There's a post that does what it says on the tin

Reward missions/ challenges/ subtlety instead of violence; add consequences and, if it's 4e, move the roleplaying section in DMG 2 to the front of DMG 1
Won't work.
Of McRae's 4 points, only one is necessary to make the case, and it's the one you can't do anything about.
1) "Extensive rules for combat, making it seem like an expected activity.":
This may be true, but even if it was false, it wouldn't be sufficient to change anything. Besides, its exactly backwards. RPG's tend to have extensive rules for combat because it is an expected activity. RPG's don't have combat as the expected activity merely because there are extensive rules. This is a subtle distinction, but its an important one. Combat is the expected activity because the precursors to formal RPG's feature combat. I'm not talking about merely the war games that were the formal precursors of D&D, but the entire attitude of role-playing and of play in general is about combat. Before RPG's, I played cops and robbers, cowboys and indians, spacemen and aliens, and pirates. When I wasn't shooting something, I was whacking at it with a stick or cane pole that stood in for a sword. The sword games were easy. You knew who was dead by who got whacked with the stick. On the other hand, they weren't safe. The shooting games were much safer, but you couldn't tell who was winning and there were often frequent metagame arguments like, "BANG! I shot you! No you didn't, you missed. BANG! I shot you again. No, I dodged. BANG! I shot you." RPG's formalize a structure for resolving the typical conflicts of role-playing. The reason you have extensive rules for combat is that its only at combat that you really need extensive rules. For non-combat roleplaying, consensus and sense of the individual players is generally enough to resolve conflict. You don't need rules for talking. It's acceptable and possibly even highly desirable to simply allow each player to decide how they respond.
2) "It's all, or mostly, what the PCs are good at."
But even if this isn't true, it doesn't mean that the game won't be about killing. Besides which, it's not even true. Quite arguably, even in 1e, the general sense you would have about the starting characters were you not expecting combat is that if it did happen only the fighter is any good at it and even he's not that good as a single hit is likely to leave him dead. The vast majority of the characters abilities and assumptions about those abilities would not be about combat, and a person without a combat assumption bias would look at the 1st level spell list and see that the combat spells were quite weak but that the non-combat spells have apparantly extraordinary and indeed world changing utility. Clearly, this is not a combat game.
Likewise, without the bias in favor of combat, a player looking at the rules of 3e would see that gain skill at combat is hard, but that its quite possible to have extraordinary even supernatural skill outside of combat at relatively low levels. Clearly this isn't a combat game either, as combat is discouraged by the rules!
And this is to say nothing of games like GURPs or others where the list of non-combat abilities vastly exceeds the list of combat abilities. Yet, these game remain largely about the killing and the vast majority of rules concern combat.
3) "XP for killing."
But in every edition of the game, you've got XP for other things as well. In 1e, the XP you got for killing was for most campaigns a tiny fraction of that recieved for hauling off valuables. Quite obviously, by this logic then, 1e was an archeology game and killing things was intended to be a minor part of it. Likewise, in 3e you don't get XP for killing things - you get XP for overcoming obstacles. If there was a bias toward killing things, it therefore had to be introduced because you could just as easily set up adventures where every obstacle was diplomatic or social and you could have given out just as much XP.
Moreover, this doesn't address the large number of games that don't give XP for killing, and yet which still manage to be about killing things.
5) "Lawless setting, like the Old West."
But this is certainly not universally true. And more importantly, even when it isn't true, the fact that it isnt' true doesn't make the game less about killing. You can play in a setting that is the opposite of lawless - some tightly regimented police state - and yet don't be surprised if killing of some sort breaks out in short order.
4) "Violence works."
This is in and of itself sufficient to explain the presence of violence regardless of setting or system. The problem here is inescable. Violence solves everything. This is the reason humans resort to it. Violence is the cutter of gordian knots. Violence is the hammer in the tool chest. The only way to prevent violence in a game session is to present the characters with difficulties that can't be solved through acts of destruction, and inherently this cannot be done in a setting that features as its primary content conflict with monsters, villains, foils, and antagonists. In such a setting, destruction works. Sooner or latter, with enough destruction, the problem just goes away.
The problem that you are going to run into here is that its not merely, as Umbran intelligently and perceptively expounds, the desire to have conflicts that are different from those commonly experienced, but precisely this desire to confront problems that can be solved through acts of destruction that is the primary motivating force in interest in the setting. If you present people with a game setting that requires laborous acts of creation to resolve problems, most people are going to be terribly uninterested both with the concept and with the resulting gameplay. It's precisely this difference between the sorts of problems we normally encounter and the sorts of stresses that we normally experience that explains the attraction of violence in our games.