I often agree with Randall, but in this case I think he's (mostly) wrong. Yes, ideas are tested by experiment. Properly constructed experiments. That means repetition, controls, statistics, the whole nine yards. If scientists used Mythbusters-style experiments we'd still think light objects inherently fall faster than heavy ones (after all, most lighter objects do fall slower than heavier ones, thanks to air resistance). You don't think people in the "unscientific darkness" didn't actually try out a lot of the things they got wrong? Of course they did. They got it wrong because they ran their experiments improperly. And Mythbuster's often does as well. To be fair, "it didn't work this time, lets try it out 99 more times to make sure" doesn't really make entertaining television, and that's generally all Mythbuster's is: entertainment. They have the seeds of science (experimentation), but science is far more than that alone. The problem is, when people look at what they do as actually being science, they end up thinking you can confirm a scientific theory with a single experiment run with 20 minutes of work. And the conclusion to that thought process is looking at the weather report and dismissing global warming because it's a particularly chilly summer. Or saying "hmm, well [some action] didn't kill me this time, it must be perfectly safe."
Scientifically educated people don't come to that conclusion, of course, but those people aren't the problem.