So many good comments on yesterday's post that I can't really do them justice in response, so they are likely to be fodder for a few more posts. So with that in mind, I want to start with one point that a few people raised about a distinguishing feature of combat, and that is of course, risk.
Classically, RPG combat includes a risk that the players might lose and die. This gives the player strong personal stakes in the outcome, even if the fight is otherwise uninteresting. The fight might be the most random, railroad-y, cookie cutter encounter you can imagine, but if it could kill your character, you pay attention.
Now, there are two things worth noting about this. First, there is nothing about this that suggests that non-combat _can't_ be compelling. It's certainly easier to draw players in with personal risk, but it's far from the only way. Skill scenes, especially athletic and social ones, can often carry dire risks, and that's easy too, but that's only part of it. ANY scene can get player investment if it's interesting and if the stakes interest the players. Yes, if your players are only interested in survival then you'll have a hard time coming up with ways to interest them without threats, but I think most players are a bit more broad-minded than that.
Second, and more profoundly, it's all kind of a farce.
There have been games where the lethality of combat was a real consideration, and speaking as a rolemaster fan, I can say that they can be a lot of fun, but that is not the mode of most modern games, and it is especially not the mode of 4e and the various 3.x derivatives. While they keep the trappings of combat and risk, it's usually built on top of a resource management engine. The risk of a TPK does not come from the opposition so much as it comes from the danger of a badly designed encounter.
Now, this is not to say there's no risk. Heck, some games (like Gamma World) have done a good job of bringing risk back to the table, but they do so by embracing ideas that reduce the pain of character loss. But the point is that even in risky games, the generally expected outcome (and, in fact the desired outcome) is that the PCs will win. It may be costly, but they'll pull it through.
On some level, I feel like a bit of a grognard here, in that what I'm saying is not too far off from "It's not very convincing to talk about risk if you're not playing a game where a crossbow through the eye will kill you dead" but that's less kind than my perspective. I think games have improved in a lot of ways in their understanding of the balance of risk vs playability. But I also think a lot of that improvement depends on some sleight of hand. The GM wants the players to FEEL like they're at great risk, but doesn't want to the risk to actually be that great because it's disruptive to play. The best GMs are the ones who really can create the "Die Hard" victory, where heroes are bloody and broken but unbowed as they manage to land the final deathblow and win the day.
And that's awesome. I'm totally not knocking it. But what I am saying is that the idea that combat is materially different than other challenges is a well-constructed fiction. The things you think of that make it different are just tricks, and tricks can be used on other things too.