Pretty much, though in some cases testing means further observation. For instance, the prediction that inflation should leave a non-uniform level of background radiation could only be tested by observing it, not by reproducing the Big Bang.
In some cases. In others, it might be hard to completely falsify. Consider the hypothesis that the asteroid wiped out the dinosaurs. Finding a few fossils after the kT boundary doesn’t really falsify it, since a very few survivors can be expected. The drop in the population doesn’t prove it, since there might be other reasons for extinction. Yet this is clearly a scientific hypothesis.
An interesting question. Technological hypotheses are clearly falsifiable, since you build them and if it doesn’t work you have falsified it. But I can see people objecting to the idea that the application of science, for instance in the design of advanced semiconductor processes, is scientific. But that would be a discussion not involving falsification.
Because the utility of science lies in its predictive power. If you only include data from past experiments, you are basically only curve-fitting, which is relatively easy and not that useful.
It’s not always bad–Keplerian mechanics didn’t really have any new predictive power; it just explained the motion of the planets in a much simpler and more accurate way. It took Newton, and then Einstein, to really solve the problem, though.
At any rate, prediction enhances falsifiability, because you can say that your new theory lives or dies on the result of the new experiment. Our strongest theories–relativity, evolution, etc.–are tested every day via experiments that the original scientists could not have even anticipated. That they continue to succeed is what gives them their strength.
Because if you look only at past data - esp. if it’s a narrow set - you can easily have three or four different hypotheses that all explain the past data. So you need to devise a test that rules out hypothesis 1, or confirms it.
That’s also why scientists are not angry (well, not too much) about a failed test . If you see that a ball thrown up describes a parabola and then bounces up 80% of the first height, you can formulate a law on how a ball behaves. Then you test it not with a basketball, but a medicine ball, which flops instead of bounce, and you revise your hypothesis. You do another test with a flubber ball which bounces twice as high, and you again revise. Until you end up with half a dozen different laws for each material. Or one big law:balls do whatever they like, but they will come down again if you wait long enough.*
There’s a Harry Potter fanfic over at fanfic.net, Harry Potter and the methods of rationality. (You can download it in one big pdf). As fanfic, it’s very bad because everybody is acting out of character half the time.
But as “science explained using Pottervers” it’s very interesting. There’s one chapter where Harry (who’s supersmart and knows more books than Hermione) meets Hermione on the train and explains to her how important it is to devise a test not only that validates your theory, but one that invalidates it. His test is
saying three numbers, and finding out the rule. After Hermione tries several modifications, he finally reveals an even simpler rule, and says only about 20% of people get it. (that’s because, as the Cracked article explains, we’re hardwired psychologically to be right. Subconscious, we don’t want to be proven wrong.)
*I just watched a Mythbusters special: 25 best parts of the show, and they had segments talking with them. In one clip, with failed set-ups, Adam stressed that to a scientist, a busted test is still very important, because negative data is still data.
** And then a bird flies by and catches your ball, and you amend your law again
Cecil addressed this once, when commenting on a different (though related) topic. His explanation is as good as any (start at “Exception Stupidity, Part Two”):
Or, as some people like to say, “Not even wrong.” It’s nonsensical. It cannot be addressed in any meaningful way.
A friend of mine is currently trying to convince me of the existence of the human soul. But the terms are so negatively defined – massless, intangible, invisible, undetectable, cannot be stopped by any object or force – that the object is the subject of “nonsensical” definitional terms.
There is a classical truism in confirmation theory: “You cannot prove a negative.” This is generally true, but I would like to point out a class of exceptions. You can prove a negative in such limited cases as permit an exhaustive search of the experimental space.
I cannot prove that there is no “Bigfoot” – an undiscovered large anthropoid living in the American north-west. But I can prove that there isn’t one in my living room, because I have the ability to search my living room exhaustively (at least at one the scale required.) The American north-west is simply too large for that expedient.
(The last time I said this, someone said, “What about a microscopic bigfoot?” So, to quote Ray Bradbury, I hit him.)
Another class of things that it would be possible to disprove the existence of (that is, prove not to exist) would be things purported to have contradictory properties, like a cube that is red all over and green all over, or a four-sided triangle, or thing that has everything in it, or an omniscient, omnipotent being.
