A Complete Ongoing Guide to Abusing the AI

[MadDjinn]

Anyone know if the Greeks make pike armies pretty much the same way others make archer armies (when at war). Do does the higher hammer cost slow them down enough that you can attack and take your time and not have to worry too much about them armying up?

With other civs you have the option of warring, letting them send units and then you make your HAs vet or elite and kill their archer armies. However for Greeks I can see how that would backfire if they army up their pikes while you are trying to become elite.

in 99% of the games, they start with making a hoplite first. so you can walkin at a later date than most. the other 1% of the time they do warriors for absolutely no reason and forget to make the hoplite.

otherwise, they'll act like other civs and make warriors rather than hoplites if you're at war with them for a long time early game. They also like to grow rather than tech or make units. I've watched greek AIs go to 6 pop right away, they do get BW first for no reason, just on growth with 0 tech or forests worked after .

[FF-GTR]

otherwise, they'll act like other civs and make warriors rather than hoplites if you're at war with them for a long time early game.

What determines if an AI makes warriors or archers? Do they only build archers if you are attacking their city, and otherwise build warriors to send out?

[Hellogoodbye123]

What determines if an AI makes warriors or archers? Do they only build archers if you are attacking their city, and otherwise build warriors to send out?

if you declare war on them before around 3000bc, they will not tech at all and just make warriors.. could help to get them with horses, and vet your horses.. but again.. you might want 25g..

[FF-GTR]

if you declare war on them before around 3000bc, they will not tech at all and just make warriors.. could help to get them with horses, and vet your horses.. but again.. you might want 25g..

How early do you have to war on them to put them into "warrior mode"? Since I've definitely forced them to make archer armies sometimes instead (bad).

And with the greeks, would it be the same dates for hoplite armies?

[Hellogoodbye123]

How early do you have to war on them to put them into "warrior mode"? Since I've definitely forced them to make archer armies sometimes instead (bad).

And with the greeks, would it be the same dates for hoplite armies?

as early as possible i guess, i never make them do warriors, i usually like the 25g instead...

and idk about greeks.. the like to make pikemen remember, so walk in are fairly easy if you moved around with your civ

[HydroDragon]

I rarely see greek hoplite armies, except when they start on an island or something.

They usually start off by building a hoplite then building warriors to explore, from my experience.

A greek capital is a love/hate thing. On the one hand, 7.5 defense will stop your warrior rush but on the other hand they'll only have the one hoplite so if you do get past it you get an extremely versatile city and maybe even Democracy.

[Salmeen10]

How early do you have to war on them to put them into "warrior mode"? Since I've definitely forced them to make archer armies sometimes instead (bad).

And with the greeks, would it be the same dates for hoplite armies?

I use this tactic alot, and it helps alot against civs with good starting techs or if you had a bad start and didnt get your HA early, or even if you want to take them with warriors. You have to declare war before they get BW, which they usually get at 2400bc I think. I declare war in 2900-2800 if I cant get walk-in and try to take them as soon as possible. If they are romans or anyother civ with good starting tech you wont have to get BW yourself just to get their good tech. You have to be ready to kill around 3 warriors, which can make your army vet/elite. Dont camp their forest otherwise they will continue researching BW. The aztecs and chinese get BW early(maybe 2900bc) so you have to be carefull with them.

[Pedal2Metal]

Morte,
Is this thread already in the strategy archive? It seems like a worthy addition with generally useful information for the community if not. I know you've covered a lot of this already but it's convenient to have it all in one thread.

best regards,
Pedal2Metal

[FF-GTR]

I second the notion this gets a link in the strategy archive. Along with the other thread P2M pointed out.

Also to add something. China does not work like the other civs. Most likely it has to do with being 3 pop (perhaps HGB Thebes is the same). I just declared war on them around 3300 to stall their BW. However they continued to research. Perhaps because they saw I had no unit nearby. But in any case, they never sent any warriors my way. I wound up killing 2 warriors and 3 archers with my HA only to die 9v5 around 2000. What a frustrating game that was!

I also notice that the chances of getting a walkin on Bejing between 3500 and 3000 seem way lower. I'm not convinced they are standard 3500 and 3000 as said in the first post. I have happened to run into them on peninsulas a lot later (In this case, for any AI civ the first warrior often ends up back in the cap circa 3200 after hitting a dead end in their first direction). However I'm not sure it's just the peninsula. Thoughts on China AI?

[TyShine]

China does not get bw faster but it will get AAs without having war declared on it. It works balanced to start but can do things faster if war is declared upon them. They almost always have two archers and two warriors within their cap if not a full arche army. They are an AI to kill early or be left alone. Unless it is very necessary do not risk a 6 (nonvet ha) v 5 archer as the hit points are not in your favor. If you can take them early great, but if not find something better to do with your horsearmy.

[TyShine]

I might do a player match with china as my teammate to see how they act. I'll post my results when or if I do.

[FF-GTR]

China does not get bw faster but it will get AAs without having war declared on it. It works balanced to start but can do things faster if war is declared upon them. They almost always have two archers and two warriors within their cap if not a full archer amry. They are an AI to kill early or be left alone. Unless it is very necessary do not risk a 6 (nonvet ha) v 5 archer as the hit points are not in your favor. If you can take them early great, but if not find something better to do with your horsearmy.

I got it up to vet against a warrior and his first 4 def archer. So when it died it was a 9-5. I knew it was risky. However I didn't see anything nearby better to do and felt I needed an AI cap to be competitive. I also wanted to get the HA up to elite to then hit Grizz with it. Was hoping for march or blitz. I knew if I just turned one non vet HA against Grizz, who already had a walk in and more gold than me, I would lose that matchup most likely. However with an elite HA and 4 pop bejing, maybe I could have pulled something out of the bag.

Anyway I'm sticking with my story. China and I were in a state of war since around 3300 (maybe 3200), yet he gained BW. I was not camping his trees. This is unusual AI behavior from the understanding in this thread.

I wish I had paid more attention to which year exactly he switched from 1 to 2 techs. I was so convinced all I needed to do was declare war that I didn't check every turn to see his tech count. But when I noticed he was working two water tiles even though we were at war I got worried. I went to peace and then war again, but he already had BW by the time my HA got there.

And he definitely never sent a warrior my way. He did have another warrior roaming a different direction.

[ScottieX]

China does not get bw faster but it will get AAs without having war declared on it. It works balanced to start but can do things faster if war is declared upon them. They almost always have two archers and two warriors within their cap if not a full arche army. They are an AI to kill early or be left alone. Unless it is very necessary do not risk a 6 (nonvet ha) v 5 archer as the hit points are not in your favor. If you can take them early great, but if not find something better to do with your horsearmy.

doesnt that depend on the state of china. if they jsut have a archer and a warrior (lets say) and not able to build another next turn you have jsut over 1 in 4 to get the archer whils reserving hte ability to retreat and then the cap and probably vet and probably no damage. if you get the 3/4 wounded then you waste 2 turns with the ha (1 attk, one rest).

Of course if he has mroe archers you might want to rethink it but even then if your going to go vet or elite and no immediate need for the HA it might still be good.

and aside from that in the case of 2-3 A im thinking walkout... I'll look for a chance to practise that.

[TyShine]

doesnt that depend on the state of china. if they jsut have a archer and a warrior (lets say) and not able to build another next turn you have jsut over 1 in 4 to get the archer whils reserving hte ability to retreat and then the cap and probably vet and probably no damage. if you get the 3/4 wounded then you waste 2 turns with the ha (1 attk, one rest).

Of course if he has mroe archers you might want to rethink it but even then if your going to go vet or elite and no immediate need for the HA it might still be good.

and aside from that in the case of 2-3 A im thinking walkout... I'll look for a chance to practise that.

If you have a vet Horsearmy then by all means go for it. However, risking a non vet horse army against a fortified archer is foolish unless its absolutely necessary.

What are you talking about 3/4 wounded? I was referring to horsearmies. If you want to risk warrior armies go for it because they are too slow in most cases to be used else where and only cost 30 hammers.

[FF-GTR]

doesnt that depend on the state of china. if they jsut have a archer and a warrior (lets say) and not able to build another next turn you have jsut over 1 in 4 to get the archer whils reserving hte ability to retreat and then the cap and probably vet and probably no damage. if you get the 3/4 wounded then you waste 2 turns with the ha (1 attk, one rest).

Of course if he has mroe archers you might want to rethink it but even then if your going to go vet or elite and no immediate need for the HA it might still be good.

and aside from that in the case of 2-3 A im thinking walkout... I'll look for a chance to practise that.

I agree in most cases attacking 6 v 5 is a bad idea actually. In that case better to come with two armies, or try to go vet up, or just threaten for gold.

I had 6 v 4 then 9 v 5 and felt I needed to take some chances given the map situation (good human opponent who already had an early cap, I was Russia so no bonus help coming up in later eras).

[TyShine]

I got it up to vet against a warrior and his first 4 def archer. So when it died it was a 9-5. I knew it was risky. However I didn't see anything nearby better to do and felt I needed an AI cap to be competitive. I also wanted to get the HA up to elite to then hit Grizz with it. Was hoping for march or blitz. I knew if I just turned one non vet HA against Grizz, who already had a walk in and more gold than me, I would lose that matchup most likely. However with an elite HA and 4 pop bejing, maybe I could have pulled something out of the bag.

Anyway I'm sticking with my story. China and I were in a state of war since around 3300 (maybe 3200), yet he gained BW. I was not camping his trees. This is unusual AI behavior from the understanding in this thread.

I wish I had paid more attention to which year exactly he switched from 1 to 2 techs. I was so convinced all I needed to do was declare war that I didn't check every turn to see his tech count. But when I noticed he was working two water tiles even though we were at war I got worried. I went to peace and then war again, but he already had BW by the time my HA got there.

And he definitely never sent a warrior my way. He did have another warrior roaming a different direction.

Go ahead and attack with vet horsearmies. 9-5 is a rare loss. I wouldnt declare war on them as their three pop works different. Ive found avoiding contact is best with them.

[FF-GTR]

Go ahead and attack with vet horsearmies. 9-5 is a rare loss.

That's what I was thinking! I had 6 v 4 then 9 v 5.

It was however his 3rd or 4th archer or something that killed me. He got some gold from a barb or hut or something, and I had to sell a caravan to him to get the money for the HA.

Actually, maybe he popped BW from a hut.

Oh well I think Grizz had that game anyway.

[ScottieX]

If you have a vet Horsearmy then by all means go for it. However, risking a non vet horse army against a fortified archer is foolish unless its absolutely necessary.

What are you talking about 3/4 wounded? I was referring to horsearmies. If you want to risk warrior armies go for it because they are too slow in most cases to be used else where and only cost 30 hammers.

hmm sorry I wasn't clear.
you attack with your HA and retreat if you get wounded, if you get wounded you make peace and rest (try to get your 25g or whatever you want). there is a chance (somthing like 25% or 20%) that you will beat him with no wounds. if that happens you then beat his warrior and take the cap.
So most likely (3/4... maybe 4/5 or so) you end up retreating and you waste one move on the HA and one turn of moves on the HA. If you are lucky you get the cap.
0% chance of loosing the HA.

only applicable if you know he only has one archer and cant build the next next turn - because then it becomes a different problem.

[FF-GTR]

3 pop definitely makes a difference in the AI behavior.

Some ranked H2H game today, I declared war on the germans very early, 3300 or something. Once they hit 3 pop, I noticed they started working two water, even though we were at war (both zulu and I had nearby units as well, although not adjacent to his city).

I wound up camping his water a bit with a galley to slow down his archers, and got an HA there before he had an AA (but not before single archers). Zulu opponent quit right before I was about to take Berlin.

But anyway, that's why china's different, I think. 3 pop worker management for AI is not the same as 2 pop.

So I see declaring war not as a way to halt BW research, but just slow it down.

[Salmeen10]

3 pop definitely makes a difference in the AI behavior.

Some ranked H2H game today, I declared war on the germans very early, 3300 or something. Once they hit 3 pop, I noticed they started working two water, even though we were at war (both zulu and I had nearby units as well, although not adjacent to his city).

I wound up camping his water a bit with a galley to slow down his archers, and got an HA there before he had an AA (but not before single archers). Zulu opponent quit right before I was about to take Berlin.

But anyway, that's why china's different, I think. 3 pop worker management for AI is not the same as 2 pop.

So I see declaring war not as a way to halt BW research, but just slow it down.

Weird. I do it very often and they always switch to 4 prod and 2 food. If I camp their forests they research BW, otherwise they dont. I dont remember the AI doing something different. Maybe it depends on how many warriors they produced and how much gold they found.

[elthrasher]

Weird. I do it very often and they always switch to 4 prod and 2 food. If I camp their forests they research BW, otherwise they dont. I dont remember the AI doing something different. Maybe it depends on how many warriors they produced and how much gold they found.

AI behavior just isn't that consistent. A couple of weeks ago I watched the AI move an archer out of a city leaving it empty for my March warriors. How often does that happen? What possible reason could there have been?

[FF-GTR]

AI behavior just isn't that consistent. A couple of weeks ago I watched the AI move an archer out of a city leaving it empty for my March warriors. How often does that happen? What possible reason could there have been?

I actually think it's 100% consistent, if only we could figure out the rules. Now I'm explicitly distinguishing between "consistent" and "sensible".

"Rules" may also be too complex to keep in your head, if its heuristics and such are tricky or evaluate a lot of state over time. Although they generally don't seem to be that complex when we figure them out.

This is in contrast to something like attacking 9v6, which is only consistent over a large sample - you can't predict reliably in a single instance. (Due to the fact it is affected by random seed)

But I've thought about this way too much...

[elthrasher]

I actually think it's 100% consistent, if only we could figure out the rules. Now I'm explicitly distinguishing between "consistent" and "sensible".

"Rules" may also be too complex to keep in your head, if its heuristics and such are tricky or evaluate a lot of state over time. Although they generally don't seem to be that complex when we figure them out.

This is in contrast to something like attacking 9v6, which is only consistent over a large sample - you can't predict reliably in a single instance. (Due to the fact it is affected by random seed)

But I've thought about this way too much...

You don't think there's any inexplicable randomness in what the AI will do? Honestly, in 2+ years of playing this game, I had never seen the AI completely walk out of a city before. Twice in that time I've rolled up on an AI cap in 2000 BC to find it empty. These things just aren't consistent with 99.9% of my experiences. I mean, yes, there may be triggers that cause the AI to abandon ship, but I can't imagine what they might be. I think it's got more to do with glitchy behavior or just random weirdness. That's what I mean by inconsistent.

[FF-GTR]

You don't think there's any inexplicable randomness in what the AI will do? Honestly, in 2+ years of playing this game, I had never seen the AI completely walk out of a city before. Twice in that time I've rolled up on an AI cap in 2000 BC to find it empty. These things just aren't consistent with 99.9% of my experiences. I mean, yes, there may be triggers that cause the AI to abandon ship, but I can't imagine what they might be. I think it's got more to do with glitchy behavior or just random weirdness. That's what I mean by inconsistent.

There might be some weird artifacts that seem unpredictable to a human point of view, but I feel they would be repeatable under the same circumstances. So not connected to randomness the way battle results are.

I can't prove it, but feel strongly due to what a few of us have discussed in the seed thread and also just my intuition from having had to analyze a lot of code. But I could be wrong of course.

Also even if I'm right that's not the exact same as being able to leverage the predictability either. From a practical point of view, if it's too complex or ridiculous to predict as a human, it might as well be random.

[FF-GTR]

Honestly, in 2+ years of playing this game, I had never seen the AI completely walk out of a city before. Twice in that time I've rolled up on an AI cap in 2000 BC to find it empty. These things just aren't consistent with 99.9% of my experiences. I mean, yes, there may be triggers that cause the AI to abandon ship, but I can't imagine what they might be. I think it's got more to do with glitchy behavior or just random weirdness. That's what I mean by inconsistent.

Oh and agreed on glitchy (just not random).

But I have seen the AI walk out a setter with their archer, leaving cap empty, and then move settler farther than archer, leaving cap empty and settler unprotected. If you had limited POV and couldn't see settler square, it might appear this was just walking the archer out of the city for no reason.

Or just a glitchy artifact of the AI programming.

And to be fair, I've seen human players do equally dumb things, and even done so myself once or twice, when taking a risk early and incorrectly evaluating the risk of walk-in. Perhaps the AI has a heuristic function to evaluate the risk of cap attack, and under a few circumstances it just gets it wildly wrong.

[elthrasher]

But I have seen the AI walk out a setter with their archer, leaving cap empty, and then move settler farther than archer, leaving cap empty and settler unprotected. If you had limited POV and couldn't see settler square, it might appear this was just walking the archer out of the city for no reason.

It could be. In the case of the archer that walked out, it wasn't the capital and it was well into the ADs. I don't recall seeing the population drop. It could have happened, but I think I would have noticed as I was playing the Mongols so any settled AI city was of interest to me. In this case, I had a March warrior army two tiles away but didn't want to risk it as I could get my Keshik there in two turns.

The times I found an empty AI cap in 2000 BC, I can't see how it could have been because of a settler. The AI never builds settlers that early.

[ScottieX]

It could be. In the case of the archer that walked out, it wasn't the capital and it was well into the ADs. I don't recall seeing the population drop. It could have happened, but I think I would have noticed as I was playing the Mongols so any settled AI city was of interest to me. In this case, I had a March warrior army two tiles away but didn't want to risk it as I could get my Keshik there in two turns.

The times I found an empty AI cap in 2000 BC, I can't see how it could have been because of a settler. The AI never builds settlers that early.

The AI doesnt differentiate between settlers they built 100g and ones that jsut wandered close to the city. That's why my trick to get archer armies to walk out by giving them a settler works. Also my trick of holding open walkouts with a bait unit.

The AI can build settlers that early - they could pop 100g.

but i have also had the situation where the ai walked out its only unit that didnt involve bait. There may be a chance of that in the programe, did the ai have any units roaming were you at peace?

[elthrasher]

The AI doesnt differentiate between settlers they built 100g and ones that jsut wandered close to the city. That's why my trick to get archer armies to walk out by giving them a settler works. Also my trick of holding open walkouts with a bait unit.

The AI can build settlers that early - they could pop 100g.

but i have also had the situation where the ai walked out its only unit that didnt involve bait. There may be a chance of that in the programe, did the ai have any units roaming were you at peace?

If the AI had had 100 gold, I would've gotten a fair share of that upon galloping in and I don't think that happened. No, I think the city was just empty for no apparent reason.

What happened in both cases was I saw the outlines of the city without having made contact at all. I readied my army and charged in hoping to catch the AI with just a single warrior defending. That happens sometimes if the AI spent that whole time thinking it was the only civ in the universe. Instead I found the city empty. I believe it was the Arabs in both cases and I found untouched barbs nearby as well, so I seriously doubt the AI had 100 gold. I think instead it just never produced anything due to some glitch. It never sent a warrior in my direction and missed barbs immediately on the other side. It must've got its prod stuck on Stonehenge or something. I don't know.

Here's the post where I wrote about one of these games:

http://forums.2kgames.com/forums/sho...postcount=1146

So you see I got barbs on both sides of Tripoli. No way the AI had 100 gold. If you keep reading, you'll also see some tips on surviving bombers from a really bad player. What a quality post!

[FF-GTR]

Here's the post where I wrote about one of these games:

http://forums.2kgames.com/forums/sho...postcount=1146

Maybe a wandering 3rd flag barb killed his initial warrior(s)? He might have had a warrior out and about when you took the city and it dissapeared. Perhaps the AI prioritizes having a wandering unit above a defended cap when not at war or aware of other civs. If the wandering unit dies, perhaps it will send out the cap defending unit.

[elthrasher]

Maybe a wandering 3rd flag barb killed his initial warrior(s)? He might have had a warrior out and about when you took the city and it dissapeared. Perhaps the AI prioritizes having a wandering unit above a defended cap when not at war or aware of other civs. If the wandering unit dies, perhaps it will send out the cap defending unit.

I suppose those things are possible, though you can't get an AI to walk out its 2nd warrior by killing the first and then going to peace (I know that's not quite the same). Normally when I come across an AI in this position 20 turns in, it has two warriors, one at home and one either exploring or at home.

I'm still going to favor glitchy behavior here. Maybe it had a warrior and it just vanished. Buggy code and unexpected output can NEVER be discounted as the cause for the unexplained in this game.

[FF-GTR]

I'm still going to favor glitchy behavior here. Maybe it had a warrior and it just vanished. Buggy code and unexpected output can NEVER be discounted as the cause for the unexplained in this game.

Absolutely.

Even could be an interplay between the two especially given the game progression to described. Aztec killed the arab wandering warrior, then aztecs get eaten by barb and dissapear, Arab buggily "forgets" there is danger in the world now that the aztecs are erased, sends warrior on a leisure stroll as if it were 3500 instead of 2500. Or some such nonsense.

[Zefelius]

I was away during the summer when this was first posted. Just wanted to say it's an excellent thread. Very nice work Duke---you've made some great contributions to this game, and this is just one example of that.

[FF-GTR]

Another tidbit maybe it's been said. If you settle next to an AI warrior, and don't rush your own (which would force a "meeting") and your workers are not next to his warrior but instead on the other side of your cap (which would also force a "meeting") then the AI will walk in on you. I think earlier it had been said AI won't walk in on a friendly human who has never warred except for a two step zulu AI - well it turns out there is another exception! Ouch!

Basically if they haven't "met" you they will walk in.

[Salmeen10]

Another tidbit maybe it's been said. If you settle next to an AI warrior, and don't rush your own (which would force a "meeting") and your workers are not next to his warrior but instead on the other side of your cap (which would also force a "meeting") then the AI will walk in on you. I think earlier it had been said AI won't walk in on a friendly human who has never warred except for a two step zulu AI - well it turns out there is another exception! Ouch!

Basically if they haven't "met" you they will walk in.

Today I noticed(from walking 9 turns ) that even 4 flags barbs dont take your initial settler. Also you cant sell your spies before you settle, but the barbs wont even steal them. you cant meet the AI with your initial settler, so they cant declare war and take it, instead they just stare with their warrior.

[sn1p3rk1ll3]

Today I noticed(from walking 9 turns ) that even 4 flags barbs dont take your initial settler. Also you cant sell your spies before you settle, but the barbs wont even steal them. you cant meet the AI with your initial settler, so they cant declare war and take it, instead they just stare with their warrior.

4-flag barbarians DO get out and grab yer settler!

Happened to me quite some times when learning the path of the settler walker

[ScottieX]

4-flag barbarians DO get out and grab yer settler!
Happened to me quite some times when learning the path of the settler walker

yeah Sal you were luicky but I suggest you DONT start a habbit of putting your settlers next to 4 flag barbs or our games will get even shorter than they already are...

[Zefelius]

I mean, yes, there may be triggers that cause the AI to abandon ship, but I can't imagine what they might be.

Maybe the AI picks it up from humans. I've played human opponents who leave their cities open for me even though we've made contact. Sometimes I'll think they're a freezer, but Nope, I roll in with a HA around 2000 or even later and they quit soon afterward!

[elthrasher]

Maybe the AI picks it up from humans. I've played human opponents who leave their cities open for me even though we've made contact. Sometimes I'll think they're a freezer, but Nope, I roll in with a HA around 2000 or even later and they quit soon afterward!

Those guys are great. Saved me again and again in my FFA days. Who the heck builds zero units when there are three human players who might wander up at any time? I'm totally on board with the idea of cutting corners to make your empire sleeker, more efficient, but when the whole thing can be toppled by a single warrior showing up at 0 or later, you've got issues.

[Star428]

I'll rephrase. As long as you are not at war, the AI won't walk into you city or take your settler. Probably until around 2000 is my guess.

That's exactly what i thought for a long time. Until one day when i was in a SP game and left one of my cities empty for just one turn and AI Spanish walked-in on it.

[ScottieX]

That's exactly what i thought for a long time. Until one day when i was in a SP game and left one of my cities empty for just one turn and AI Spanish walked-in on it.

yeah I've had that in multi before when I was doing a them or me sort of gamble. (ie i have a warrior by their cap they have one by mine - i figure they will be peaceful and let me kill them ) usually its worth the gamble...