Methodological naturalism 'presupposes' that there is no external cause, and that the explanation for the existence of a creature (or organism), the earth, or the cosmos as a whole is found within the creature, the earth, or cosmos itself.
Not strictly true. This sounds closer to ontological naturalism, which proposes that supernatural does not exist. Methodological naturalism proposes that empiricism is the key, that we must be able to test our hypotheses and reasonings before calling it science. This puts the supernatural out of its remit. It does not state that the supernatural does not exist, only that the supernatural is by definition untestable. This leaves things nicely open - things which we call supernatural now will be considered natural if they turn out to be testable.
When trying to define something as pivotal as 'science', should we presuppose as a convention anything at all, other than the search for the truth?
Methodological naturalism is exactly that - it is a method that has been developed to find the truth: using empricism and testing. It goes one step further even than that - by maintaining tentativity.
You seem to be proposing methodological supernaturalism. The problem with methodological supernaturalism is that there is no 'method'. How does one pursue the supernatural? The biggest rational argument at the moment is that if it cannot be explained using the current discoveries of methodological naturalism, then the supernatural can be invoked as an explanation. However - this assumes that methodological naturalism has all the answers already and that new discoveries and ideas do not form. It also falls into the trap that if something is no longer confirmable by science for some reason, the supernatural could be involved. It is the pitfall of the GodDidIt argument. When is it valid to invoke a supernatural agent and when is it invalid? This is where the method of supernaturalism so far fails.
™ Version 4.2
