Big Data is using the principle of correlation over the principle of causation. This principle is used in data informed disciples and is going to change our life. Big Data uses mathematical proves and has unleashed a wealth of powerful methods that combines knowledge with data. But correlation cannot be causation. A rooster’s crow is highly correlated with the rising of the sun but it does not cause the sun to rise. Data likewise is dumb. It can tells us that the number of people who took the medicine were cured but it cannot tell us why. May be those who took the medicine did so because they had money and might have recovered just as fast without it. Mere data , therefore, is enough. Yet we can see how Big Data enthusiast continue to chase data intelligence. But it is not like that these data scientists believe in voodoo. Judea Pearl says that we have entered what he calls Causal Revolution.

The Causal Revolution did not happened in a vacuum. It has a mathematical secret behind it which can be best described as the calculus of causation which answers some of the hardest question that we have asked about cause and effect. Pearl says that the calculus of causation consists of two languages: casual diagrams to express what we know and symbolic language, resembling algebra to express what we want to know. Causal diagrams are simply dots-and-arrow pictures that indicate our existing scientific knowledge. The dots stand for quantities of interest called variables and arrows represent known or suspected causal relations between those variables. May be we could say we which to find which variable listen to which others. These diagrams are like the maps of one way street and are extremely easy to understand. Data scientists use other models than the causal diagrams too.
Side by side with the diagrammatic ‘ language of knowledge’ we have the symbolic ‘ language of queries’ to express question to which we look for answers. For instance, if we are interested in the effect of drug (D) ON lifespan (L) , then our query can be symbolically written as : P(L |do (D)). It says: what is the probability of P that a typical Patient would survive L years if made to take the drug D. In some cases we may wish to compare P(L |do (D)) with P(L | do (not-D)). This refers to the patients that were denied treatment with the drug D. Thus, we are doing with intervention operators and not simply passive observations. This means we observe intervention operator do (D) and see what impact it has on patient when he/she is given the drug (D). Mathematically, then, we write the observed frequency of lifespan L among patients who voluntarily take the drug as P(L|D). This expression stands for the probability (P) of Lifespan L conditional on seeing the patient take the drug (D). We have to note that P(L|D) may be totally different from P(L |do (D)). This difference is between seeing and doing. Seeing may not lead to the doing. This why we do not regard the falling barometer as the cause of the coming storm. Seeing the fall of the barometer gives us an indication of probability the coming of the storm. If we force the fall of the barometer, it will therefore, not cause the probability of the coming of the storm.

The Causal Revolution deal with the do operators. This enables us how to predict the effects of an intervention ( do operator) without actually enacting it. As with predicting the interventions, in some cases we may do what we call retrospective thinking with an algorithm that takes what we know about the observed world and produce an answer about the counterfactual world. this algorithmization of the counterfactual is another gem by the Causal Revolution. Continuing with our above example: suppose Joe took the drug and died in a month. The question of our interest in this instance is whether the drug has caused his death. To get the answer we need to construct of counterfactual scenario wherein Joe was about to take the drug but changed his mind. Therefore we may ask: would he have lived? Counterfactual reasoning deal with what-ifs and appear to be hypothetical and our empirical observation cannot confirm nor refute the answers to such questions. Yet we do admit such thinking as reasonable. Take the counterfactual like the rooster does not crow one morning, it does not keep the sun from rising. Even in that scenario the sun would have risen. Algorithmization of the counterfactuals thus, enable the machines to participate and excel in this unique human way of thinking about the world.

There are even more complex models of causality , like the inference engine. What we have discovered in this context is rather simple. But it gives us an insight in the black box of data analytics. What we know ad dam data is raw material for us to causal models and from raw data into usable knowledge. We use the observation, intervention and counterfactuals in the ladder of causation. What we do with algorithimization is automatize human level intelligence. We call it Artificial Intelligence. This is why we can communicate to machines and machine can communicate with us. We are smarter than our data. Data does not understand causes and effects. We humans do. When we let machine participate in this understanding , we are doing what we call data analytics. Thus, what data analytics is doing can reveal what we are. It can reveal an anthropology. We have amplified our abilities in the machines so that we can better understand the data both big or small. Data analytics done using the computation power is simply the power mind over data.