Organizing city capture by AI in the middle of the game

[keu]

I want to simulate some historical actions.
The task is to give to AI plenty units to capture specified cities but not too many!

In the middle of the game I do not know how these cities are defended.
So I think to create for an AI plenty 'conqueror' super-units (say 6 or even more) with very high attack value and kamikaze flag. But how to neitralize unused 'conquerors' if defender civ had a few defender units in the captured city? I do not want AI to use these 'conqueror' super-units later in the game.

I thinked of some solutions but still don't test it.
1) make conqueror air unit with very short range (1 or 2) hoping stupid AI will go out of fuel. Will it?
2) assign conquerors to special home city which will be destroyed or change terrain to disband conquerors via lack of support.
3) upgrade conquerors by Leonardo's Workshop to some harmless.
4) don't use conquerors but diplomat hoping AI to incite a revolt.

All of these solutions are not as well as I want. May be there are another ways and I shouldn't to re-invent the wheel?

I'm sorry for my terrible English. I hope you understand what I mean.

 

[McMonkey]

I think the short range air units would make most sense, except air units cannot capture cities! You would have to include some normal land units to enter the city. I think you would need to test this to see if the AI plays along!

 

[Ingvar VII]

Try playing around with unit costs too. I have never tested to confirm but I think the computer likes to disband units that have disproportionally high shield costs versus actual stats.

 

[Nikoagonistes]

Well, in my Sobieski scenario, when I wanted the AI to decidedly take a city, I did the following:

1. Find a place where the AI is all but guaranteed not to go (preferably one of the distant corners of the map.)
2. Ensure there are two adjacent, empty squares in that place.
3. Place a unit belonging to the attacking civ in said place. The unit should be able to move only in the two squares.
4. On the turn where you want the attacking civ to take the city, have a barbarian
"trigger unit" spawn in the two-square space. (This unit can either be a unique unit or, if you're running out of unit slots, a weak, no-attack unit that the barbarians normally wouldn't have.)
5. Have an event where, upon the attacking civ destroying the barbarian trigger, large armies belonging to the attacking civ will spawn around the cities you want it to take.

It's basically a more elaborate version of the "spawn enemy units right next to the desired city" technique, except by having the invading armies spawn at the trigger, it ensures that the attacking civ can move the new units while it's his turn to move. A simple "@IFturn=,@THENCreateUnit" event, on the other hand, would simply go off at the beginning of a turn and thus might give the defending civ a chance to attack the attacker's exposed unit stacks.

Did I actually make sense there, or am I clear as mud?

(BTW: This idea isn't originally mine. Some others suggested it in my Sobieski thread back at Apolyton - I think I might check there.)

 

[keu]

I think the short range air units would make most sense, except air units cannot capture cities! You would have to include some normal land units to enter the city. I think you would need to test this to see if the AI plays along!

It is a strange behavior: AI want not use air congueror units were created near the enemy city. Units just stay at place, don't attacking and don't going out of fuel. Yes, normal land unit able to capture city was created as well.

Conqueror, nil, 1, 1.,1, 15a,1d, 3h,3f, 4,0, 0, Plu, 001000000000000

Just change domain from air to land and city become attacked and captured.

UPDATE: 1-move air units without kamikaze flag work better. They either attack a city or goung out of fuel next turn. It seems that your sense is good
Thank you, McMonkey!

 

[keu]

Try playing around with unit costs too. I have never tested to confirm but I think the computer likes to disband units that have disproportionally high shield costs versus actual stats.

I have tried. AI don't want to waste 900-shields-cost kamikaze units and just not use it for attack but drawing back

 

[keu]

It's basically a more elaborate version of the "spawn enemy units right next to the desired city" technique, except by having the invading armies spawn at the trigger, it ensures that the attacking civ can move the new units while it's his turn to move. A simple "@IFturn=,@THENCreateUnit" event, on the other hand, would simply go off at the beginning of a turn and thus might give the defending civ a chance to attack the attacker's exposed unit stacks.

I've understood your idea. But is it necessary if attacking civ have less number then defender, i.e. moving first in each turn?

 

[Ingvar VII]

Hm, I guess it wasn't so good an idea.
In Kobayashis Dominion War scenario many AI planets start out with A "Shield" unit. They almost invariably disappear before the city is attacked. I have never really tried to find out why exactly, but the only reasonable explanation is that the AI does indeed disband them. They cost much more than they are worth anyway.
Perhaps it has something to do with role?

 

[AGRICOLA]

Well, in my Sobieski scenario, when I wanted the AI to decidedly take a city, I did the following:

Did you use the name "GhostOfDisco" at Poly?

Sobieski is an excellent scen and a lot of fun to play.

 

[Nikoagonistes]

Did you use the name "GhostOfDisco" at Poly?

Yes, I did. (I think I might add that detail to my sig.)

I'm currently planning to release the complete Sobieski scenario on Sept 12 of this year. (The 12th because Sobieski won the Battle of Vienna on that day bck in 1683, but I'm not sure, as it's right after the 10th anniversary of 9/11...)

 

[McMonkey]

The 12th sounds like a good plan to me!