I really like that! ''Provisional.'' So, science never says ''yes,'' it can at best offer a strong "maybe" or an emphatic ''no?"
I'm inclined to say that it might be almost as hard to give an "emphatic no" as an "emphatic yes". No matter what evidence we generate that seems to contradict our hypothesis, the hypothesis still might be true if our evidence is mistaken or if we readjust some of the other assumptions that we are making in our reasoning.
I wonder if someday, maybe centuries from now even, if every scientific theory will ''fail'' a new set of experiments, and give way to a new theory?
In the philosophy of science that's called the "pessimistic induction". If many/most of the 'scientific' beliefs of the past have subsequently been shown to be wrong, even ones that were supported by what was believed to be excellent confirming evidence (such as geocentric astronomy and the heavenly spheres) why shouldn't we expect that future science will reject much/most of what we currently believe?
I'm inclined to agree with Exchemist in seeing science as a huge intellectual
model (actually a bunch of interacting ones), that embody our best current understanding of how the physical universe behaves and how it's put together. It isn't the last word, it's an ever-more-accurate
approximation, the best that we have at the moment.
The way I look at it is that 'true' and 'false' rarely if ever indicate absolute
apodeictic certainty. These truth-values probably should be thought of as having likelihood-weights. Beliefs that seem highly certain get high weights, perhaps 99% (a small fraction below absolute can't-be-wrong certainty). Beliefs that are more like guesses, mere shots-in-the dark, might only rate a 51% rating (a small fraction above flip-a-coin).
Logic's 'true' and 'false' truth-values might be intellectual idealizations more than something that applies to real-life reasoning. And that in turn implies a rejection of classical logic in favor of something like many-valued fuzzy logic.