Thursday, November 4, 2010

You Never Prove Anything

You can never prove anything.

2, 4, 6... 8, right? How about 10? (2+4=6, 4+6=10, or because I was pulling numbers out of a hat, and those were the ones that I happened to grab)

2, 4, 6, 10... 16, right? How about 12? (2, 4, 6, skip to ten any way you want, then start counting by twos again)

There's always the possibility, if not the likelihood that something unanticipated will explain your data in a way that invalidates your theory. (Everything is a theory.)

Still, there's a hierarchy of plausibility. Nobody is going to jump off any cliffs soon, expecting to fly.

We see something, think about what might make it happen, and get some ideas. Those ideas are hypotheses. We test a hypothesis by predicting things that would happen if it were true. "Objects will not float off the table." "Objects will fall if I drop them." "Planets will maintain certain orbits around the Sun." "Because Neptune's orbit isn't what we expect, we can expect another planet (Pluto) here."

When a hypothesis has a record as a reliable predictor, call it a theory.

No comments:

Post a Comment