Through the Broken Window

We're told since middle school that conducting a proper scientific experiment is conceptually simple. You observe a phenomena, you formulate a hypothesis to explain it, divine some testable predictions from that hypothesis, then finally test those predictions using experiments. If enough experiments confirm your predictions, you have yourself a scientific theory. We call this the "scientific method".

The important criterion here is falsifiability. For a theory to be scientific it has to yield testable, falsifiable, predictions. If you can't test it, it ain't science. This is a very reasonable sounding criterion for what is scientific that falls apart upon closer inspection.

What I mean is that if we follow this criterion to its logical conclusions, we would have to dismiss all familiar scientific theories as pseudo-science. Why? Because one can never "disprove" a hypothesis with empirical evidence--no hypothesis is falsifiable. The problem arises because we can't test single hypotheses one at a time. When we conduct experiments we have to test a slew of hypothesis all at once. For example, If I were to test my hypothesis that a force called gravity will cause an apple to fall to the Earth when it's dropped I am also testing my hypothesis that no force greater than gravity will push the apple away from the Earth.

If I drop the apple and it flies into space, does that disprove my hypothesis that a force pulls it to the ground? Not at all! I can easily say that it only disproves my second hypothesis. There must be some force stronger than gravity pushing the apple into space!

This is an important insight made a long time ago by philosopher of science Pierre Duhem that still hasn't fully been appreciated by our legal system or high school science teachers.

The moral of the story: Searching for truth about the world around us is VERY hard and empirical evidence only takes us so far.