Data Divination: Conceptual Background

Data Mining seeks patterns in data to better predict future events in the domain under analysis. It is a process which distills the human experience of reality into atomisable data that can be related and parsed for patterns. These patterns hold sway over our lives.

The revalation of patterns as an organising force has a distinguished history in human practice. Interpreting patterns from our environment via ritualised settings has guided decision making from quotidian levels to upper echelons of government. Collectively these human mediated, pattern predicting practices are known as divination.

This project explores the overlap and dissonance between different systems of patterns attaining significance. It does so by constructing databases of how both the human and the machine apprehend tasseomancy (divination by tea leaves)

explanation of the projects genesis at METAL Residency


There are two important threads at work here: the first concerns the combination of machine learning methods and statistical analysis that accompanies ‘knowledge discovery in databases’ (more commonly known as data mining). It is by these means, and some occasional contextual or expert knowledge (as provided by human’s priveleged enough to be included in the knowledge loop) that patterns in large datasets are unearthed, patterns which are equated with the discovery of novel patterns of information and hence ‘new knowledge’.

The data mining machine, through the methods detailed above, looks for patterns in our digital traces. With enough data patterns can be discerned that humans alone would not be able to detect by themselves or in concord with one another.

From the get go of my research in this area this idea struck me as off. The idea that masses of data can be productive if only are arranged in the correct architecture and with the correct combination of KDD algorithms prompted me to think of data mining in terms of knowledge alchemy.

Aside from alchemical practices there were other esoteric practices I felt were relevant to bring into proximity to this newly emergent oracle in our lives. There is a long precedent of human’s divulging agency to objects in order that events of the future might become more palpable and apprehendable to them. Collectively such practices are known as divination (distinct from divining): a collection of practices by which patterns in seemingly random objects or events can determine what is yet to come.

Aside from the surface level parallel of pattern seeking systems between divination and data mining I feel there is a deeper element to be explored. Many of these practices operate at a surface level: fortune telling. But the advanced practitioners in all areas will encourage a level of divining known as scrying. Scrying is less prediction based (it doesn’t rely on the codification typically ingrained in tea leaf pattern dictionaries), instead it is a revelatory mode of operation. The object becomes an enabler for your intuitive knowledge to reveal something to you. And I think finding ways to contrast human intuitive knowledge against the juggernaut number crunching pattern revealing prowess of data mining is an essential endeavour

But you have to remember that the machine is looking for patterns: the patterns found influence our life.