The issue is that intent aside initiative (particularly individual initiative) does tend to play out like characters frozen in time in practice, despite the books saying otherwise and claiming that actions are assumed to take place simultaneously. I have seen characters evade a fireball blast because they hadn't taken their action yet so they weren't within the blast radius when the fireball went off, yet an ally standing right beside them at the beginning of the round did take the fireball blast cuz they "won" initiative so they already had time to move into melee before the fireball happened.
Pre-declared spellcasting resolves this: Everyone knows that the wizard is casting Fireball at point X. The character who beat the wizard's initiative notices this in time to move clear of the blast zone. The character whose initiative is worse than the wizard's reacts too slowly and gets fried.
I've also seen systems which address this sort of thing by saying that whoever has the
worst initiative goes first, but anyone with better initiative is able to interrupt actions by someone with worse initiative. So, again, the wizard casts and the character with better initiative can interrupt his action to move to a safe place, while the character with worse initiative than the wizard does not have that option. But that method is more complex and easily leads to sequencing confusion when you're four interrupts deep and have to remember who interrupted who interrupted who interrupted...
And before you say "but none of that is simultaneous!", that objection ignores the duration of actions. If casting a Fireball spell takes, say, 5 seconds, and moving clear of the blast zone takes 2 seconds, then there is time for someone with sufficient situational awareness to notice that you're casting (i.e., better initiative) and then move clear concurrently with the casting itself.
I like the idea ao sitting everyone down in initiative order and going around asking what people are going to do from low to high and then resolving things from high to low. I'm not sure how effective or fun this would be.
I've seen this used (both as a RAW suggestion and as a table convention) in systems with fixed initiative orders, such as the Strike/DEX Ranks in most BRP variants. I have a hard time seeing it work well in systems with rolled initiative (unless you only roll once at the start of the session and that order holds for the entire evening's play) because you'd constantly be playing musical chairs.
Oh, one more I didn't like, but I was just observing and not playing, and I'm not sure it was being run right. I watched a couple people playing Melee, where the faster guy always got to run around his opponent and attack him from behind. That seemed really absurd (given that it wasn't some kind of super hero or super martial arts game).
That's a common way to not run Melee right, which usually seems to result from taking the idea of initiative from other games and apply it to Melee, which actually has two different initiative systems working side-by-side.
First, there is rolled initiative, which is a straight d6-v-d6 side initiative system. The side that rolls higher wins initiative for the round, but that only means that they get to choose whether they want to move first or second. As you observed, in 95% of cases, you'll want to choose to move last, so that you can flank or get behind your foes. But this is
only movement at this point.
Second, after everyone on both sides has moved, attacks are made in order of descending DEX. So the guy with high DEX does always get to attack first, but he may or may not be in optimal position to make that attack, depending on whether his side moved first or second, which in turn depends on which side won initiative for the round.
(Another thing they did wrong is that Melee has strong engagement/ZOC, so that, once you're adjacent to an enemy and in one of that enemy's front hexes, you can only move by making a one-hex "shift", similar to the D&D3.x 5-foot step, so you can't run around to get behind your opponent at that point in any case.)
I think there's some talking past about different mechanics here. There are a number of different possible initiative systems
(1) Individual rolled initiative like D&D
(2) Fixed action order like HERO or GURPS, going in order of character stat
(3) Group rolled initiative like older D&D
(4) Arbitrary order like clockwise among players, usually in groups
(5) Rerolled initiative like Savage Worlds, where order is randomly determined every round
(6) Declare all actions, then resolve actions - like RuneQuest
(7) Individual initiative bids each round, like EABA2
In EABA2, at the start of each combat round, everyone secretly bids for their initiative and then acts in descending order, with ties resolved as simultaneous actions. Anyone who makes no initiative bid goes on initiative 0.
So, what's to stop you from bidding a million for initiative every round? Because you're rushing your action, your initiative bid is applied as a penalty to all actions taken that round. Bid 2, and you're doing everything at -2. Most of the time, you'll want to bid 0 initiative, then, to avoid taking penalties unless you either completely outclass your opponent (making the penalties irrelevant) or you really, really need to go first (forcing you to eat the penalties), and even then, you want to bid as low as you can while still beating the opposition's bid.
EABA1 also had an interesting twist on methods #1/5/6. You first declared the general type of action you wanted to take that round (e.g., "I'll shoot" - not who you're shooting or any other details) and then make an unmodified skill roll for the skill that will be used for your action, which functions as your initiative roll. The logic behind this being that an expert will generally be able to act more quickly than a novice, in addition to more effectively.