Popper's principle work was in the philosophy of science. He championed a definition of scientific hypotheses which is something like: A hypothesis is scientific if and only if it is falsifiable. That we can never prove a scientific hypothesis, merely not yet disprove it.
Previously, the primary view was one of induction: that experiments give inductive evidence for a hypothesis. That is, if I drop a bowling ball and it falls at 9.81 M / s^ 2, the Popperian view is that this is evidence that does not disprove Newtonian gravity. The inductive claim is that this is evidence in the positive for Newtonian gravity. Our experiments should always try to disprove that which we believe, in the Popperian view.
There are strong echoes of this in classical statistics, such that we discuss the "null hypothesis", which do either "reject" or "fail to reject". In classical statistics, we never prove things, and there's no notion of strength of empirical belief the way there is in Bayesian epistemology -- merely a lack of falsification.
That's the 3-paragraph version. I once wrote a 20-page paper regarding falsification and the Popperian worldview.
Well, you can't prove it's not about him.
ReplyDeleteI don't get it.
ReplyDeleteI don't get it either and now I could go to "explain xkcd", but I prefer to read your explanation :D
ReplyDeletePopper's principle work was in the philosophy of science. He championed a definition of scientific hypotheses which is something like: A hypothesis is scientific if and only if it is falsifiable. That we can never prove a scientific hypothesis, merely not yet disprove it.
ReplyDeletePreviously, the primary view was one of induction: that experiments give inductive evidence for a hypothesis. That is, if I drop a bowling ball and it falls at 9.81 M / s^ 2, the Popperian view is that this is evidence that does not disprove Newtonian gravity. The inductive claim is that this is evidence in the positive for Newtonian gravity. Our experiments should always try to disprove that which we believe, in the Popperian view.
There are strong echoes of this in classical statistics, such that we discuss the "null hypothesis", which do either "reject" or "fail to reject". In classical statistics, we never prove things, and there's no notion of strength of empirical belief the way there is in Bayesian epistemology -- merely a lack of falsification.
That's the 3-paragraph version. I once wrote a 20-page paper regarding falsification and the Popperian worldview.
.
ReplyDelete