Even in science2.0, correlation is not the same as causation, and once we detect correlation, it is still very valuable to try and discover the causes. That is the essence of “Why the cloud cannot obscure the scientific method“, a reaction to “The End of Theory: The Data Deluge Makes the Scientific Method Obsolete“.
I concur. However, I think it is wort emphasizing two slightly more subtle points that often seem to get lost:
- Even when we just detect correlation, and we don’t understand the underlying causes, we can still do some smart engineering of tools. As an example, I don’t think that Amazon proclaims to understand my interests, but it can make pretty good suggestions, that are as useful to me as those that my personal librarian would make – if I could afford one.
(There is a version of the Turing test in this realm – as I mentioned before.)
- We now have way more opportunities for exciting discoveries of correlation, which presents us with nice new challenges for understanding common causes. That is why I fundamentally agree with Chris Anderson that “The Petabyte Age is different because more is different“, or, in my own words: “scale changes everything”
It does seem to me that science2.0 may be more fundamentally different than many seem to expect, not because causation doesn’t matter anymore, but because the data available for interpretation and the methods to do that interpretation have exploded and there are now many orders of magnitude more of them than most of us could even have imagined…