Forward flow vs backward flow, using experimentation and curiosity to develop hypotheses rather than having a hypothesis and using that to design an experiment. The backwards flow is more likely to succumb to biases.