Those interested in the philosophy of science should be aware of Karl Popper[0], and his view that what makes a theory scientific is its falsifiability: If a theory can be false in some way, then you must be able to construct some sort of empirical experiment which could demonstrate its falsehood. This is in contrast to the verificationism that was popular in the Vienna Circle during during his time, which took the view that positive experimental results was evidence towards the truth of a theory. So in his view, science is a process that produce theories that converge on the truth, but never quite get there. For him, "scientific fact" is an oxymoron.
He also famously got into a heated debate with Wittgenstein over the nature of philosophical inquiry. So heated, that Wittgenstein almost beat Poppers ass with a fire poker, or so the story goes[1].
Also of importance in this area is Kuhn's [0] the "Structure of Scientific Revolutions" [1], which not only first coined the term "paradigm shift", but is also probably the best account of how science really works.
Your comment does not make any sense to me. 'The Structure of Scientific Revolutions' does not just propose 'theories': it mainly shows how the Popperian and other accounts of 'how science works' are simply wrong, as a close consideration of historical evidence shows that science just does not work that way. It goes on to explain why it would also be impractical if science did work that way. Those parts are all pretty uncontroversial.
The account of how science does work is still being debated. Perhaps Kuhn disavowed his account later, but that doesn't matter with respect to the most important parts of the book.
It's true that he, along with Kuhn, are probably the two most influential philosophers of science. But in many respects his philosophy of science is ahistorical--naive falsificationism isn't really present in any of the major scientific revolutions. (One might make the tongue-in-cheek remark that his descriptive theories have been decisively falsified.)
His defenders, of course, would claim that that naive falsificationism is a misreading of him.
An alternate way of expressing the same idea is that science is falsehood-finding process and not a truth-finding process. In other words, the results of each new experiment tells you which theories are incorrect. The others are all possibly correct. Occam's razor recommends you should go with the simplest one.
An another comment pointed out, this is extremely obvious from the bayesian point of view.
It's a good approximation but it misses the mark. String Theory is not verifiable but it's still a scientific theory. Bohmian Mechanics is, some say, by definition not verifiably different from Quantum Mechanics but it's a scientific theory. You might call them "speculative" scientific theories to save the idea but at the end of the day the Popperian [sic?] view of science is a bit too simplistic to be useful.
I don't understand why should String Theory be labeled as "scientific". Can you explain? Why cannot we be happy just calling it philosophy?
To me, "scientific" means "uses scientific method", which in turn is a set of tools that are pretty successful in understanding universe. Scientific method is certainly not fixed set of tools - for example, computer simulation became very useful only recently.
Maybe doing whatever string theorists are doing will become useful in the future, and as such their method will become part of scientific method; but since we aren't really sure if string theory itself is true, we can't consider their methodology to be useful (or useless) and so part of scientific method.
For me looking into String Theory's status as scientific/not-scientific is a good way to learn that the Popperian view of falsifiable==scientific is a bit too simplistsic and that classifying things as scientific/not-scientific may not even be a productive activity. It's just not that simple.
A good quote from that thread which may answer your question: "Anyways, my point is that String Theory is not just some whacky idea "what if everything was strings" that exists in a vacuum, but is rather a very natural and conservative extension of Quantum Field Theory to a more general and less arbitrary framework."
And while your're at it, I suggest taking the Bayesian approach for the criterion of a scientific method, where falsification is a particular (and simplified) case of Bayesian reasoning.
A great introduction may be found at [0], if one is already well versed in Bayes theorom, then an excerpt [1] should be enough.
He also famously got into a heated debate with Wittgenstein over the nature of philosophical inquiry. So heated, that Wittgenstein almost beat Poppers ass with a fire poker, or so the story goes[1].
[0]: https://en.wikipedia.org/wiki/Karl_Popper
[1]: https://www.youtube.com/watch?v=vBmt2sCdd5E