Gripping tactical battles is not an outcome.
Well, it is an outcome in the sense I intended it. I know it can be a desired outcome of play to enjoy gripping tactical battles, because I have that desire for that outcome.
I can tell you one fantasy RPG that does not produce gripping tactical battles, at least in my experience: Runequest. Very few tactical choices have to be made by a player either in character build or play, and the resolution of combat turns purely on the dice rolls. Runequest has a lot going for it as a system, but gripping tactics is not one of those things.
The fact that I can describe an RPG that doesn't deliver the desired outcome reinforces my conviction that it is an outcome of play that it makes sense to desire.
The outcome in this case would be to win or lose that gripping tactical battle.
Well, that's also an outcome. But as GM I don't particularly care whether the PCs win or lose - I just want gripping tactics! And, conversely, Runequest will also deliver wins or losses in battle, but it won't deliver gripping tactics.
Winning and losing battles is not the outcome that I'm interested in. It's the fact that the battles involve gripping tactics that I care about, and that (for my group) 4e delivers on (as does RM, but in a different way - and at the moment at least 4e is more satisfying for my group).
If those mechanics facilitate a path toward a destination known from the start then you don't really have a game.
This just isn't true. The mechanics of chess facilitate a path towards a destination know from the start - namely, resignation by one player (or checkmate in some cases) because the position is one in which his/her king cannot be defended from the other player's attack. It's still a game.
Cthulhu-based roleplaying games facilitate a path towards a destination known from the start, namely, revelations of cult activity that lead to confrontations between bookish investigators and strange creatures that threaten to overwhelm those investigators both physically and psychically. It's still a game.
Hard choices come when the actual outcome of the game is riding on them.
Well, I talked about a game in which the desired outcome is "finding out what it means to be a fantasy hero". Such a game of necessity will involve hard choices - if no hard choices are put in front of a player, s/he is hardly going to find out what heroism means! But those hard choices won't be ones that threaten the desired outcome - they will ensure that it is achieved. For example (a choice that has already come up in my 4e campaign), Do heroes make deals with slave traders to buy back their slaves, or do they rescue those slaves by beating up on the slave traders? My players went for the first option. Other players might very well feel that dealing with slave traders is never justified, even if it's the most convenient way of freeing the slaves in question. That's a hard choice (not hard like "Should I have kids or not?", but hard enough for a fairly low key recreational pursuit). The gameplay turns upon that choice. Ingame outcomes ride on it (eg do the slavers live to continue their evil trade, or not?). But the occurence of that choice does not put the overall goal, of finding out what it means to be a fantasy hero, into question. It makes realising that goal possible.
And what I've just described manifestly involves playing a game - it's a pretty bog-standard instance of playing an RPG.