Andrew Rowell over at ID in the UK has done a very good job of exposing the problems with having methodological naturalism as the exclusive methodology for the natural sciences:
The faith of the methodological naturalist.
The basic articles of faith for a methodological naturalist go something like this:
We have found excellent naturalistic explanations for many phenomenon in nature.
Therefore
we believe every phenomenon in nature will have a naturalistic explanation.
Therefore
we make it a strict rule that science is exclusively the study of possible naturalistic explanations for what can be observed in the universe.
Science is not the search for the truth about the origin, operation and destiny of the universe it is limited exclusively to purely naturalistic explanations of the origin, operation and destiny of the universe.
The methodological naturalist will choose a naturalistic explanation over a meta-nature explanation to be taught as the truth in science lessons even if it is not actually true.
Thus for a methodological naturalist it is perfectly reasonable possibility that in science lessons it will become necessary to teach children what is in fact not true and what is in fact known to be untrue for the sake of meeting the methodological naturalism criteria laid out by the grand assembly of the interplanetary science council.
The real truth can only be taught in a new subject called meta-science lessons and it is a perfectly reasonable possibility in the future for the syllabus in these lessons to contradict the science syllabus and for the meta-science lessons to be teaching the truth and the science lessons to be teaching what is known to be wrong.
MN has been a very productive, beneficial methodology for science. It has led to a great deal of knowledge about nature, but the presumption that every natural phenomenon can always be reduced to unintelligent natural processes is not grounded in empirical science. Such an assertion can only be made from a position of faith.