Lecture at Gesellschaft für Wissenschafts- und Technikforschung (Society for Science and Technology Studies), annual conference
The model distinguishes between what is data and what is non-data. It is a reference to reality, much more than the algorithm.
Instead of focusing on »the algorithm«, the trinity of Modell-Data-Algorithm should be taken into account, since the algorithm alone is an indication of nothing, while data is a reference to real world events, and the model is the decision about what get’s included and what get’s excluded from computation.
Prediction from this perspective is statistical correlation. The lecture further demonstrated a few exsamples, how prediction can be tricked by introducing smut or fake data, which is beyond the scope of the model.
Further Info at: http://gwtf.de/ (In German)