Last year we wrote an editorial on defining the scientific method and were somewhat surprised to see that it was one of our most highly viewed pages. We are pleased that our readers share our interest in some of these less technical considerations as well.

Spurred by a Correspondence we’ve just published, this month’s editorial discusses the inference of causal relationships from purely observed data, without perturbation or prior knowledge to order the events. It seems that this is possible in some settings, but given the necessary assumptions, is it practical? And what may its implications be?

It’s worth noting that there is a large body of work dealing with causality that we could not even touch upon in this context. We welcome your contributions to the discussion.