Because HP doesn't determine my (or anyone elses) actions in the fiction. (Unless "die" is an action, I guess).
Having hit points above zero is generally a necessary condition for declaring an action in respect of a character. So hit points seem to me to play a big role in determining actions in the fiction.
Also, because a wide range of actions - jumping over a cliff, running through a burning building, trudging through a desert, etc - cause hit point loss, the availability of hit points is important to opening up various fields of action declaration.
HP measures how long I get to keep kicking before I drop.
That would be true if hit points dropped at a steady rate (and so were some sort of clock). But in fact hit point loss is heavily determined by actions declared (eg jumping over cliffs, charging into the rabble, etc). Which is to say, hit points are a resource that players can, in effect, elect to spend. (They often don't know the precise amount, because it will depend upon dice rolls - so it's like a gamble.)
The fiction (what the character sees) could describe hp damage as a bunch of equally valid things.
This seems to me to be a different matter. There is also the possibility of distinguishing between events of hit point loss, and the ongoing hit point total. For instance, many players who don't like martial healing seem to assume that if hit points that were lost have now been recovered, any wound caused has now healed. Whereas most players who like and use martial healing tend to take it for granted that while the
loss of hit points might mark an event of wounding, the
recovery of hit points - even of those very hit points (if that makes sense - it might not) - is an independent event that doesn't tell us anything about the wound that was taken, but tells us something about (say) the character being spurred on to greater effort by a word of benediction from a cleric, or a shout of encouragement from a warlord.
Come and Get It dictates the story. Glaringly.
CaGI it doesn't dictate the story any more than (say) rolling a successful grapple attack, which dictates that a target is grappled. Or rolling a successful Diplomacy check, which dictates that a target is friendly. Or declaring the casting of a magic missile spells vs a 1 hp giant rat, thereby dictating (with no need for dice rolls) that the rat is dead.
Unless the game is entirely determined by the GM, all player declarations of action dictate the story. That's what they're for. And D&D has never mediated all of them through dice rolls (eg spell casting in D&D typically succeeds with no need to roll dice - contrast most simulationist reactions against D&D, like RQ and RM, which do require dice rolls for spell casting).
CaGI is perhaps the single worst power ever written in D&D history. If there was ever a poster child for everything wrong with 4e, CaGI would be it. A quick review why.
* Unlike a large chunk of fighter powers, which can be described as "I hit and do X", CaGI is a very specific unique maneuver. That means EVEN IF we are willing to accept the "daily powers/encounter powers are just unique circumstances that arise, rather than some special maneuver I can only use once a fight/day", CaGI means once per day, like clockwork, the fighter meets the unique conditions to use that maneuver. Only once, mind you. If he uses it on the first room full of goblins in a dungeon, every other goblin in the complex is now magically immune to this trick. Until tomorrow, when they magically forget until a group falls for it again.
* CaGI overrides my control of NPCs without specific magical compulsion.
<snip>
* Its broken. In their infinite wisdom, the WotC team not only thought it was a good idea to give the fighter's PC control over my monsters, they didn't even think this was worthy of a Will save at first. Thankfully, the errata fairy fixed that one.
* Its unfair. You'd NEVER use this ability on a player!
CaGI it isn't especially unique. The fighter in my game shifts between multiple weapons - generally a maul and a polearm. When he uses the polearm, CaGI generally (but not always) correlates to him using his superior polearm skills to wrongfoot his enemies. When he uses his maul it might represent his enemies charging in. Plus there are other narratives too, depending on circumstances.
Furthermore, given the incentive in 4e to build characters around a "theme" (eg to exploit feat and item synergies), it is likely that a PC with CaGI probably has other forced movement powers as well. Which means that the manoeuvre performed using CaGI is sometime performed, in the gameworld, via different means at the table, namely, use of those other powers. (In my game, the player has Footwork Lure, plus several other close bursts that move their targets in various ways.)
Nor does it happen every day, nor every 5 minutes. It happens at best once per encounter, and the frequency of encounters is not even steady in the real world, and certainly doesn't occur in any metronomic fashion in the gameworld. Even if it did, would it matter? In AD&D combat between 1 HD combatants, no warrior ever kills two enemies less than 1 minute apart. Yet the game, and those who play it, seem to have survived this crime against verisimilitude. The game has an action economy and pacing mechanics - only OotS-like tables are going to read that stuff back into the minds of the inhabitants of the fiction.
As for overriding NPCs - killing NPCs overrides my control of them. Grappling does. Reaction rolls and diplomacy do. The point of being a player of D&D is to impact the gameworld. Combat rules are one means to this. Moving NPCs around is hardly a bigger impact on the gameworld than killing them.
Nor is CaGI broken, in my experience, because the forced movement only works if the target can come adjacent to the fighter, so it can't be used to pull enemies into pits etc. Changing it from auto-move followed by an attack roll, to a Will attack followed by auto-damage, just needlessly limits the narrative space of the power (eg it can no longer model skilled polearm work, given that it makes no sense that this would involve an attack vs Will) and also makes the STR bonus to attack illogical (why does your STR make your enemies more likely to close with you?).
Come and Get It is a good example of a power that would have gotten a tiny fraction of the ire it got if it had instead been on a magic using class instead. As a paladin power, for example.
I don't think it was a good idea for it to be in the first PHB.
Yes and no. Its absence might have made 4e marginally more palatable for some of those who didn't like it. But for me (and perhaps others like me, if there are any such people), it was a clear marker that WotC knew what they were doing with their game design and were serious about it. It was part of the assurance I was looking for in buying into the game (both literally and metaphorically).