AI and DOW logic

joncnunn

Senior Java Wizard
Retired Moderator
Joined
Mar 17, 2008
Messages
8,621
Location
Missouri
One area in which the AI is still weak in this mod is that it's still extremely transparent about "surprise war". (e.g. In early Industrial Era ending a turn with four embarked units on the boarder plus quite a few more right behind) It would have been better off having that convoy stop just outside of visibility that turn to make it more likely the defender wouldn't shuffle his army that's guarding multiple possible fronts into position against the one it chose the turn before the war.

In addition, it appears the AI still doesn't reevaluate weather it's a good idea to actually DOW someone after they've started moving the troops when they arrive at the point to see if their assumptions about the defenders military strength are still valid. (e.g. that the target's military strength hasn't dramatically climbed from discovering a military tech and mass upgrading / that the target hasn't spammed a bunch of defensive pacts the turn before)

These are both things seen in the base game, but it would make the AI a stronger opponent if it would do these (the second one a higher priority.)
 
One area in which the AI is still weak in this mod is that it's still extremely transparent about "surprise war". (e.g. In early Industrial Era ending a turn with four embarked units on the boarder plus quite a few more right behind) It would have been better off having that convoy stop just outside of visibility that turn to make it more likely the defender wouldn't shuffle his army that's guarding multiple possible fronts into position against the one it chose the turn before the war.

In addition, it appears the AI still doesn't reevaluate weather it's a good idea to actually DOW someone after they've started moving the troops when they arrive at the point to see if their assumptions about the defenders military strength are still valid. (e.g. that the target's military strength hasn't dramatically climbed from discovering a military tech and mass upgrading / that the target hasn't spammed a bunch of defensive pacts the turn before)

These are both things seen in the base game, but it would make the AI a stronger opponent if it would do these (the second one a higher priority.)

1.) The AI can't know what you see. That would be cheating.
2.) The AI does this. A lot, actually.

G
 
I sometimes got caught up by surprising DoW. So i think AI is quite good at this.

And they are quite smart to reevalute dow if you move many defense troop to your border where they are attacking you. Observe a bit deeper and you will be surprised how good the AI is now.
 
The AI is very human :) If they smell weakness they go to war, even with not-overwhelming forces if they suspect they have a chance (and they do, usually, or just delay my plans badly).

From my ongoing (random/deity/continents/standard) game, here I was suspecting Maria got enough of me trying to wonderspam and settle in her proximity (notice my 'army' is busy elsewhere):
civ5lp59.png


She then tried the 'peaceful' approach :p

civ5lp59bis.png


Needless to say, she should have been less greedy :p That really gave me no option but go to war.
I rush bought a defender and built some forces, then fought 2 waves of archers/spearmen, some horsemen; terrain definitely played a factor, making me feel I could hold that assault forever; I was getting ready for a counter offensive when Casimir came knocking. Those two guys NW are the typical vanguard and I had to sign a white peace (feels bad for Harun UA :( ) to defend the unguarded front.
civ5lp85.png


Quite fun if you ask me; you are right in that 'surprise' attacks are easy to spot 1-2 turns ahead, but it's not like you can do much if not signing a peace on the other front/rushbuy some stuff, and otoh the enemy has to approach somehow with the slower units.

Also, while I was busy with Casimir in the west, Maria made me a demand for my truffles, peacefully sitting with some skirmishers on my southern border. I felt compelled to accept :D
 
The AI still makes demands? I haven't seen it in this patch at all. Unless it's a difficulty thing - I'm currently (attempting) Emperor
 
And they are quite smart to reevalute dow if you move many defense troop to your border where they are attacking you. Observe a bit deeper and you will be surprised how good the AI is now.
It's always fun when you get an intel report that an army is marching on one of your cities, and then you have to hold your breath to see whether they're actually going to follow through. :)

Demands seem relatively rare, though I do get them occasionally. Does the AI have logic for deciding whether to try making demands before it declares war, or is it just random?
 
It's always fun when you get an intel report that an army is marching on one of your cities, and then you have to hold your breath to see whether they're actually going to follow through. :)

Demands seem relatively rare, though I do get them occasionally. Does the AI have logic for deciding whether to try making demands before it declares war, or is it just random?
"I like that Venice city.. but his military might hamper my military.. what if he just paid my demands? win-win."
 
It's always fun when you get an intel report that an army is marching on one of your cities, and then you have to hold your breath to see whether they're actually going to follow through. :)

Demands seem relatively rare, though I do get them occasionally. Does the AI have logic for deciding whether to try making demands before it declares war, or is it just random?

Depends on whether or not the AI is hostile towards you v. war towards you in approach. War = conquest, hostile = damage. And damage = denounce, pillage, bribe against, deceive, spy on , demand, etc.
 
I sometimes got caught up by surprising DoW. So i think AI is quite good at this.

And they are quite smart to reevalute dow if you move many defense troop to your border where they are attacking you. Observe a bit deeper and you will be surprised how good the AI is now.

If it makes a difference; this was an ocean boundary against an established coastal city; we didn't share a land border.
The reasons I strongly suspect AI Napoleon didn't reconsider were that:
1. Backdating the minimum number of turns he'd have had to launch that taskforce indicated it was before I hit that key very late Rean tech that allows several unit upgrades.
2. His apparent target the turn he "surprise" DOWed was actually the best defended city I had at the moment (in large part because my forces had starting moving that turn; this city would have been about my fifth best protected the turn before.)
3. I had indeed signed 3 Defensive Pacts the turn before; one with his neighbor (Maya) ; the other two with my own neighbors. (Not that I really needed the help ; I just wanted to ensure Napoleon wouldn't bribe my neighbors and also given the same distance constraints and tech I'm not really in position to take French cities either.)

I did see a bit more info on the diplomacy screen; Napoleon actually thought his army was superior to mine; it might have been a good move by him if he had actually shared a land border with me instead of being on the other major land mass 10+ turns away.
He did respond the next turn by attempting to withdraw his forces (extremely unsuccessfully since that's when the rest of my Frigates that weren't involved the first turn showed up) ; I had visibility into the Mayan navy annihilating what would have been his second wave. (Knocked him down to "near equal")
Next turn he was down to weaker than my army ; and I started redeploying my army back to where it was, in part because my "too easy, this could be a feint" alarm bells went off.

So I think it's mostly the military of those that would be brought in by Defensive Pacts it doesn't give adequate weight towards.
Come to think of it a few turns later, I'm now more concerned that the Mayas are going to grow too big by conquering France (along with the ex-Roman territory that France removed from the game; Portugal brought back as a 1 city vassal, and Maya eliminated again) than anything else; but unfortunately the AI still appears locked out of making peace due to the old locked-war for X turns after its declared ; his military rating is now "very weak" compared to mine.

One thing I was impressed is how fast the AI put hammers into the treasure fleet; it was completed in only four or five turns.

On the first screen shot, that's indeed to me the part of the game in which the AI is most challenging.
 
Top Bottom