Let's look at (a particular application of) bioinformatics. I sequence the DNA of a bunch of humans and a neanderthal. I apply some statistics and proclaim that humans and neanderthals admixed. Later, I sequence more humans and another neanderthal. Applying the same statistics, things look a bit messy now. So I proclaim that there were two admixture events, but I also change the estimated amount of admixture and pretend that this is still consistent with the first claim. Years later, another genome, another admixture event, more hand waving excuses for the numbers that don't seem to fit the big picture.
Is that science? Some kind of trial is there. It is appropriate given time and setting. Some kind of value (a p-value) is produced. Even the theory is there, even though calling it "formal" is a stretch. So this is science? I don't think so.
Let's look at (a particular application of) bioinformatics. I sequence the DNA of a bunch of humans and a neanderthal. I apply some statistics and proclaim that humans and neanderthals admixed. Later, I sequence more humans and another neanderthal. Applying the same statistics, things look a bit messy now. So I proclaim that there were two admixture events, but I also change the estimated amount of admixture and pretend that this is still consistent with the first claim. Years later, another genome, another admixture event, more hand waving excuses for the numbers that don't seem to fit the big picture.
Is that science? Some kind of trial is there. It is appropriate given time and setting. Some kind of value (a p-value) is produced. Even the theory is there, even though calling it "formal" is a stretch. So this is science? I don't think so.