We are talking here about a leisure activity, and so moralised or mandatory language (should, ought, would be wrong to, etc) is largely out of place.
Rather, it makes sense to talk about what means are well-suited to what ends. Given the hobby we are talking about, relevant means will include GM techniques, player approaches, choices of setting and theme, etc. And relevant ends must encompass the known variety of RPG play, as well as anything new that someone thinks up and is seriously engaged in.
In the play of a 4e game it makes no sense for the PCs to be fundamentally opposed to one another. The game has no tools for supporting PC vs PC conflict in any seriouos fashion; and its conflict resolution mechanics, particularly for combat, only really show their strength when the PCs are working together.
The contrast here with (say) Burning Wheel is pretty marked. I know from play experience that BW can support PCs who are opposed to one another, or who shift back-and-forth in their allegiances.
This doesn't make one system superior to the other in any objective sense.
Actually, it objectively does if all other things are equal: BW is marked as superior in that it has the flexibility to handle something 4e cannot. Both systems can happily deal with co-operative play but only BW can also deal with non-co-operative play.
But, not knowing all the ins and outs of either system, I don't know if there's an area where 4e is objectively superior to BW in a similar way, which would tend to cancel out the BW advantage noted above.
It does mean that you wouldn't try and duplicate the 4e play experience using BW, nor vice versa.
True; I'd rather find (or create) one system that could equally well provide both experiences.
Setting is just another tool like anything else. It has no magical status, and no special role in setting priorities. If a player wants to play an X, and the GM has conceived of the setting as X-less, then the basic social situation is no different whether X is a gnome or a ninja. If the GM has conceived of the PCs as non-Xs, then likewise the basic social situation is no different whether X is an evil elf, an uncooperative loner, a serpent worshipper, or a demigod. There is no a priori rule of RPGing that says that just because X exists in the setting it is therefore fair game as a player character. This is all part of the process of setting things up.
There's (at least) three different things here than can in theory be restricted.
PC race is one: fair game, the DM can restrict this to suit the setting*.
PC class is another: fair game, the DM can restrict this to suit the setting*.
PC personality (or alignment, or characterization) is the third, and my position is that the DM has no right to restrict these
unless that same restriction applies to all such characters in the setting. (e.g. if my PC Elf can't be evil then by extension there are no evil Elves in the setting; just like my PC Paladin has to be good because all Paladins in that setting are by definition good)
* - or, in some cases, to suit the mechanics in order to prevent 'broken' builds in some systems.
A concrete example: the setting of Cthulhu Dark clearly contains Great Old Ones, various alien/old one races, evil cultists, etc. But none of these is fair game for being played by the players. That would be at odds with the whole genre and theme the system is meant to support.
No Great Old Ones or alien races as PCs comes under PC race restrictions, which are fine. But if I want to play an otherwise-allowable PC as a hidden evil cultist (or a spy for same, or a wannabe who is still trying to gain acceptance), where's the harm? Sooner or later I either get found out or I don't, and I either succeed or I don't, or I have a change of heart, or whatever. Won't know until it all plays out...