Briefly, yes, assuming he declares something more complicated ("charge") than your initial proposal, and the guard doesn't also declare something more complicated than your proposal (like "I'll run away, shooting back if I don't need to Dash"--I generally wouldn't have a guard declare something this complex unless he was a veteran of lots of mobile skirmishes). If the Barbarian wins initiative, he gets an attack and an opportunity attack but winds up not in melee; if he loses initiative, the guard gets a free shot at him this round but next round they're in melee.
That outcome seems pretty reasonable to me on two counts, based on my fencing experience--you can almost always charge someone successfully (although you might get stabbed on your way in); and those who are thinking ahead about their options react faster and better than those who are just responding to what's right there in front of them in the instant. The Barbarian's player is being rewarded for player skill, which sounds reasonable to me. (And the Barbarian's player probably won't develop that skill right off--the first three or ten combats will probably play out as per your original post, which again is why I wouldn't have the guard automatically use sophisticated tactics either.)
If it doesn't sound reasonable to you, don't allow it.
^I didn't notice this post at the time, but this is an interesting conversation. Perhaps it would help to explain what I'm looking for in an initiative system and why:
I've played D&D with hardcore gamers who like to optimize, master a system, and play to win. I've played D&D with complete newbies who are just excited about roleplaying, and wouldn't so much as skim a rulebook even if I asked them to. In both of these groups I've played with, the cyclic initiative system has always been a problem.
For the roleplayers, rolling initiative broke the flow of the story. It broke the cohesiveness of the group, isolating each player to a one-on-one microconversation with the DM, rather than the group dynamic that makes the whole rest of the game so fun. It forced them to learn rules they didn't care about, to avoid the embarrassment of getting something wrong while the whole table is waiting for them to finish their turn. They weren't "listening," they were "waiting for their turn to talk." They didn't want to play a wargame (
certainly not a Chess game), they just wanted to go on an imaginary adventure with their friends. D&D combat is not fun for them.
For the powergamers, it slowed down the game. It made it a Chess game. They would wait for their turn to come around, assess the game state, figure out the most efficient use of their resources, and make the most optimal move. And that made combat take forever, it was really boring for everyone when it wasn't their turn, and everyone ended up complaining that "combat takes too long," and bought new editions on a promise of "faster combat." The thing is, I don't think these people really
wanted to play Chess any more than the roleplayers did--but the most
optimal way to play the game is to play it like Chess.
If the game is asking the player to choose between "winning" and "having fun," the game is poorly designed. The powergamer's goal is not to play the game in the most fun way, their goal is to win. If the path to winning isn't fun, it's not a fun game for them. Plot twist: I'm one of these powergamers. D&D combat is not fun for me.
The reason I really like the concept of declare/resolve is that it improves the experience for both of these groups. For the roleplayers, they can interact with combat the same way they interact with everything else, they can work together more easily, and the mechanics feel more like a story. For the powergamers, it changes the strategy (without removing strategic depth) from a Chess-like game that rewards forethought and mechanical precision, to a Poker-like game based on chance, probability, and bluffing. I don't know if it'll be any "faster" than combat with cyclical initiative, but I don't think that's even the real problem.
In other words, I think more hidden information and higher variance can lead to a game that feels more exciting and engaging for everyone, is easier for new or casual players to get into, and is just as strategic and interesting for more invested players.
those who are thinking ahead about their options react faster and better than those who are just responding to what's right there in front of them in the instant. The Barbarian's player is being rewarded for player skill, which sounds reasonable to me. (And the Barbarian's player probably won't develop that skill right off--the first three or ten combats will probably play out as per your original post, which again is why I wouldn't have the guard automatically use sophisticated tactics either.)
All I'm trying to say is, yes, it's good to reward player skill, but this is not a skill I want to reward. The more you reward forethought and mechanical precision, the closer it gets to Chess. The whole point of separating declaration and action (to me) is to reduce information and add variance. If you give a powergamer an opportunity to reduce variance and base their decision on more information, they'll take it, and try to "break" it. The process of learning that skill and being rewarded for developing it is great, but once you've mastered it, it makes the game more tedious (so it's actually kind of a punishment for everyone). And if some players have mastered it and some players haven't, the game becomes unbalanced. I want my game to be robust at all levels of play.
But I admit, I am using hyperbole to make a philosophical point, as a game design purist.
In a general sense, overkill is a positive feature, especially with ranged attacks. It's not like you know when you fire your arrow that somebody else is going to have fired another arrow at the target in the meantime and killed it. Nor will you necessarily even realize immediately when something has been fatally wounded, not until it topples over. Speaking as a DM, I'd be more resistant to someone trying to declare contingency ranged attacks than contingency melee attacks vs. dash, although I don't have any codified methodology to draw a bright line between what I would and wouldn't allow there.
Declaration/Resolution makes melee weaker since the enemy may not even be there. With range, it much less likely to matter if the guy moves first or not. And 5e already slightly favors ranged.
This is a very good point. How's this for a rule:
Ranged attacks are considered to be "fired" at the very beginning of the Resolution phase. The attacker has no control over the attack after declaring it. So, for example, if the target has moved behind total cover before the archer's initiative comes up, the attack misses. If another creature moves into the line of fire before the archer's initiative comes up, there is a chance for the attack to target that creature instead (let's just say equal chance for each possible target--if you're shooting into or across a scrum, there's a very low chance you'll hit who you're aiming at--and equal chance of hitting allies than enemies).
On the other hand, melee attacks are not executed until the initiative result at which they are resolved. An action declaration such as "I'll run in and attack that guy with my axe" can be given more flexibility during resolution (e.g., if the target is no longer there, the attacker can change his movement and attack a different target, or even take a Dash instead of an Attack action to get to the intended target).