Originally, Elemental was going to have continuous turn combat. That effectively meant real-time. Ultimately, after playing around with it, we decided to implement turn based (simultaneous turns based on combat speed) with tiles.
(1)
WINNER. TAKE. ALL. This is the part where we want to hear your opinions. We do ask that you keep an open mind on what we ultimately go with. My opinion is that the attacking player has the onus to finish the battle in N turns. After N turns, the attacker morale starts to go lower and lower at which point the defender can come out and make mince meat out of them. The question is, what should determine what N is? Or should we allow retreating? Should we allow draws? I’m against retreats or withdraws because it’s one of those things that allows the game to drag on. It’s a strong personal preference of mine that two men enter, one man leaves. (Your heroes will tend to escape though).(2)
Controlling the length of a tactical battle. We believe that users should have a lot of control over how in depth they want their battles to be. Should a tactical battle finish in less than a minute or should they last 2 hours? How do we make it so that players can control this?(3)
Randomization vs. Richness. I won’t lie to you, we have a trade off in front of us and it’s a big one. We can randomly generate the battlefields in tactical combat OR we can have it pick from a series of pre-made tactical battle maps. The randomly generated ones won’t be as interesting but they’ll more accurately reflect the local terrain. I’m preferring the pre-made ones because we can add some spectacular strategic when we’re crafting them and have hundreds to pull from. (4)
(1) I really wish for a clear and straightforward explanation here, because it's non-obvious to me what we're talking about. Maybe I'm just dense, but before we had 'continuous turn based' which was a label that did nothing but confuse me (Seriously. If it was 'effectively real time' then why haven't we just been calling it 'real time, but you can pause?) Now 'simultaneous turns'. What does that mean? There's an initiative order, where I move a unit, then you move a unit? We both choose our actions, and then the actions happen simultaneously when we're both done? What?
(2) and (3) On the one hand, you want to force the battle to fight to completion, so that one force must be completely wiped out before the battle ends. On the other hand, you want to give the users the ability to control how long the battle takes. Okay... I see these two goals as working at cross purposes. Will having winner-take-all battles with no retreat and no surrender really make the game shorter, overall? Even if it reduces the number of turns on the worldmap, it's potentially making the tactical battles drag on and on while I hunt down and kill every single enemy unit, even after they're clearly beaten. That introduces a 'mop up' phase in every single tactical battle, which I see as making the game overall play slower.
On the other hand, if you want to give players control over how long the battle lasts... a retreat option gives that. Being able to pull the plug when the battle has gone on long enough and it's obvious you're going to lose is, again, much quicker than having to fight until every single troop is dead. And it's much less frustrating than having to sit and watch my entire force be systematically destroyed - where's the fun in that?
I'm with the people who have suggested letting tactical battles continue over several strategic turns. After so many tactical turns, it cuts back to the world map and continues the next turn, with those units locked in combat until the fight can continue.
I also think that, if you want retreats to not drag the game out, then find a better way to handle them rather than just not having them at all. What you want to avoid is the scenario where you attack, the enemy retreats, and you're forced to chase them across many turns, trying to whittle them down. So, to prevent that: Don't make retreating a trivial exercise. It shouldn't simply be a matter of moving your units to the exit and canceling the tactical map, it should affect what you can do with that army afterwards.
I think it might be a good idea to move the after-battle mop-up phase to the strategic rather than the tactical level. When a commander sounds the retreat, their troops all run away independently - they're no longer a single, disciplined army, but rather scattered small groups who need to regroup and consolidate before they're a fighting force again. Maybe have a calculated percentage of your troops desert entirely, maybe have the troops return to the nearest town after a time, maybe have them scattered around the world map in small units that the enemy can more easily hunt down and destroy.
(4) Accurately representing the local terrain, and providing a variety of different landscapes to fight on, is much more important to me than having hand-crafted battlefields with lots of detail. Reflecting the landscape of the world map means that there's an important strategic element to where you fight your battles, as opposed to 'Oh, I'm fighting on Map #6 this time'
However, I'm not so keen on pure randomness. In particular, if I fight a battle at a particular location on one turn, I don't want to come back the next turn and fight another battle on the same location only to get a different randomly generated map. I'd prefer for the second battle to be recognizably the same terrain as the first. So my preference is to have the battle map reflect the terrain as accurately and reliably as possible. If you can do that with hand-crafted maps, by making enough variety of maps that the game can pick one that will closely represent the local terrain - okay. If not, then go with the random solution.