Fafnir13
King
My major mode of conquest (early game) is to build up a few units to uberness (often heroes) and rely on them to take out the tough guys in opposing cities. Only, of course, when the odds are in the 98-99.9% range. If they aren't, a constant stream of sacrificial units is kept on hand to whittle them down until my big guy can gain the all important percentage of win.
What usually happens is that the tough guy will end up getting so strong that he gets 99.9% in almost all cases. This allows more of the sacrificial guys to fight the weaker opponents, building them up until they become uber city breakers as well. It's a really nasty and fun snowball effect.
If only the AI could use this. I had a game recently where, a nasty stack of Fawns lead by one with many stars and Orthus' Fricking Axe of Doom. They wasted my little army of defenders, but then collapsed into their usual stupidity. Fawns went wandering back toward their homeland, apparently forgetting the whole conquest thing. The really powerful ones ended up attacking into situations that left them horribly open on the next turn. Bit by bit, I was able to amass another army and take out their divided forces. After that, they didn't have any big guys left and all they could do was feed my guys free XP. Twas a real shame as they could have royally screwed me had their initial stack been reinforced and continued its advance into my lands.
Anyways, here's a few thoughts that might improve on the general AI performance in early game stackamancy.
1. Designate a heavy hitter and guard them well. This would be the guy with 5 stars and a few shock promotions. The AI should only be attacking with him on the high end of 90's, attacking with their low end units until their designated heavy can get the kills.
2. Designate the sacrificial minions accordingly. Weak guys with few to no promotions are often best sent in first if the big guys don't have the best odds. Of course, given poor enough odds (of 0.1% or so) chances are you'll end up not doing enough damage ( if any) to make it worth while. This leads into the next group.
3. Designate the mid-level troops and know when to use them. They've got a few promotions, but not enough to make them city breakers quite yet. When the sacrificial minions are looking at 0.1% odds, these guys might be in the 20-30%. Odd are they're dead, but at least they should go out doing enough damage to the opposition to allow the city breakers to do their work. Occasionally they even get lucky, netting fat stacks of XP which can help then climb the ranks to city breaker.
4. Count and compare stacks with defenders including odds and expectations. If I've got five warriors (on breaker, one mid, and three sacrificial) and the city's got three and my odds with the best guy are only 50% and my odds with my worst guys is only 13% but they've only got one tough warrior in the city with rest being a no star hacks, what should I do? My intuition in this situation tells me to back off and get more troops. Attack with the mid would probably weaken the tough guy, but not enough for the desired 99.9% on the breaker, meaning another attack afterwards with a sacrificial. This leaves me with two guys dead and my breaker's action taken up killing a weaker guy (as they would probably become the strongest in the city's stack). That means I've got two sacrificial guys to take on 1 opponent of equal strength and 1 weakened tough guy. It would probably take both of them to take out just one, meaning next turn I've only got my breaker and a sacrificial with a promotion to spend versus a tough guy in the city who probably just gained a level thanks to taking out a couple units. I could probably use the sacrificial to weaken again and take with the breaker, but that leaves him open for a counter attack. Worse yet, what if the city builds a warrior next turn? Complete loss at that point.
Hrmmm....where was I going with this? Oh yeah! The Ai thing. Um...I'm not sure how to make the AI go through the convoluted thought process I just meandered through. I imagine it can check the odds of various units so thresholds could be set. I'd also imagine it could count opposing units and it's own and make comparative estimates. Maybe not. I'd just be glad if it would look beyond the odds of a single attack and consider the future.
5. Attack coherently, not just at whatever's convenient, and think about where the attack will leave the unit. I cannot count the number of tough units I've lured onto deserts or off of hills and forests with a lone sacrificial unit. It is as though the AI does a little check of combat odds nearby and goes for the best one, ignoring the stack of slavering scariness that's just waiting for it to show a hint of weakness. Captured workers also work, but don't have the advantage of weakening the tough guy by a few, possibly crucial points. If the AI is attacking with a city breaker, it ought to be looking around at what can attack said city breaker on the next turn and what the potential odds of those attacks are. Say, strongest opponent within reach with a modifier of the total number of enemy units with additional modifiers for how many units the AI has to support the city breaker all coalescing into something that determines whether or not it will attack into a given square. I'm sure it's far more complicated than that, but at least that's a gist of it?
6. Set a goal and go for it. I've seen the AI do some really silly things with its units. The strongest guy will leave a city to escort a settler all by himself in the middle of a war whilst my stack of doom is not two tiles away. I've seen stacks of their powerful units charge off for my distant and well defended cities whilst a weakened stack of my own sits outside their capital. Worst yet was the aforementioned stack of fawns led by wielding that horrible Axe. They could have wiped out a few more of my cities and caused havoc had they only advanced instead of seemingly forgetting even the idea of conquest. Were their barbarians or something near their cities that made them decide that units needed to be sent back? Were they trying to assemble another stack, taking units not set on city defense back to rally point? Dunno.
If the AI is bent on conquest, it should be funneling units out to the nearest of its conquered cities, re-amassing the stack their and sending it out on its merry way to the next target. Newly captured cities don't need to be that well defended. Leave a few sacrificials and call it good. Any opposing attack will mean fewer troops bent on taking out your advancing stack and your minions are getting the bonus of being on the defensive. This may mean that you lose a newly conquered city and have to backtrack, but there are always more sacrificial minions marching in to reinforce and the enemy is expending effort in not fortifying their other cities.
If it's bent on defense, it should try to get as many troops inside the cities closest to invaders and launch stacks when appropriate. The target of these stacks needs to be the most powerful of the opposing units. Once you take out their city breakers, their left without the brute force necessary to take on a defended city. Mop up is what ensues at that point, leaving the defender with a good deal of strong units ready to make a counter attack. Ideally, only close troops or ones within a turn or two of a given threat should be altered significantly in their courses. No point in having your stacks in distant lands do a yo-yo act as threats far away appear and are dealt with before it has a chance to help.
7. The most important for early game AI: know when to build troops. All this talk of stacks and odds and city breakers versus sacrificial minions is meaningless if the AI only has a couple warriors and scouts guarding its cities. Some additional troop production should be happening around the turns that barbs start to advance on the settled world (tends to be turn 100-120 for me). If nearby Civs happen to have aggressive as a trait, a few more warriors on the defense is not a bad plan. Especially when said aggressive Civs adopt apprenticeship. I know I've rocked more than a few AI's with a quick rush to 1 star + Shock warriors. Them guys is lethal against anything caught unprepared.
I have noticed that the AI will build up troops once it goes to war with you. If you've properly set up your invasion, it's far too late.
Heh, just as an aside, I used to have rather embarrassing arms races consisting of me amassing warriors right outside their capital whilst they amassed them inside. I later realized that if I can't come in with overwhelming odds right from the beginning it's only going to lead to horrible stalemate that slows down both Civs while the rest of the world continues to advance. Live and learn.
I think that's about it for my random ramblings for the time being. Most of this (probably all of it) is probably basic stuff to most of you. I'm just hoping that by dissecting it a bit, it might help get some ideas flowing on how to get the AI a little bit better at the early game conquest that ends up influencing how the rest of the game goes.
If it does actually prove helpful, I'll pleasantly surprised.
What usually happens is that the tough guy will end up getting so strong that he gets 99.9% in almost all cases. This allows more of the sacrificial guys to fight the weaker opponents, building them up until they become uber city breakers as well. It's a really nasty and fun snowball effect.
If only the AI could use this. I had a game recently where, a nasty stack of Fawns lead by one with many stars and Orthus' Fricking Axe of Doom. They wasted my little army of defenders, but then collapsed into their usual stupidity. Fawns went wandering back toward their homeland, apparently forgetting the whole conquest thing. The really powerful ones ended up attacking into situations that left them horribly open on the next turn. Bit by bit, I was able to amass another army and take out their divided forces. After that, they didn't have any big guys left and all they could do was feed my guys free XP. Twas a real shame as they could have royally screwed me had their initial stack been reinforced and continued its advance into my lands.
Anyways, here's a few thoughts that might improve on the general AI performance in early game stackamancy.
1. Designate a heavy hitter and guard them well. This would be the guy with 5 stars and a few shock promotions. The AI should only be attacking with him on the high end of 90's, attacking with their low end units until their designated heavy can get the kills.
2. Designate the sacrificial minions accordingly. Weak guys with few to no promotions are often best sent in first if the big guys don't have the best odds. Of course, given poor enough odds (of 0.1% or so) chances are you'll end up not doing enough damage ( if any) to make it worth while. This leads into the next group.
3. Designate the mid-level troops and know when to use them. They've got a few promotions, but not enough to make them city breakers quite yet. When the sacrificial minions are looking at 0.1% odds, these guys might be in the 20-30%. Odd are they're dead, but at least they should go out doing enough damage to the opposition to allow the city breakers to do their work. Occasionally they even get lucky, netting fat stacks of XP which can help then climb the ranks to city breaker.
4. Count and compare stacks with defenders including odds and expectations. If I've got five warriors (on breaker, one mid, and three sacrificial) and the city's got three and my odds with the best guy are only 50% and my odds with my worst guys is only 13% but they've only got one tough warrior in the city with rest being a no star hacks, what should I do? My intuition in this situation tells me to back off and get more troops. Attack with the mid would probably weaken the tough guy, but not enough for the desired 99.9% on the breaker, meaning another attack afterwards with a sacrificial. This leaves me with two guys dead and my breaker's action taken up killing a weaker guy (as they would probably become the strongest in the city's stack). That means I've got two sacrificial guys to take on 1 opponent of equal strength and 1 weakened tough guy. It would probably take both of them to take out just one, meaning next turn I've only got my breaker and a sacrificial with a promotion to spend versus a tough guy in the city who probably just gained a level thanks to taking out a couple units. I could probably use the sacrificial to weaken again and take with the breaker, but that leaves him open for a counter attack. Worse yet, what if the city builds a warrior next turn? Complete loss at that point.
Hrmmm....where was I going with this? Oh yeah! The Ai thing. Um...I'm not sure how to make the AI go through the convoluted thought process I just meandered through. I imagine it can check the odds of various units so thresholds could be set. I'd also imagine it could count opposing units and it's own and make comparative estimates. Maybe not. I'd just be glad if it would look beyond the odds of a single attack and consider the future.
5. Attack coherently, not just at whatever's convenient, and think about where the attack will leave the unit. I cannot count the number of tough units I've lured onto deserts or off of hills and forests with a lone sacrificial unit. It is as though the AI does a little check of combat odds nearby and goes for the best one, ignoring the stack of slavering scariness that's just waiting for it to show a hint of weakness. Captured workers also work, but don't have the advantage of weakening the tough guy by a few, possibly crucial points. If the AI is attacking with a city breaker, it ought to be looking around at what can attack said city breaker on the next turn and what the potential odds of those attacks are. Say, strongest opponent within reach with a modifier of the total number of enemy units with additional modifiers for how many units the AI has to support the city breaker all coalescing into something that determines whether or not it will attack into a given square. I'm sure it's far more complicated than that, but at least that's a gist of it?
6. Set a goal and go for it. I've seen the AI do some really silly things with its units. The strongest guy will leave a city to escort a settler all by himself in the middle of a war whilst my stack of doom is not two tiles away. I've seen stacks of their powerful units charge off for my distant and well defended cities whilst a weakened stack of my own sits outside their capital. Worst yet was the aforementioned stack of fawns led by wielding that horrible Axe. They could have wiped out a few more of my cities and caused havoc had they only advanced instead of seemingly forgetting even the idea of conquest. Were their barbarians or something near their cities that made them decide that units needed to be sent back? Were they trying to assemble another stack, taking units not set on city defense back to rally point? Dunno.
If the AI is bent on conquest, it should be funneling units out to the nearest of its conquered cities, re-amassing the stack their and sending it out on its merry way to the next target. Newly captured cities don't need to be that well defended. Leave a few sacrificials and call it good. Any opposing attack will mean fewer troops bent on taking out your advancing stack and your minions are getting the bonus of being on the defensive. This may mean that you lose a newly conquered city and have to backtrack, but there are always more sacrificial minions marching in to reinforce and the enemy is expending effort in not fortifying their other cities.
If it's bent on defense, it should try to get as many troops inside the cities closest to invaders and launch stacks when appropriate. The target of these stacks needs to be the most powerful of the opposing units. Once you take out their city breakers, their left without the brute force necessary to take on a defended city. Mop up is what ensues at that point, leaving the defender with a good deal of strong units ready to make a counter attack. Ideally, only close troops or ones within a turn or two of a given threat should be altered significantly in their courses. No point in having your stacks in distant lands do a yo-yo act as threats far away appear and are dealt with before it has a chance to help.
7. The most important for early game AI: know when to build troops. All this talk of stacks and odds and city breakers versus sacrificial minions is meaningless if the AI only has a couple warriors and scouts guarding its cities. Some additional troop production should be happening around the turns that barbs start to advance on the settled world (tends to be turn 100-120 for me). If nearby Civs happen to have aggressive as a trait, a few more warriors on the defense is not a bad plan. Especially when said aggressive Civs adopt apprenticeship. I know I've rocked more than a few AI's with a quick rush to 1 star + Shock warriors. Them guys is lethal against anything caught unprepared.
I have noticed that the AI will build up troops once it goes to war with you. If you've properly set up your invasion, it's far too late.
Heh, just as an aside, I used to have rather embarrassing arms races consisting of me amassing warriors right outside their capital whilst they amassed them inside. I later realized that if I can't come in with overwhelming odds right from the beginning it's only going to lead to horrible stalemate that slows down both Civs while the rest of the world continues to advance. Live and learn.
I think that's about it for my random ramblings for the time being. Most of this (probably all of it) is probably basic stuff to most of you. I'm just hoping that by dissecting it a bit, it might help get some ideas flowing on how to get the AI a little bit better at the early game conquest that ends up influencing how the rest of the game goes.
If it does actually prove helpful, I'll pleasantly surprised.
.
