AI STEAMROLLING TEST

I’ve never seen any behavior tree stop working because of some „higher thinking”.
And the fact that some of them get stuck (e.g. also settlers sometimes stuck and GWAMs) also confirms that. Because if there was such a mechansim eventually a stuck tree would be canceled and re-started. But nope.
I think, we agree! But there is a small misunderstanding: I did not mean eg. a running city attack operation to be stopped or disrupted.
With operation could be dismissed or delayed (again and again) I mean, that the „higher thinking” AI 'chews' about city attack operation (among several possibilities), but then dismisses that and does something other - so the (possible) city attack operation is not initiated this time.
That decision is again and again made against city attack operation: this operation is not started at all. So the (possible) city attack is postponed for too long and for too often.
 
So I came back from work today to see what happened and I was still on the mail loading screen (first stirrings of life), I forgot to click start game before I left, but that ended up being a good thing for my second test as I got to watch this in full!! So test #2 had the following result before the map got filled with GDR's.

pangea-test2.jpg

I started the game with 8 civs on a small pangea map to speed up the action (Deity difficulty). Unfortunately it ended being a crappy map (almost continent, with a smaill peninsula of mountains separating the two landmasses). I also wanted to change a few things in the test: I changed Walls up to Urban Defenses to 25-25-25-25 for strength. Basically a quarter of the strength as normal because I wanted to nerf ranged unit bombard strength to 15-30-45-60 from archers to machine guns so they couldn't one shot a city anymore. I did strengthen seige units and changed them slightly. All of their base strength and bombard strength +5. I also allowed artillery +3 range so they don't have to rely on observation balloons. I also did quite a few changes to the Operations.xml file including making larger maximum attack forces to go to war. Also made more requirements to bring seige units and use more air units if it's part of the attack force. Also adjusted naval slightly so they don't send 1 to 2 units on a suicide mission against a stronger military to try and take a coastal city. I also changed parameters to allow the AI to go to war even if they weren't strong enough and to stay at war and keep sending units at their particular target until they were a 7-1 underdog.

Other than that, I changed the entertainment complex to allow 25 amenities and the arena 25 amenities to a 10000 tile range. Basically copied that Escondido wonder or whatever and replaces 2 with 50.

So, I chose Canada as the human guinea pig in the autoplay to see if the AI would ever attack them. While I am typing this the screen shot above has changed and Canada has been wiped off the mat by Russia.

pangea-test-2(a).jpg

Game started off pretty normal, civs attacking city states, building up their army ect. Then India took a quick lead by taking two cities from Korea (India was on the west side of the right hand land mass, Korea was on the south west side of the same landmass). Then Spain (originally located on the South East of the right landmass quickly forward settled India. That caused a joint war with India and Norway (Located South on the right landmass) vs. Spain. Spain got completely swallowed by both except for two cities before they peaced out. The other civ on the right land mass was Kongo & Russia (Located far East and far North East respectively)

In the meantime on the left land mass, Shaka started on the far West of the second landmass and Brazil was far south. Zulu and Brasil expanded rapidly until Shaka got his Zulu corps and took 2 of Pedros cities before peace.

Then India DOW's Korea again and takes their capital and every city but 3 tucked away in mountains to the south. At this point I thought there was no way India could lose game as they almost had a third of their landmass.

I was sooo wrong. India gets DOW immediately by Russia, Brazil, Canada & Korea (although they were still at war). India then quickly takes a canadian coastal city with naval (eventually losing it to loyalty) and I thought the steamroll was on. Russia then starts ripping through India like a hot knife through butter, taking 5-6 cities within 10 minutes. At the same time, Korea takes back their capital and one other city.

Now Norway DOW's India and takes a couple cities that India conquered from Spain. They peace out for a bit and then Kongo gets into the action and DOW Russia, takes one city quickly, then Russia absolutely crushed them leaving only Kongo's Capital and that lowly city you see they still own at the bottom right on the second pic.

Before that war was even over Brazil DOW India and takes two cities just off the coast and actually maintains loyalty in them. India is only left with 2 cities. Soon after Shaka DOW Brazil and takes 3 cities before peace.

At this point I thought it would come down to Russia and Zulu.

Then Norway declares on Spain and wipes them out. Out of the blue, Brazil and Russia declares on Kongo and Brazil steals their capital with help from their unique battleship.

Things get peaceful for a while (a short while) before Zulu again declares on brazil. Well that was a mistake, Brazil just got Jet Bombers and GDR's and get massacred by Brazil and they literally lose 10-12 cities and are completely wiped out.

Now Russia must be jealous of the action because they DOW their BFF Korea and start shredding through them with thrir own GDR's, but now Norway declares on Korea as well and takes two coastal cities while Russia grabs their capital. Korea is gone!

Next up, Russia polished off Indias final 2 cities.

Only 5 civ's left: Brasil, Russia, Canada, Norway & Kongo on their own little island.

Russia then starts going after Norway with their GDR's and sends them packing.

The final battle then ensues (first pic to second pic). Russia takes back everything on their landmass and the nukes start flying. Canada gets involved and Russia walks their GDR's over there and take a couple cities from Canada but eventually lose them to loyalty. Russia peaces out with Brazil, stays at war with Canada and goes back across the pond and finishes them off (second pic).

Conclusion: This was by far the most entertainment iv'e seen in a game. Thinking one civ (India) has a guaranteed victory to getting wiped off the planet was quite enjoyable. And to see efficient use of naval, air & nukes was amazing. Unfortunately, in the end the only units left are about 100 GDR's, Jet Bombers and missle cruisers but I guess that's to be expected.

I going to try a continents game next on Deity and ACTUALLY play myself to see how I do.
 
So I came back from work today to see what happened and I was still on the mail loading screen (first stirrings of life), I forgot to click start game before I left, but that ended up being a good thing for my second test as I got to watch this in full!! So test #2 had the following result before the map got filled with GDR's.

View attachment 519247

I started the game with 8 civs on a small pangea map to speed up the action (Deity difficulty). Unfortunately it ended being a crappy map (almost continent, with a smaill peninsula of mountains separating the two landmasses). I also wanted to change a few things in the test: I changed Walls up to Urban Defenses to 25-25-25-25 for strength. Basically a quarter of the strength as normal because I wanted to nerf ranged unit bombard strength to 15-30-45-60 from archers to machine guns so they couldn't one shot a city anymore. I did strengthen seige units and changed them slightly. All of their base strength and bombard strength +5. I also allowed artillery +3 range so they don't have to rely on observation balloons. I also did quite a few changes to the Operations.xml file including making larger maximum attack forces to go to war. Also made more requirements to bring seige units and use more air units if it's part of the attack force. Also adjusted naval slightly so they don't send 1 to 2 units on a suicide mission against a stronger military to try and take a coastal city. I also changed parameters to allow the AI to go to war even if they weren't strong enough and to stay at war and keep sending units at their particular target until they were a 7-1 underdog.

Other than that, I changed the entertainment complex to allow 25 amenities and the arena 25 amenities to a 10000 tile range. Basically copied that Escondido wonder or whatever and replaces 2 with 50.

So, I chose Canada as the human guinea pig in the autoplay to see if the AI would ever attack them. While I am typing this the screen shot above has changed and Canada has been wiped off the mat by Russia.

View attachment 519250

Game started off pretty normal, civs attacking city states, building up their army ect. Then India took a quick lead by taking two cities from Korea (India was on the west side of the right hand land mass, Korea was on the south west side of the same landmass). Then Spain (originally located on the South East of the right landmass quickly forward settled India. That caused a joint war with India and Norway (Located South on the right landmass) vs. Spain. Spain got completely swallowed by both except for two cities before they peaced out. The other civ on the right land mass was Kongo & Russia (Located far East and far North East respectively)

In the meantime on the left land mass, Shaka started on the far West of the second landmass and Brazil was far south. Zulu and Brasil expanded rapidly until Shaka got his Zulu corps and took 2 of Pedros cities before peace.

Then India DOW's Korea again and takes their capital and every city but 3 tucked away in mountains to the south. At this point I thought there was no way India could lose game as they almost had a third of their landmass.

I was sooo wrong. India gets DOW immediately by Russia, Brazil, Canada & Korea (although they were still at war). India then quickly takes a canadian coastal city with naval (eventually losing it to loyalty) and I thought the steamroll was on. Russia then starts ripping through India like a hot knife through butter, taking 5-6 cities within 10 minutes. At the same time, Korea takes back their capital and one other city.

Now Norway DOW's India and takes a couple cities that India conquered from Spain. They peace out for a bit and then Kongo gets into the action and DOW Russia, takes one city quickly, then Russia absolutely crushed them leaving only Kongo's Capital and that lowly city you see they still own at the bottom right on the second pic.

Before that war was even over Brazil DOW India and takes two cities just off the coast and actually maintains loyalty in them. India is only left with 2 cities. Soon after Shaka DOW Brazil and takes 3 cities before peace.

At this point I thought it would come down to Russia and Zulu.

Then Norway declares on Spain and wipes them out. Out of the blue, Brazil and Russia declares on Kongo and Brazil steals their capital with help from their unique battleship.

Things get peaceful for a while (a short while) before Zulu again declares on brazil. Well that was a mistake, Brazil just got Jet Bombers and GDR's and get massacred by Brazil and they literally lose 10-12 cities and are completely wiped out.

Now Russia must be jealous of the action because they DOW their BFF Korea and start shredding through them with thrir own GDR's, but now Norway declares on Korea as well and takes two coastal cities while Russia grabs their capital. Korea is gone!

Next up, Russia polished off Indias final 2 cities.

Only 5 civ's left: Brasil, Russia, Canada, Norway & Kongo on their own little island.

Russia then starts going after Norway with their GDR's and sends them packing.

The final battle then ensues (first pic to second pic). Russia takes back everything on their landmass and the nukes start flying. Canada gets involved and Russia walks their GDR's over there and take a couple cities from Canada but eventually lose them to loyalty. Russia peaces out with Brazil, stays at war with Canada and goes back across the pond and finishes them off (second pic).

Conclusion: This was by far the most entertainment iv'e seen in a game. Thinking one civ (India) has a guaranteed victory to getting wiped off the planet was quite enjoyable. And to see efficient use of naval, air & nukes was amazing. Unfortunately, in the end the only units left are about 100 GDR's, Jet Bombers and missle cruisers but I guess that's to be expected.

I going to try a continents game next on Deity and ACTUALLY play myself to see how I do.

So bad AI unable of taking cities is seemingly largely result of AI being programmed to play as intended in a weird set of rules.

At this point I don't even know if it's better or worse than our previous explanations of AI quality
 
Presumably an easy fix to the AI so it could actually handle itself in war is just increasing their amenities then? Like, +1 amenity in every city?
Probably no "easy fix". Something static could be implemented easily along Elhoim's UPDATE GlobalParameters SET Value = 20 WHERE Name ='CITY_AMENITIES_FOR_FREE'; But I'm sure, regardless whatever number is chosen, the cities will compensate the extra aminities with natural growth and gain (nearly) nothing (as we know, really big cities / playing tall is not very rewarding).

With trying to discover for a Civ 'in case of necessity' extra resources / amenities I mean a context with specific reasons, for example an emergency event, in which a participating Civ should appear more lively & competent. In the worst case you developer might get the babysitter blues ... maybe you have to give appropriate units, say musketmen, then you'll add also some niter (for repair & immersion of the unbelievers), also at least one source of niter inclusive improvement within their territory, some gold for immediate maintenance of the units and in order to attract their interest some of the new extra amenities along a source of that inclusive appropriate improvement within their territory ... *** [and you give the tech, Military Engineering]
So bad AI unable of taking cities is seemingly largely result of AI being programmed to play as intended in a weird set of rules.
Why 'a weird set of rules'? Do we want AIplayers, who can conquer cities and then are totally bogged down, because they have a dozen amenities less than needed for equilibrium?
To me it looks like the set of rules is ok in principle - changing some parameters you get a wide range of outcomes:
The current state of AI city-taking is atrocious. Look at the replay, and only four cities ever were captured? (And they were all city-states...)
to see efficient use of naval, air & nukes was amazing. Unfortunately, in the end the only units left are about 100 GDR's, Jet Bombers and missle cruisers but I guess that's to be expected.

.

I going to try a continents game next on Deity and ACTUALLY play myself to see how I do.
I would be most interested in a triplet of onlyAI games:
- amenities & the other military stuff
- only amenities (exactly same values as above)
- only the other military stuff (exactly same values as in first)
played on the exactly same map etc. - same turn 0 save file, but 3 different sets of XMLs.

[*** Edit: of course I forgot something :D:p]
 
Last edited:
So I came back from work today to see what happened and I was still on the mail loading screen (first stirrings of life), I forgot to click start game before I left, but that ended up being a good thing for my second test as I got to watch this in full!! So test #2 had the following result before the map got filled with GDR's.
<...>

This is amazing read and the end result minimap reminds me of how Civ3 games usually ended: relentless brutal elimination warfare till the end. Not sure if I truly want it brought back in that form, but present AI performance is really unsatisfactory. This could be basis for more options re AI aggressiveness in the game settings.

Is there some code which defines how much the AI value amenities? If yes, maybe similar results could be achieved by instructing the AI value them considerably less than now?
 
Thanks, @Engeez for these experiments.

I'm quite encouraged by these results, as it suggests that the current AI behaviour is not a result of a fundamental inability of the Civ 6 AI to play the role of interesting foils. I'm sure @cvb is correct and that this still isn't an easy fix to the AI's passivity, as changing one factor has knock on implications that need to be dealt with. But at least it seems that it may be possible to encourage the AI to be a more aggressive opponent in the end game.
 
I would be most interested in a triplet of onlyAI games:
- amenities & the other military stuff
- only amenities (exactly same values as above)
- only the other military stuff (exactly same values as in first)
played on the exactly same map etc. - same turn 0 save file, but 3 different sets of XMLs.

I would be glad to do these triplet tests - if someone gives me the mods to play with...
 
I would be glad to do these triplet tests - if someone gives me the mods to play with...
Great! Having more test samples will probably help to see even clearer.

The easiest way would be to make save copies of the original ..\SteamLibrary\steamapps\common\Sid Meier's Civilization VI\Base\Assets\Gameplay\Data\Units.xml | Buildings.xml | Technologies.xml | Districts.xml files and then overwrite them temporarily with versions modified by @Engeez
Same turn 0 save file &
a) all original files
b) just modified Districts.xml
c) just modified Units.xml & Buildings.xml & Technologies.xml
d) all modified (Districts.xml & Units.xml & Buildings.xml & Technologies.xml)
Quartet. The vanilla version as reference is interesting too!
 
I tried an actual human game with this and found out that since I added a bombard value to the ranged units they also attack other units like seige units do, as in never. So with the main end goal of this to make it fun for the Deity level player, I will keep testing this AI vs AI first then AI vs. human. I read through alot of the xml files for the original and two dlc`s and came up with the following ideas:

1. Make the AI Leaders Warmongers.
a) Re value many of the grievances
b) Re value many of the diplomacy modifiers
c) Re value if one of your spies get caught on a mission, better prepare for war!
d) Re value parameters for AI to go to war

2. Tactically:
a) Give seige units extra movement, more base strength, more bombard strength
b) Make them build more seige units (I find they build less and less as the game progresses, perhaps make bombards and artillery non resource dependant)
c) Adjust their 'walled city attack" teams to include more seige units and bombers if available.
d) Don't send a naval support unless they have enough to take a coastal city.

3. Walls and buffs:
a) Trying to think of a proper wall system. 25-25-25-25 (25% strength altogether), seems fine AI vs AI but human vs. AI is another story. Almost considering making Unique melee units with battering ram abilities for the AI. Ideally it would be great to strengthen walls of AI and weaken Human walls.
b) Perhaps make units cheaper to purchase and build then buff then penalize human players production and gold because human knows how to create production (chopping). I don't like making everything cheaper like wonders or you can never beat an AI to it.
c) Find a way to give AI unlimited amenities while still valuing trades and make humans work for it.
d) Buff AI housing slightly so they can grow a little faster, perhaps even food.
e) Change key districts (Campuses & Theatres) production value so game is slightly slower and you makes the player not want to spam (this will definitely require some testing)
e) Change population values for production, science and culture (more like civ 5). Make tall cities powerhouses.
f) Make having power worth it. Major buffs to buildings with power. Especially IZ buildings
g) extend factories, zoos ect. regional to 8 so you can actually have cities futher apart.

In general as a Deity player the biggest immersion mechanics to me are being scared to death from start to finish of being destroyed while I fight and claw my way through the game. And also seeing AI's fight consistently. I also want the AI's to try and stay competitive throughout the game even if they fail to get resources and completely abandon science.

Moderator Action: Please do not use symbols in words to evade the autocensor. Changed your text to reflect the forum rules. leif
Please read the forum rules: http://forums.civfanatics.com/showthread.php?t=422889
 
Last edited by a moderator:
Engeez, you have terrible many really fantastical ideas. A lot to come back later and look at it as reference "what else could be done" ...

Alone your point 3. Walls and buffs c) Find a way to give AI unlimited(?!?) amenities while still valuing trades and make humans work for it. will be a whole Lua file.

But for now - we have the saying "take two steps at once and fall on your jaws" - to keep the thing quite manageable, I would prefer to take a SMALL set of parameters and test with them without changing them and see what happens by several users.
I want to ask you to provide the modifications you described some posts above. I think, right now the details are not so important and V. Soma et al could do some tests, to demonstrate the potential of the ideas.

How much is the lack of amenities the critical path on the way to a conquering AI?

Btw, I'm a fan of Antoine de Saint-Exupéry and his musings about the perfect machine ...

.

PS. I find the line "1. Make the AI Leaders Warmongers" not so clear, the word should express, that it is a person, who is easily to disdain and hate with deep emotion, lowlife & wicked.

Moderator Action: I chose the word "Warmongers" because the original word he chose was in violation of the forum rules. edit - PDMA removed. leif
 
Last edited by a moderator:
Uh, I try to run the AutoPlay mod for testing but it does not seem to work for me - do I have to alter it for GS?
 
I tried an actual human game with this and found out that since I added a bombard value to the ranged units they also attack other units like seige units do, as in never.

This is an odd omission to the basic combat routine. I can understand Siege units targeting cities or encampments as their priority target, but it seems odd that their second choice wouldn't be to attack enemy units when possible, ideally after Ranged units attack and before Melee units do. I wonder how challenging that will be to amend?


Find a way to give AI unlimited amenities while still valuing trades and make humans work for it.

I wonder if there's a way to give amenities based on Population? Sort of like the old Civ 5 Tradition policy, but applying to all AI cities? If the Deity AI got one amenity for every 2 Pop, then combined with their luxuries and districts that might provide a buffer that would give them the confidence to attack cities (assuming this is a real relationship, which obviously is the objective of further tests). As a bonus knock on effect, it would also make it much tougher to Loyalty-flip AI cities, and their yields would be boosted from their happiness, both of which would contribute to toughening up the Deity opposition.
 
This is an odd omission to the basic combat routine. I can understand Siege units targeting cities or encampments as their priority target, but it seems odd that their second choice wouldn't be to attack enemy units when possible, ideally after Ranged units attack and before Melee units do. I wonder how challenging that will be to amend?




I wonder if there's a way to give amenities based on Population? Sort of like the old Civ 5 Tradition policy, but applying to all AI cities? If the Deity AI got one amenity for every 2 Pop, then combined with their luxuries and districts that might provide a buffer that would give them the confidence to attack cities (assuming this is a real relationship, which obviously is the objective of further tests). As a bonus knock on effect, it would also make it much tougher to Loyalty-flip AI cities, and their yields would be boosted from their happiness, both of which would contribute to toughening up the Deity opposition.

I’ll slow down and next time I get a chance to play I’ll do autos with same parameters (one shot ranged) and unlimited amenities, same civs, same seed. Then set amenities to default and post results. I do notice the AI sometimes surrounds the city with ranged and one has to die before a melee unit can capture the city.
 
I was running a game as France last night. It really wasn't turning out very well, not great placement, hemmed in by mountains. I was only able to get three cities going. I did at least have lots of iron on hand. Was only up to the Medieval era and had swordsmen, pikemen and crossbowmen. Hungary had already taken out Egypt earlier and I was next on the list. They brought in 4 catapaults, about 6 swordsmen, 4 crossbowmen (we were militarily equivalent) and attacked one of my cities. I wasn't really all that happy with the game anyway so I wanted to see what the AI could do if given a chance so I did try to defend, just not too hard. I would fire back at them with my crossbows and city bombardment, had my infantry defending and had horsemen (just got knights and was building them but my funds were low so couldn't upgrade) harrassing, especially the catapaults. The AI used the catapaults and crossbows fairly efficiently, with the catapaults set up on the other side of a lake and took their infantry around the flanks and engaged my pikes and swordsmen supported by crossbows. You could see their strategy and it wasn't bad but their swordsmen would attack a couple of times and pull back or would be wiped out by a counterattack. Incoming reinforcements would do strange things like move back and forth between two hexes for a couple turns then come in. I deliberately left openings to see if they'd bite and it took awhile for the AI to realize it had an opening and finally took it. Overall it took them about 15-20 turns (I kind of lost count after awhile) to finally take the city even though the catapaults were constantly bombarding and kept the city reduced nicely. Of course that fight involved most of my army so I was pretty open after that. They took my capital next, but it still took about 10 turns to do so because they were very slow at moving troops in. And I still gave a fight to them but they did have numbers on their side. And then they took my final city in about 5 turns because I really had little left.

Overall, it does seem that the AI doesn't take players out that much because it's hesitant and it's attempts at co-ordination are there but clumsy. It also does some odd things with troop movement that slow down it's advance. With the same resources I would have taken the first city in 5 turns or less. And had I tried a bit harder it really wouldn't have taken much to destroy the first assault. But if the AI had pushed hard at the beginning it really would have taken my city easily even if I had tried harder. Still it was interesting to see how the AI operates on city attacks because I haven't seen it that often against me directly. There is some promise there if it would only push a little harder even with the clumsiness. So I am happy I lost because I've not been conquered once in Civ VI and it was refreshing to actually lose like this.

Please note that I'm not a high level Civ gamer. I've played Civ casually since Civ I but I play casually because I enjoy the game and it's many facets. This game was at King on a standard continents map, really a very ordinary setup except for the horrid start location. Hungary really could only come at me one way due to the mountains and I couldn't really expand past my three cities without going to war with Hungary myself which I was prepping for but got surprise DOW'd by Hungary...I actually was surprised because we were being best buds up to that point. Yes I was going to betray him but I didn't expect him to do it first :P
 
Overall it took them about 15-20 turns (I kind of lost count after awhile) to finally take the city even though the catapaults were constantly bombarding and kept the city reduced nicely. Of course that fight involved most of my army so I was pretty open after that. They took my capital next, but it still took about 10 turns to do so because they were very slow at moving troops in. And I still gave a fight to them but they did have numbers on their side. And then they took my final city in about 5 turns because I really had little left.

Overall, it does seem that the AI doesn't take players out that much because it's hesitant and it's attempts at co-ordination are there but clumsy. It also does some odd things with troop movement that slow down it's advance.
One thing that Firaxis seems either completely incapable of or uninterested in learning the AI is the value of holding back its melee units. Usually they'll bring in some melee units and some ranged/siege units against your city, which is fine. Any human player will know to fortify the military units around the city while the ranged/siege units take down the walls and depletes health, but AI will mindlessly charge in with its melee units from turn one, outright killing them or bringing them low on health and making them easy to cherry-pick with city garrison attacks and archers.
 
@kaspergm Excellent observation. And most funny thing is that actually the behavior tree for city siege has two separate phases once the units get to the city. First one uses node Siege City and the second Alpha City + Attack City. I would expect that Siege city would be doing what you described and once the walls are down, the second phase would capture a city. In reality however, the first phase is executed only for 1 turn (there is no loop, or condition here) and the next one triggers and thus units attack the city as instructed.
 
@Infixo that is interesting. And yes @kaspergm that is exactly what they would do with melee against my defending units fight until they could be easily taken down by the city bombardment or my ranged units. I could have repulsed them but again I played half-heartedly deliberately to see what they would do. The siege units were used well and kept keeping my city nicely reduced but lack of follow through on the melee front made it take forever. The human player usually smashes these attacks because of this suicidal behavior and with fewer units than the AI. And they aren't usually as well coordinated (for the AI anyway) as this one was, they usually go in with fewer units or more piecemeal and get smashed to pieces. Had they played their melee and crossbows better, they would have likely beaten me actually trying and not with the half-hearted defense I was throwing up. When there is little opposition though they take so long bringing units in to finish the job. It took forever for them to move their units on my capital after the first city and as I was busy trying to build more I got units into play that shouldn't have gotten finished had they played it properly considering how little opposition I had left to defend my capital. The couple survivors from the first city fight had time to fully heal before they showed up at the capital, in a more piecemeal fashion this time dashing their melee against my hill fortified melee which didn't go so well for them until their crossbows finally got there and their catapults got into position.

Overall they just can't move their units well. They can't follow through unless you let them and so they lose most of the time unless they surprise you because they don't come in with that overwhelming unit advantage like they should unless the terrain is in their favour and even then it leaves room for improvement. Or they don't use the units at hand very well. Really the first city attack was fairly impressive for the AI but the other two not so much.

This is why the stacks of doom from the older Civ's were so effective because they would overwhelm you with these ridiculous hordes of units. They didn't use any real strategy other than overwhelming numbers which made things hard for the player (at least from what I remember from the older Civ's anyway). I don't know if the AI knows how to move and coordinate the units properly most of the time. They can do it sometimes but not most of the time. If they did that it would up the difficulty of AI attacks on us players.
 
This is why the stacks of doom from the older Civ's were so effective because they would overwhelm you with these ridiculous hordes of units. They didn't use any real strategy other than overwhelming numbers which made things hard for the player (at least from what I remember from the older Civ's anyway). I don't know if the AI knows how to move and coordinate the units properly most of the time.

It has to be easier to program the AI if it has numbers on it's side. Melee attacks of attrition would then up the combat difficulty on Immortal+ without impacting enjoyability for people who prefer the mid-level difficulties.

I'm modestly disappointed that the dev team resolved the 1:2:1 Range issue by moving Machine Guns to range 2 rather than moving all Ranged units to the Slinger:Machine Gun range of 1 combined with an automatic free attack against the first unit to attack the Ranged unit each turn. The current rules for Ranged units are heavily tilted in favour of the player, and there's no need for it, flavour-wise or fun-wise.
 
Back
Top Bottom