Imagine an ISIS jihadist with a spell that detects faithful Coptic Christians, ones who really live their beliefs. Having an reliable way to Detect Good (Coptic) doesn't necessitate believing that Christians are actually RIGHT about anything.
I'm a bit wary of real-world examples, but with due caution will respond to this.
First, the proper interpretation of the requirements of any major religion (including Christianity, and including its Coptic denomination) is a matter of contention. It's analogous to my example of
beauty, upthread. Human beings do not have
immediate epistemic access to facts of beauty - our immediate access is to perceptual facts (facts about colour, shape, visual texture etc), and beauty supervenes in some very complex way on these more basic facts.
Similarly for religious observance: the basic facts, to which we do have immediate access, are facts of behaviour, and (some) facts of motivation (trickier, because of the "problem" of other minds, but let's put that to one side for the moment). The fact of being a
good Copt supervenes on these in some complex way, and in the typical case actual human beings don't have un-mediated access to this fact (absent supernatural revelations, which are themselves, among human beings, often contested matters).
To elaborate a bit more on that, let's move away from real-world religion and into fantasy. Consider, for instance, a fantasy game set in Middle Earth, wanting to evoke some of the thematic content of the Akallabeth. One question that is going to come up in the game is, What are the right and wrong ways of honouring the Valar, and their various spiritual servants, and of the human kings who have been endowed by providence with the right to rule, etc? The facts of value relevant to answering these questions have the same sort of supervenience structure as described above; so does epistemic access to those facts. So, for instance, the characters in the gameworld can have immediate epistemic access to a fact of human sacrifice (eg they can see and hear it taking place, perhaps perform it themselves, etc), and their epistemic access to the value fact that
a wrongful event is occurring, which shames both humans and the Valar who have tried to guide them is mediated by that basic fact.
Suppose, in this hypothetical Middle Earth game, we then give certain characters access to a magical ability
Detect Faithful. We are now positing that - much like your "Detect Copt" example - the character in question has un-mediated epistemic access to facts of value, without having to epistemically engage with the more basic facts on which those value facts supervene.
For this to work, in game, the participants (led, presumably, by the GM)
do have to form a view about what is right: they have to form a view about what behaviours and motivations are or are not consistent with being faithful to the Valar (given that the value fact of being faithful to the valar supervenes upon such more basic facts).
In games being run with fairly broad-brush or B-movie morality, that is probably not a big deal. Human sacrifice is obviously a breach of faith, doing your duty in the war against Sauron is keeping faith, etc. But it doesn't take very much subtlety for the issue to become a bigger deal. For instance, being
prideful in the performance of your duty (Denethor, Saruman, and sometimes Boromir) is a way of breaching faith with the Valar, but what counts as
pridefulness might be something on which members of an RPG group could reasonably have differing views. My advice to groups wanting to run a Middle Earth game that wants to be subtle enough to explore these sorts of issues would be to not have characters (PC or NPC) having the supernatural ability to Detect Faithfulness.
Similarly, positing a Detect True Copt ability raises the same issues. I wouldn't recommend it for a game in which issues of religious interpretation or faithfulness are meant to be engaged with to even a fraction of a degree of the seriousness with which mainstream art and literature do so.
Second, even if we put all the foregoing to one side, Detect Good in D&D (or Know Alignment, etc) is not a detector of religious conviction or propriety. It detects moral facts. What a succubus learns when she scans a paladin is not that the latter is (say) faithful to Bahamut, but rather that s/he is
good.
In fact, it is
because the paladin is good that s/he counts as one of Bahamut's faithful. (It is "law and good deeds [that] are the meat and drink of paladins" (Gygax's PHB p 22), after all, not fidelity to some god's decrees.) In other words, D&D in its standard or default presentation of 9-point alignment rejects Hobbes' answer to the Euthyphro: goodness is loved by Bahamut because it is good; it is not constituted as good in virtue of being loved by Bahamut.
(I think this is consistent with Gygax's DMG p 23: "alignment does not necessarily dictate religious persuasion, although many religious beliefs will dictate alignment".)
Are you arguing that the Succubus must of necessity share the axiological values of whatever force/entity is responsible for assigning Know Alignment results? If so, why do you think that?
I think it because it's what the game texts say: for instance, the 3E SRD describes alignment as "general moral and personal attitudes", and goes on to say which of these are
good and which are
evil. Similarly, p 33 of the 5e Basic PDF says that alignment "broadly describes . . . moral and personal attitudes. Alignment is a combination of two factors: one identifies morality . . ."
So, for instance, when Gygax says that, for lawful evil persons, "life, beauty, truth, freedom and the like are held as valueless, or at least scorned" I don't read him as saying that
evil is just a shorthand for "scorner of life, beauty, truth, freedom and the like". I read him as telling us that such individuals are
evil ie are committing moral error.
One can also look at it from the paladin's point of view. The paladin uses Detect Evil (classic version, not 5e version which is different), and the succubus registers. As I read the traditional alignment text, the paladin learns not just that the succubus scorns life, beauty and truth, but
also that the succubus, in doing so, is wrong. Which is to say that using Detect Evil is, for the paladin, not
just a way of identifying those who scorn the things the paladin values, but is
also a source of moral comfort: it affirms the paladin's values.
There's nothing stopping anyone from adopting a different approach: that
good is just a shorthand label for "respects and fosters the welfare of others, plus beauty" and
evil a label for "disregards both beauty and the welfare of others". But that's not how I've ever read the alignment descriptions; they seem to me to affirm that there is
reason to respect and foster the welfare of others, and hence that is why it is good. The evil, being
evil disregard these reasons.
In any event, I think that a game that adopts the shorthand label approach has a reasonable chance of encountering one or both of a couple of issues. There is the first issue I described in relation to the "Detect Copt" example: a subtle game may put pressure on issues like "what is welfare" or "what is beauty", and hence bring to light any disagreements over such matters among the game participants, and hence make the use of Detect Evil or Detect Good a contentious matter even when divorced from moral evaluation - who's to say, after all, that the succubus hasn't
really gotten to the bottom of the issue of welfare (maybe the Sadeian theory of libertinage is correct), in which case perhaps
she should register positive to Detect Good!
I've run campaign in which the PCs (and their players) form the view that the (so-called!) Heavens are wrong, because they have formed the view that the gods' conceptions of welfare are corrupt and self-serving. In such circumstances, how do we determine who registers to Detect Good. What sense would it make, in Planescape say, for the Seven Heavens to fail to register to Detect Good? The whole notion seems incoherent to me.
Second, there is the Euthyphro problem, which in practical D&D play tends to manifest itself in the collapse of moral distinctions into team distinctions. Rather than angels and paladins being
good, and devils and demons
evil, we get angels and paladins valuing
this stuff, while demons and devils scorn
the same stuff, and so they come into conflict. But we then lose our ability to frame the conflict as asymmetric, because we've deprived ourselves of the necessary moral vocabulary by simply equating it with certain orientations (either valuing or scorning) towards
this stuff.
My general solution to these issues, for any game that is going to raise subtle issues: use alignment as a broad metagame descriptor of NPC personalities/allegiances; but drop alignment as an ingame phenomenon (no Detect/Know Alignment spells, no aligned outer planes, etc). 4e basically went down this route; so does 5e in its character and monster building, but not so much in its cosmology.