In the second conference which comprises William James' Pragmatism, he explains why this philosophy does not in principle entail a preference for materialism over idealism, "no such materialism bias as ordinary empiricsm labors under". He writes:
If theological ideas prove to have a value for concrete life, they will be true, for pragmatism, in the sense of being good for so much. For how much more they are true, will depend entirely on their relations to other truths that also have to be acknowledged.
The Absolute of transcendental idealism performs a concrete function, insofar as its believers draw practical consequences or guides to action from it. He emphasizes these consequences are good, and insofar has they are good, he writes, who could deny the truth of the Absolute?
Truth, James' says, is "a species of good", not a category distinct from it. "True is the name of whatever proves itself to be good in the way of belief". "What would be better for us to believe" is proposed as a candidate definition for truth. Something is true if it "fits every part of life best and combines with the collectivity of experience's demands (...)".
There is a special kind of relativism here, one which is not only different from traditional epistemological relativism (truth does not truly exist independently from a subject's perspective) but which in fact contradicts it entirely. It is of course conceivable that an idea $\lambda$ may be consistent and fruitful within the 'system' of person $A$ while being inconsistent or unfruitful within the 'system' of person $B$. (By a person's 'system' I mean here the factors which James' deems important in the formation of belief: the pre-existing collection of ideas the individual has and the demands which experience imposes upon them.) So far nothing problematic exists: in logics, for instance, $(a)$ different sets of axioms or $(b)$ different models may associate to a given formula $\varphi$ a different value, and the semantic of $\varphi$ is undefined except with regards to any such set or model. However, the statement "$\varphi$ is true under model $X$ but false under model $Y$" is true independently of all models, i.e. it is a meta-logical truth, and in fact the whole first order theory of types would make little sense if one could not make such a claim.
Does James' notion of truth allow for, so to speak, meta-subjective evaluations of truth? If we assume his theory is correct, could we determine the truth value of a sentence such as "$B$ is wrong in believing $lambda$"? If he is right, such question could only be resolved through reference to $B$'s believes, actions and experiences, and the sentence does not imply "$A$ is wrong in believing $\lambda$". This seems to entail the phrases "$\lambda$ is false" and $\lambda$ is true" lack any meaning, insofar as they lack reference to a subject. This issue is only solved when $\lambda$ is an idea which cannot, under any circumstance, be fruitful, productive or coherent. Though there aren't many such ideas, epistemological relativism is one.
The fundamental issue with epistemological relativism, as well as with scepticism, is that it is strictly academic. Everybody acts as if some things were certainly true and some things were certainly false. The most radical relativist has rather strict notions of true and false, as well as of right and wrong, when he takes his car to the mechanic or goes to the hospital to mend a wound or treat an illness. Furthermore, the relativist or the sceptic do not fail to act as if their believes were true due to hypocrisy or lack of consequence: they fail to act in accordance to their philosophies because doing so is contrary to life itself. If any given species, at some point in natural history, was not hard-wired to act as if certain things were facts; if a predetermined universe of truths was not embedded in their brains and expanded through experience, Darwinian evolution has surely erased every trace of it. For what kind of existence can endure without action? And what is belief but rule for action?
But I digress. We must accept at least that we act as if things were true, and thus land again on questions arising from James' theory. Any theory of truth which does not lead from "$A$ is right in believing $\lambda$" to "$B$ is wrong in not believing $\lambda$" is problematic. The fact that a subset of ideas is saved does not reduce the problem. Could one say Galileo and Bruno were wrong in their beliefs? Such beliefs were $(a)$ certainly inconsistent with facts as recognized in their times (this was accurately pointed out in Feyerabend's Against method) and $(b)$ pernicious to their lives. Are the flames which extinguished Bruno's life a refutation of his theses?
Many examples such as these ones come to mind. James' theory of truth is unsatisfactory. Necessary reference to both a subject and a certain (rather undefined) kind of utility makes reasonable debate almost impossible.