Using simulation "send troops" gives siege defenders incredible advantage compared to actually playing the mission where defenders barely have any advantage. So for example 1200 attackers vs 600 defenders on lvl 3 walls. In simulation the defenders win, barely sustaining 300 casualties. If I choose to play the mission instead the defenders stand no chance no matter how many extra kills I get as player or if I try to command the defense by myself without the help of AI. How am I supposed to play a game that encourages me to not actually play it ?
You need to somehow add that huge advantage that defenders have in battle simulation into the actual mission as well if player chooses to play it otherwise whats the point of ever commanding a siege defense or using battle simulation while being the attacker.
To give context to your statement that in autoresolve, defenders can beat a force twice their size with ~50% casualties, I ran a series of custom battles with the following parameters:
Battle size was set to High (500 spawned) on some battles, and Very High (600 spawned) on others. (Defenders did better on High.)
Defenders had 400 men (Empire 25% inf 25% ranged 25% melee cav 25% ranged cav), level 3 fortifications, two ballistae, and two catapults.
Attackers had 800 men (Empire 25% inf 25% ranged 25% melee cav 25% ranged cav), siege ram, siege tower, one ballista, and one onager.
I just let the battles run without interfering, but a good player could inflict 1-100 more casualties depending on how good they are with the catapult.
On Vostrum, defenders won with 159 men left (~60% casualties).
On Charas, defenders won with 68 men left (~80% casualties).
On Jalmarys, attackers won with 13 men left (100% casualties/routs, 10 remaining defenders retreated to the keep).
On Syronea, defenders won with 105 men left (~75% casualties).
On Quyaz, attackers won with 20 men left (100% casualties/routs, 5 remaining defenders retreated to the keep).
On Seonon, defenders won with 106 men left (~75% casualties).
On Sibir, attackers won with 30 men left (100% casualties/routs, 52 remaining defenders retreated to the keep).
On Chaikand, defenders won with 80 men left (~75% casualties).
So I would overall agree that defenders need more advantage in real sieges to compete with autocalc sieges. Your point of being discouraged from actually fighting the battle is 100% valid.
However, we should also consider that if the defenders have
too much advantage that will make the game even grindier for the player, who has to besiege over 100 times to win the game. So, it could be argued that autoresolve defense calculation should just be reduced down to match the outcome of real battles. But that would increase snowballing by making the AI take territory faster, and would be unrealistic too - siege assaults in real life had a strong defender bias.
So I think the best thing is to meet in the middle; some small buffs to defenders in real battles, and a small nerf to the force multiplier of fortifications in autocalc battles.
Some ways the defenders can be given more tactical advantage by their fortifications:
* Make siege ladders need to be carried to the walls before they can deploy, rather than starting right at the base of the wall.
* Make it quicker and easier to push away siege ladders,
or, make siege ladders destructible. Currently they are the only siege tool which is invulnerable, but if they were destroyable and all got destroyed, the attackers would be forced to go through the front gate or retreat.
* Make AI troops have more difficulty defending themselves in melee combat while climbing a siege ladder? Human players can stay at the same level of effectiveness as now.
* Slightly increase the HP of the inner gate.
* Reduce the number of anti-arrow barricades attackers get.
I think a few of these should be enough to buff defenders in sieges to see similar results to autoresolve.
