I appreciate what you're doing. I'm attempting to respectfully disagree. I think that, as the game is presented, the metaphysical element supervenes entirely on the behavioural and dispositional elements that I'm talking about.
So
Detect Evil or
Know Alignment cast on an orc is (I assert) no different from the same cast on a Brigand (cf Bandit) or Pirate (cf Buccaneer) or random NPC who came up evil on the random alignment chart.
A
Wand of Enemy Detection may of course give a different answer. But perhaps the same answer for Bandit and Buccaner as for Brigand and Pirate. And in the Dragon article I mentioned, discussing a paladin's response to neutral NPCs, Roger E Moore says
Of course, when confronted by a band of wild bandits or buccaneers (all Neutral) you can’t just punch them in the nose and settle things. Again, the sword might be a reasonable answer
(The reference to punching in the nose follows from an earlier anecdote about how a paladin in a game Moore was GMing dealt with a dryad who tried to Charm party members.)
There is a difference between the orc and those random humans vis a vis detect evil or know alignment (I submit). In the case of the orc, it would be surprising, perhaps very surprising for the answer to be anything but evil, because the answer is generally going to be evil. So if I were a PC and I wanted to determine, for example, the extent to which I thought I could bargain with an orc in good faith rather than fighting it, or ally with it and have some hopes of not being betrayed, the spell is, almost exclusively, going to tell me what I already know - that the orc is evil, and trust is not a real option barring significant self interest and even then only as far as that goes. However, in the case of the bandit, there is a much more legitimate question involved, as the possible answers are significantly more varied, and more likely to return a result that might mitigate for a non-violent answer, even if we assume that results from the good column are vanishingly unlikely.
I would agree that the behavioral and dispositional elements play a key, if not foundational role in that ethical heuristic. However, I think expectations also play a role - the orc is expected to be evil, while the bandit may or may not be. Barring evidence of behavior there's no way to be sure with the bandit, but one can be far more sure, if not positive, about the orc. Since it is quite possible for the behavioral element to be missing we are left with the dispositional, at least in some cases. Once you factor in genre expectations, and assuming that the setting hasn't already modified the PCs view of orcs, I think the bandit and orc will be viewed differently by a given PC, or more specifically the the expectations of behavior will be different.
To allude back to the distinction between manslaughter, murder and justified killing: different legal systems will draw the boundaries of defensive violence in different places - eg what is legitimate self-defence, what is excessive self-defence (which is, roughly at least, a form of manslaughter) and what is outright murder? Many US jurisidictions are, by my standards, very permissive in this respect but still I think draw these boundaries.
These distinctions I would argue, are only present in the fiction to the extent that they are already present in the players mind. They play a role, but I don't think it's a fictional role, or at least not generally. Obviously it's possible to role play a character with differential views on the importance of these distinctions, but I don't think that is the common case. I also think genre expectations play a role here as well. In many games, and this is supported by a lot of articles on running games, the enemy does not often surrender, and players are thus not often faced with the prospect of what to do with helpless prisoners. I would not be surprised if an unexpected surrender by orcs put a lot of players in a pickle because all of a sudden the issues you raise above come into play. However, I also think that in many games, enough for me to risk calling it common, that players are not asked to make the distinction very often.
Perhaps we should take a moment to set aside murderhobo games as examples for this discussion. Those do exist, but I don't think there's a lot of profit in including them, as I don't think that ethical decision making plays even a minor role in how those games proceed.
Serious theoretical discussions of defensive violence need to consider such factors as proportionality, necessity (eg how "anticipatory" can such violence be?), motivation (eg if I kill my nemesis out of a desire for vengeance, but - as it happens though unknown to me - happen to prevent said nemesis killing another person at that very moment, am I to be condmened as a murderer or to enjoy the justification of defensive violence?), etc. I wouldn't expect most D&D games to get into that sort of detail, but we can reconstruct the implicit theory that is evinced by the way such games actually proceed.
When it comes to killing orcs, most D&D games seem to set a very low threshold for necessity, connected i part to a very permissive conception of anticipatory self-defence and/or justified warfare. That is, the mere presence of orcs is taken to create a threat that warrants the use of violence to elminate them.
I completely agree with the second paragraph, I just disagree at least in part about why that's the case. I would lean more into the dispositional half of you argument and index that to expectations based on alignment. I'm not really sure how we are supposed to view expectations about alignment from expectations about behavior as significantly different either, as the two are essentially the same. We could argue about primacy I suppose, but I don't know that's it's important to separate "orcs usually act in X fashion which I see as evil" from "I expect orcs to act in X fashion because they are evil". The difference becomes importance once you have actual behavior, naturally, but at the level of expectation, which is where many decisions about anticipatory self defense could be made, I'm not sure it's a distinction with importance.
Again, my advice for games that want to strengthen the veneer of moral permissibility without making anything too complicated would be to reinforce this by having the orcs be raiders of the Westfold (and perhaps who do not spare non-combatants, though that might also be unwanted grimness for some tables).
For sure, but I don't think a lot of games would have to reinforce this, as for a lot of games this would be the baseline expectation.