But M theory does make predictions (at least according to wiki) just that the observed consequences are in inaccessible dimensions. Making predictions is the requirement of a hypothesis, normally we can test these and move from hypothesis to theory. I am unsure what one calls a hypothesis that cannot presently be tested. For example evolution makes predictions about future adaptation that cannot be validated within any reasonable timeframe, all consistencies in the past might not be regarded as predictions - perhaps not hugely dis-similar to M theory really.
So if Darwin was able to state that his 'prediction' was that all successful adaptations would be of reproductive benefit to the individual, is it possible for M theory to predict that all newely discovered interactions at quantum or cosmological levels shall be compatible with the mathematics of said theory!
I don't think these theories are inherently untestable, it's just that we don't yet know exactly how to do it. The key event in the acceptance of general relativity was when Eddington showed that light is bent by the gravitational field of the sun. That didn't "prove" relativity, but along with some other predictions made it account more precisely for gravitational effects than Newton's less accurate law. That may well happen with M theory if it really is the bee's knees.
It's good that Hawking has come to express adherence to a multiverse theory as many creationists were using his supposed single universe views to argue for a single creation. In reality cutting edge thought in cosmology has been moving towards multiverse ideas for quite a long time.
Bookmarks