A TV detective may say “I have a theory that the murderer may be someone in this room.” The way he’s using the word “theory” isn’t the way the word is used in science. The words “fact” and “hypothesis” and “law” suffer from the same difference in perception and meaning between scientists and non-scientists.
Joe Hanson, Ph.D. of the PBS Digital Studios channel It’s Okay to Be Smart explains the way science uses these words. Understanding the difference is one thing, explaining it to someone else is hard. Hanson does a pretty good job, but it will be much easier for us to send someone the video than to explain it the way he does. -via Digg
Anyway, I think trying to ascribe specific definitions to those words often misses the important points. Inductive logic never results in absolute truths, as you never test every possibility. Statements in science, like most things, just come in a giant spectrum of confidence. That confidence shifts with each new observation, prediction and competing idea, but it rarely jumps discretely from one category/label to another.
Unfortunately, having some sense of where a theory falls on the confidence spectrum usually involves learning some of its history, what observations it is based on, and what alternatives are being pursued. That takes a lot more effort than just looking at its name.
Anyway, I don't mean to be negative about the video, as it is still roughly how those words are used. But it reminds me of oversimplifications that were in primary school textbooks, which do a good job of teaching the gist of a topic, but sometimes can be counterproductive when the person stumbles over a more complex case.