Seeing Data Science as Divination
Seeing Data Science as Divination
Why divination is actually closely related to Data Science
Since I was young, I have been interested in the literature of culture and mythology. One part of the study that interests me is the art of divination. For that reason, and during the time of confinement, I decided to take part in an online course provided by Harvard regarding divination and its system here.
From what I learned, every culture has its way of predicting what will happen in the future and how it could be accurate by the data the diviner collected. Yes, divination is closely related to data and data science. In this article, I will explain concisely and intuitively how divination could be associated with data science.
This article is just pure opinionated and for knowledge purposes, in no other means to degrade any long-standing practice that has existed before.
Divination Predictive System
According to Wikipedia, Divination is an attempt to gain insight into a question or situation through an occultic, standardized process or ritual. In a simpler term, it means knowing the future via the ritual process.
Divination itself could be classified into four broader categories depending on their input, including:
Random; Where inputs came from random processes, sometimes called “spontaneous.” For example, in the Roman Bird Augury, we need to wait for the birds to act spontaneously before making a prediction.
They were randomized, Where inputs are Human-initiated to produce a random outcome—for example, Rolling a dice.
Human; Where inputs that come directly from the diviner or input that cannot be interpreted by anyone other than the diviner; For example, possessed person.
Non-Random; Where inputs come from observations of any consistent, repeatable, predictable, and knowable process; For example, Measuring the position of the planets.
Somewhat the term used above is rather familiar in the Data Science field. There is an input, which depends on our inputs; the ritual type would also be different to produce the prediction.
This ritual could be imagined as a prediction system or “algorithm” for making a prediction. In this “algorithm,” we could say that if we input “A” to the “algorithm,” predict “B”; for example, if Jupiter is in position 273.87 degrees, predict the future romance life. Thou, just like any prediction model in Data Science, The “algorithm” is not that simple.
What is interesting in the divination system, the system itself is sometimes a bit unknown to the people outside; because there is a mysticism involved or the system itself is too complex or even because it lost in history. In Data Science predictive model, just imagine this as a Deep Learning model. It involves many statistical methods, the system is really complex, and after training, we don’t know what happens inside because of their stochastic nature. We just have the Deep Learning model, we input the observed data, and the prediction comes out. Divine!
Above is just a little forced example: Divination is similar to the predictive model. It could feel more natural if we use a step-by-step example of the system prediction.
Divination Framework
1. Observation
The first step in any divination or the data science field would always be the observation of the data. We fed this input into the “algorithm” to produce a prediction, just like when we fed data into our machine learning model. In theory, the more observation we have, the stronger the prediction would be.
2. Prediction
After observing the data, we have the prediction. While few predictions are rigid, like the position of the planets, many prediction systems incorporate ambiguity, intentional or otherwise, different interpretations. The clarity of the prediction thou is varied depending on the system.
In data science, take an example of cluster analysis in unsupervised learning; we would end up with cluster results that the system predicts should be clustered together. How we interpret the result would be up to us as Data scientists.
3. Evaluating Accuracy and Making Changes
Was the prediction correct? Like in any data science process, it is a simple question, but it is a much more complex problem to evaluate. It seems strange that divination needs to evaluate its prediction accuracy, but it is happening, although not always the case. Many of the divination systems themselves could be evolving to have better accuracy, but most of the time, we would not know about it as history is lost in time.
Take, for example, the Haruspicy from ancient Mesopotamian times. Haruspicy is the ritual where they sacrifice a sheep and extract the liver to gain a prediction. Since it is unlikely that the prediction ever yields 100% positive or negative results, a diviner might perform another ritual to reduce uncertainty. In this case, the diviner acts like a modern scientist by performing an additional experiment to acquire data.
Based on the method’s accuracy, some system changes could be made. Just like we data scientists playing with the hyperparameter, some changes in the system would be bound to happen. Although, as I mentioned before, most of the time, we would not know what is changed in the prediction system because it is lost in time.
Importance of divination concerning Data Science
I have explained to you in the passage above that divination and data science are, in some respect, similar. Many of the divination is based on the data. Tarot cards, astrology, roman bird augury, etc., are all based on the data and patterns.
Divination is seemed unscientific, but it is a human need. Divination tells us what will happen in the future, or it will interpret what happened in the past and then the present. What happens today is different from what it was yesterday, and it will be different from what it will be tomorrow; hence, the importance of the diviner in helping you understand the nature of the community, the nature of the society, the events that are taking place, and of course, how to sort of deal with those events.
In modern times, we would, of course, eliminate any opinion raised from this unscientific method. Prediction based on the dice or sheep liver would be considered bogus and not taken seriously. For that reason, we humans need more reason to believe that our prediction has a ground truth. Here we develop many scientific reasons just for assurance that what would happen is predictable in some senses. Data Scientist is just one of the roles that happen to do that. We try analysis and create a model based on the data to predict what would happen based on the inputs, just like divination.
It is different from the diviner that we consult for our personal daily life, but the need for Data scientists comes from the human curiosity that stems from long ago to know about what happens in the future. Thus, what we do as data scientists is not far from diviner.
Conclusion
This article is just my opinion and how I see divination concerning what I do as a Data Scientist. Our curiosity to know the pattern and what happens in the future as a human never changes; just the way how we try to have that knowledge is different.