the human experience were idealistically flattened onto the one-dimensional number line (think of this as the prerequisite for Factor Analysis and PCA). The hope was that if all were encoded into discrete chunks of data, the mystery of the future would reveal itself without having to actually experience it. All would be known before it actually transpired! The past and present could concretely solidify a chaotic realm of possibilities and their relative truth. Resource allocation would be a no-brainer. Trust in the methods would then generate trust itself.
However, a nihilist who has serious doubts about their own path of self-destruction may perform any of these analytic tasks extremely well. This is a good thing. Who really wants to be a nihilist, a variant of being a reductionist? Some of these people become professionals and we place value in their specializations, for they too are attempting to arrive at a reality that is cogent, useful and beneficial. Again, we tread in the realm of sales and advertising, part of the terrain of economics, which transmits data to its participants of the value of goods and services throughout the land.
This leads to what I believe is the reality when it comes to the nature of attachment: that we can only truly discern what a person feels by being in their presence, looking at their face. If you buy into the premise that the human face is the repository for the entire range of human emotion (go watch Al Pacino in Heat and tell me you are unable to figure out what he is feeling), it is in this mutual exchange of “looks” (not likes!) that we form meaningful bonds, the foundation of trust.
True attachment has a quasi-noetic quality to it, following from this premise. We can only put into words what we observe in the other by experiencing the reality of their being first. It is only from this point that cooperation and collaboration are possible, for without trust, one must function alone. The formation of positive, negative or ambivalent attachment finds itself in this universal root. Based on the polarity of attachment, it generates future social interactions, which either serve to erode or build trust. Language and data seek to refine the description of this experience of reality, a reality that all must experience, especially if you believe Aristotle. This is why I am particularly suspicious of AI and ML, because although there may be efficiency, there is no quality. It becomes a series of routines and algorithms to automate decision making. Don’t get me started on supervised versus unsupervised ML.
Now, making a huge leap, I invite you to fast forward to the present day, which should not be that radical a leap. Since this article revolves partially around the nature of Data Analytics, I, along with others, have been attempting to shed light on what I can only best describe as the illusion of certainty. How do we measure attachment, or more simply, trust, when it is encoded in disparate, discrete bits of data? What are the best predictors for this very human phenomenon? (Side note: I often chuckle that we have very precise formulas for how uncertain we think we are in a statistical model.)
Prior to the encoding of experience into discrete bits of data for use by computers, and then using algorithms to determine its meaning (sometimes its valence), art and language served this purpose in a powerful way. We could describe to the others the subjective experience of reality without that person having the benefit of the exact experience itself. We could use art and language to enhance the richness and totality of experience, to form ideas, concepts, and constructs, all for the purpose of sharing it with one another. Believe me, the irony of writing this article on a social media platform is not wasted on me. It is even possible that you, dear reader, may actually believe me, even if only just a little bit.
Due to the madness Social Media and eCommerce generates, even with its instant gratification platform, Big Data and Data Analytics are cohorts that seek to undo the quality of attachment because the picture it attempts to form is made the same way a puzzle is put together. This path of probable destruction is not entirely intentional, mind you. If you have ever bought a puzzle, you always knew what the puzzle would look like after you managed to put it all together. There was a certainty in knowing what the future would look like if you spent the time to arrange all the pieces, putting them in their proper configuration. However, the picture had to come first, not the other way around.
Another variant, a thought experiment if you will, would be this: someone gives you a pile of puzzle pieces without the benefit of the picture box in which it came. Even assume this is a gift, an invitation to arrange the pieces in the hopes that eventually a picture would emerge. One could be particularly cruel and give you a puzzle with no image at all. It would then just be an exercise in making sure the pieces all fit. How far do you go with such a task?
As it relates to Big Data, Data Analytics, and all its specializations, why do any of this puzzle assembly if we really believe we can figure out what is going on, without the benefit that face to face interactions provide? There is real utility in measuring through observation of the other, particularly our attachment to/with the other. Whether it is with our children, our spouses, our employees, our bosses or our customers, it forms the basis of how we make decisions about other kinds of utility (see social versus non-social rewards in attachment theory).
I believe the answer we are looking for in the realm of the discrete is the attempt to measure trust. How good we are at this endeavor depends heavily on our concept of what trust is and what it is for, or put more plainly its best use. If it is just for the purpose of manipulating the emotions, fanning the flames of desire and fear to force a call to action, I am of the opinion we are doing one another a huge disservice. While there may be efficiencies in the economy at large through the use of the internet, the price that must be paid is that we abandon what made us human in the first place: actually being with one another.
Can Data Analytics be useful? Of course. It can point to a reality, but it cannot create that reality. That is the ultimate problem. But analytics has its root in science, which I believe ultimately seeks to distribute useful knowledge for the greater good. Sometimes this endeavor has been abused (read Order Out Of Chaos by Prigogine).
If we are really in the business of providing value to our fellow man via this mechanism we affectionately call “the economy,” a product from the gifts of nature, stitched together with internet connections, modeled with data, analyzed with clever models, then there must be a return to actual experience, versus abandoning it and simulating it with a cheap replica of itself. What we will end up analyzing is the behavior of using the simulated version of reality. I am ultimately concerned that our interactions with others may see efficiencies through speed, but at the cost of quality.
If this is the case, all the Big Data and Data Analytics that could ever be engineered in all possible worlds will not realize the benefits it promises to generate. It will force all our interactions to be discretely quantifiable in some way so that we ultimately exist in a world that is engineered versus one that actually is. Analytics must paint a picture that, at best, points to that transcendent realm where Plato believed the good, the true and the beautiful reside. Only then can we actively participate in it with one another and share it in earnest.