Who says?
Without exaggeration, pretty much everybody! That is virtually (maybe even literally) all reflective working scientists or philosophers of science would agree unhesitatingly with this: that good scientific theories must be vulnerable to disconfirmation by the data.
Why carry the attribute of "vulnerability" into the definition of theory? A theory by defininition is simply a way of explaining data. Read the definition again, and tell me how you wring "vulnerability" out of it.
Um, are you actually saying some definition you saw or posted is more import than the way science is actually done? Why don't you tell me why in the world do you think that scientists in all fields and disciplines spend countless hours and dollars devising and performing observations and experiments? Why is any of that necessary if you only need an "explanation" and don't need to test it?!
From the general theory of intelligent design science engages in specific fields of study, much of which enjoys or requires the rigorus application of falsifiable hypotheses, etc. The best science does not shy away from entertaining propsitons that may appear absurd on the face of it. The more reliable science is engaged in direct observation, which places the notion of a 4.5 billion-year-old earth, among other notions like the spaghetti monster, on relatively shaky ground.