Big Data: correlations and causality (criminal law aspect)

Abstract

Following up on previous studies of the Big Data phenomenon and its impact on the field of law, the article examines the advantages and disadvantages of using correlations which can compete with establishing causality in making legally significant decisions. The author draws attention to the fact that correlations based on Big Data help to analyse objects and phenomena not by clarifying the fundamental principles of their internal structure, but by revealing useful statistical patterns which may not be related to causality, but which are quite sufficient in a significant number of cases. It has been shown that such correlations can influence decisions made by a person, supplement arguments to justify decisions made by a person, or contradict decisions made by a person based on his or her knowledge and experience. Correlations can be useful in those areas of law where statistical analysis is effective, and their calculation is carried out on the basis of mathematical and statistical methods with greater speed and efficiency, as well as with lower costs than establishing causation. The author argues that since correlations are not conclusive evidence of causation, they should be used in conjunction with other types of evidence and appropriate legal argumentation. Correlations should not take precedence over other evidence or arguments. Attention is drawn to the danger of the dictatorship of Big Data, when data or correlations may be given more meaning and significance than anything else, when the fascination with correlations based on non-transparent algorithms can lead to critical abuses. The author proposes to provide safeguards to protect against the use of conclusions based on correlations for all persons who may, due to algorithmic and other shortcomings, be subjected to harassment, harm or other violation of their rights and freedoms, and to establish a hierarchy between causation and correlation based on the accuracy criterion as follows 1) correlation (lower accuracy, one of the prerequisites for establishing causation); 2) causation (higher accuracy, which may be based on correlation). The author argues that it is necessary to be prepared for the onset of the Big Data era on the fundamental principle of presumption of innocence at a time when the worldview based on the basis of finding out the cause may slowly give way to the worldview based on Big Data and correlations.