What is religion?

The loose use of the term religion as though we all know and understand what the term refers to continues to infuriate and frustrate me. David Bentley Hart has a passage in an entertaining and instructive article "On the Trail of the Snark with Daniel Dennett" that explains why I am frustrated.  The article appears in a collection of articles and essays entitled: In the Aftermath: Provocations and Laments (Eerdmans, 2009).

And here, I think it needs mentioning - just for precision's sake - that religion does not actually exist. Rather there are a very great number of traditions of belief and practice that, for the sake of convenience we call "religions", but that could scarcely differ from one another more. Perhaps it might seem sufficient, for the purposes of research, simply to identify general resemblances among these traditions: but even that is notoriously hard to do, since every effort to ascertain what sort of things one is looking at involves an enormous amount of interpretation, and no clear criteria for evaluating any of it. One cannot establish where the boundaries lie between "religious" systems and  magic, or "folk science", or myth, or social ceremony. (Comment Anzac Day services?) There is not any compelling reason to assume a genetic continuity or kinship between, say, shamanistic beliefs and developed rituals of sacrifice, or between tribal cults, and traditions like Buddhism, Islam, and Christianity, or to assume that these various developed traditions are varieties of the same thing. One may feel that there is a continuity or kinship, or presuppose on the basis of one's prejudices, inklings or tastes that the extremely variable and imprecise characteristic of "a belief in the supernatural" constitutes proof of a common ancestry or type; but all of this remains a matter of interpretation, vague morphologies, and personal judgments of value and meaning, and attempting to construct a science around such intuitions can amount to little more than mistaking "all the things I don't believe int" for a scientific genus. One cannot even demonstrate that apparent similarities of behavior between cultures manifest similar rationales, as human consciousness is so promiscuously volatile a catalyst in social evolution. ...

Moreover, the task of delineating the "phenomenon" of religion in the abstract becomes perfectly hopeless as soon as one begins to examine what particular traditions of faith actually claim, believe, or do. It is already difficult enough to define what sort of thing religion is,. But what sort of thing is the Buddhist teaching of the Four Noble Truths? What sort of thing is the Vedantic doctrine that Atman and Brahman are one? What sort of thing is the Christian belief in Easter"? What is the core and what are the borders of this "phenomenon"? what are its empirical causes? What are its rationales? Grand empty abstractions about religion ares as easy to produce as to ignore. These by contrast are questions that touch on what persons actually believe; and to answer them requires an endless hermeneutic labor - an investigation of history, and intellectual traditions and contemplative lore ... (pp.192-193)

William Cavanaugh reaches a similar conclusion in his analysis of the difficulty of categorising what is and isn't a religion for the purposes of the flourishing scholarly industry over the relationship between religion and violence.  Cavanaugh makes clear in The Myth of Religious Violence  in a way that is not quite as clear in Hart's discussion the extent to which any consistent account of specific religions will end up drawing political and civic religions into the scope of scholarly work in this field.