Some positive feedback - I like the doggedness. After buying Legend I can't go back to civ because civ feels like playing solitaire - shove luxuries down the AI for GPT, fill up the blue and purple meters, and play a 400 turn game with only one AI triggered invasion that does nothing.



The Legends AI may be ineffective at times (militia obsession, city placement, tech decisions) but they go for the throat. I appreciate that amplitude made conflict a key part of the game.

Andy06r wrote:
Some positive feedback - I like the doggedness. After buying Legend I can't go back to civ because civ feels like playing solitaire - shove luxuries down the AI for GPT, fill up the blue and purple meters, and play a 400 turn game with only one AI triggered invasion that does nothing.



The Legends AI may be ineffective at times (militia obsession, city placement, tech decisions) but they go for the throat. I appreciate that amplitude made conflict a key part of the game.

I would say that the same issues you have with Civ are here. You can shove luxuries at AI for dust/strategics/research, pay them off to keep them happy and play a peaceful game. In my opinion, it is an easier AI for a warmonger. As a warmonger by trade, I have more issues with fighting on multiple fronts in Civ then I do in EL.



Being forced to fight on multiple fronts in EL changes the game because of the unit/army setup. It takes a while before you have two strong unit stacks, especially with the large unit increases due to heros, experience levels, and strategics.





I agree that certain AI need to definitely be more "dogged" in pursuing victory types. But I think they need to be more opportunistic as well. I guess im having a hard time with understanding the opportunistic nature of the AI. For example, I would consider attacking a larger/stronger empires un-protected flank as "opportunistic"... indeed it is the fastest way to improve when playing on Endless AI, take Cities that are way upgraded in comparison to your own.



This isn't something I saw from the AI in my playing, as soon as I became "strongest" player no AI would declare war on me. Maybe these opportunistic behaviors just need to be expanded.

A mechanic which causes the AI to gang up on anyone the moment they start steam-rolling would be a complete game changer in terms of the user experience and late game challenge. Its one of the things on the AI requests list which I'm most hoping to get through. That and the AI playing more in line with their races bonuses.



The trouble is how to get this to work with the diplomacy game. I liked realm divide in shogun 2 (everyone turning on you when you got close to winning) but i didn’t like how it basically made the diplomacy game irrelevant. There is definitely a balance somewhere but its a subtle one.

wilbefast wrote:
Trade-oriented AI is something we've discussed - currently the AI's "focus" is elsewhere ("direct" economy, military, etc). It's good to have the community take note of this too, so we know whether it's worth prioritising

But something like a harbor, especially in a city that contains multiple sea tiles is not simply a

"trade-oriented" building or improvement... It should not require any shift of focus.



The Harbor will:

[LIST=1]

Add +5 industry to each lake/sea tile the city exploits (Production).

Create a disembarkment point at the harbor which greatly eases moving units from land to sea and will give troops greater reach (Military).

Open the entire landmass up to trade with overseas factions it is peaceful with and yield vast increases in both dust and science trade income (Economy) & (Science).

[/LIST]



I can't imagine that AI's focus is "singular" by design? I am not sure how the game is coded but should not a single improvement (Harbor) that enhances multiple areas of focus trump whatever other single-focus objective it has?



By the way Wilbefast, I really appreciate your responses and solicitation of our feedback. I will always love Amplitude for including the players in the process. It's very refreshing!

Zenphys wrote:
It's great to get such dev interaction.



What is the game plan moving forward for the AI? Is the focus to primarily fix the issues listed in this thread (It would be fantastic if the whole of that list got delt with). But are there any other plans for the AI? Perhaps making them a little more dogged and focused or maybe introducing more long term thinking. Not being a programmer I’m not sure how easy/difficult these changes are and I imagine there is only so much time and the majority of the architecture will already be in place.

There are some things which can be fixed by tweaking the weights the AI uses to choose an option and other things which would require a restructuring of how it works. The cost of changing something is not always what you'd think! The goal is to find the right compromise between what the community wants (which defines the priority) and what things cost in programmer man-hours.



For example: a better handling faction-specific personalities is something we've been asked for on a number of occasion, and it is a shame to invest so much on fleshing out the world only to have the AI contradict the lore. This is something that we can solve mostly (a very dangerous word) through data, so it's fairly "cheap" (another very dangerous word) even if it's very tricky to get right.



I can't tell you 100% of what we're doing at the moment, but rest assured that the AI team is doing many cool things as we speak

Andy06r wrote:
I appreciate that amplitude made conflict a key part of the game.

99% of the time the AI is playing to win (or at least to survive), but when you see allied AIs turn on you it's sometimes because we do try to foster conflict for player enjoyment in the AI So the AI is also trying (in very rare circumstances) to make the game more interesting. You can't have your cake and eat it too unfortunately.



[HR][/HR]



A specific question for the community while I'm here: have you noticed any player actions which you feel are missing feedback? That is the AI "should" have said something in reaction to an action I just performed but didn't. Bearing in mind of course that there's a kind of "anti-spam" measure applied to stop players from receiving too many messages per turn, so "minor slights" might not always be noted if there are more important messages to display on the same turn.



Thanks again for your feedback



W.

wilbefast wrote:
A specific question for the community while I'm here: have you noticed any player actions which you feel are missing feedback? That is the AI "should" have said something in reaction to an action I just performed but didn't. Bearing in mind of course that there's a kind of "anti-spam" measure applied to stop players from receiving too many messages per turn, so "minor slights" might not always be noted if there are more important messages to display on the same turn.

I've got the opposite case. One very nasty thing the AI just doesn't seem to discern or care about: When I'm playing the cultists and they start razing my converted villages. The AI does not seem to recognize that as a hostile act of any sort.

I've just used an exploit regarding the AI that made me destroy easily two delvers units with my starter vaulter's army , at turn 1. The AI targets *always* the weakest units in battle. So I had the starter army, with the hero and a settler, and two marines nearby.



During deployment phase, I've placed the settler very far from the enemies... They wouldn't even reach the settlers at the end of the battle... And I used the marines and the hero to make the delvers "run the gauntlet"...



So, in battle evaluations, the units targets should take into consideration the battle length, and the ability to actually reach the target unit in time to cause damage, and eventually change the targets if it perceives that is not possible anymore.



By the way, I can't repeat my deed to demonstrate, because AUTOSAVE doesn't save THE INITIAL TURN.... Could you change that little itsy bitsy detail? This is the aftermath...

wilbefast wrote:




A specific question for the community while I'm here: have you noticed any player actions which you feel are missing feedback? That is the AI "should" have said something in reaction to an action I just performed but didn't. Bearing in mind of course that there's a kind of "anti-spam" measure applied to stop players from receiving too many messages per turn, so "minor slights" might not always be noted if there are more important messages to display on the same turn.



Thanks again for your feedback



W.

AH, yes.

I've noticed some things, all of them related to war, especially war on multiple fronts.



Situation 1:

This one is a rare occurrence, but I ran into it and it annoyed me.

If I am at war with empire A, and empire B is between me and empire A, empire B should not freak out and break peace with me when my army marches through their land, especially if empire B is also at war with empire A.



Situation 2:

If empire A is at war with empire B and is having a successful campaign, they should not continue to capture empire B's near defenseless cities when my army comes in and starts capturing Empire A's cities on the other side of the battlefront.



Situation 3: If I roll my entire army out to go to war with Empire A and Empire B has vision on my territory, Empire B should take note of this and recognize that my cities are inadequately defended. If my army is far away and busy with empire A, and relations between me and Empire B aren't all that good, and empire B is NOT currently involved in an active war, they should try to conquer my land.



Situation 4:

If I roll my entire army into someone's territory, and the layout of the land is such that certain cities are more safe than others, armies should be moved out of garrisons and sent towards cities close to where my army is to bolster defense. This should be done, ideally, before the attack actually begins.



Situation 5: If an empire has vision on my units and sees that they are headed towards their territory, assuming they have no need to defend multiple fronts, they should respond as recommended in situation 4, regardless of my relationship with them.



General Rule regarding 4 and 5:

Empires need to anticipate where attacks will come from and understand that its ok not to try and evenly defend every city during war. Go for where the action is, stop the conquest in its tracks and then retaliate if they are successful and the enemy army is decimated.

wilbefast wrote:


A specific question for the community while I'm here: have you noticed any player actions which you feel are missing feedback? That is the AI "should" have said something in reaction to an action I just performed but didn't. Bearing in mind of course that there's a kind of "anti-spam" measure applied to stop players from receiving too many messages per turn, so "minor slights" might not always be noted if there are more important messages to display on the same turn.

Some kind of communication would be appropriate if the player settles a province the AI intended to settle within the next turn (settler already in the province, probably). A situation like that would generally prompt some words in a multiplayer game

natev wrote:
Some kind of communication would be appropriate if the player settles a province the AI intended to settle within the next turn (settler already in the province, probably). A situation like that would generally prompt some words in a multiplayer game

As complicated as that could be, adding human elements to the ai like that would be very interesting... heck, it would revolutionize diplomacy in a manner no game has ever really attempted before.

Negotiating future borders, opportunities for backstabbing and misdirection... all the stuff that goes on behind the scenes in a multiplayer game brought to a single player game... amazing

I don't know if it's been mentioned before, but sometimes the AI seems to value Dust way too much compared to other resources when it comes to diplomatic deals. I attached a few screenshots of Hard difficulty when dealing with the Vaulter AI and a save file just in case.



First picture is the initial deal where I'm trying to get them to attack another empire.

http://forums.amplitude-studios.com/attachment.php?attachmentid=17647&stc=1&d=1437050240

Second picture is the same deal with all my non-dust resources on the table instead of Dust.

http://forums.amplitude-studios.com/attachment.php?attachmentid=17648&stc=1&d=1437050245

Third picture is another deal where I'm basically stealing the AI out of all their precious resources.

http://forums.amplitude-studios.com/attachment.php?attachmentid=17649&stc=1&d=1437050250



Even if the AI didn't need any of the strategic or luxury resources they'd be getting much more Dust just by selling it all to the market or trading it to others. The other AIs seem to be much less willing to trade away their resources now, but in early game too it was way too easy to trade their luxury resources in exchange for measly amounts of dust. I would have actually snatched the Wealth Harvester deed from the Wild Walkers if I would have just had a little more Influence to seal the deal back then.
Roving Clans - Turn 128.zip

The "AI" in EL is a fairly simple script. The game is complex and it probably took quite some effort just to get the AI to a point where it wasn't horribly broken. There are few crashes to desktop, or units walking in circles, or units standing still for dozens of turns, although I have seen all of this happen. If the plan is to keep the existing framework and add subroutines to pad out the complexity, I think this may not be agressive enough to make the game successful. As it stands now, the game is very repetative in single player because the AI follows the same pattern in every circumstance. Each game may have the illusion of variety because the maps, local geography, faction arrangement, random quests, etc are different but the reality is that once you get a leg up on the AI by overcoming the beginning boosts (based on difficulty) setting, the game is pretty much over and the only hard decision is which victory condition you want to pursue.



Before I get into the big picture stuff, I want to illustrate my point about the simplicity of the AI by using combat as an example. Based on my 2 dozen or so games, this is my take on the combat AI (some of this is guesswork):



1) Each unit identifies the enemy unit it has the greatest damage potential against and targets it. It keeps this target until the target is dead. [edit] According to this thread, unit types may also look at other factors like whether the enemy units are injured or isolated.

1a) If the unit is a healer and there is a friendly unit at less than 50% health, the target is the damaged unit; if multiple units are damaged, it chooses the most damaged or closest. If the unit is a buffer, it will buff a friendly unit the first round, choosing the strongest unit based on the type of buff.

1b) This includes all modifiers based on unit type, so an army of mixed type and weapons may have different targets, but often it's the same target. I don't believe morale and georgraphy modifiers are applied here.

2) Each unit, on its turn, will move toward its target as efficiently as possible until it is in attack range.

2a) It will end on a geographically advantageous tile if possible, prioritizing forest, high ground, and then morale.

2b) If the unit's target is destroyed, it will select a new target first.

3) If the unit can attack its target, it will. If it can't, it will attack another enemy unit in range. If it can't do that, it will heal or buff a friendly unit. If it can't do that, then it will end its turn.



I think that's it, a few simple rules with some caveats. You can tell what the script is because each round before you give your own instructions, the AI has defaulted to the above script. As you can see there are numerous decisions that the AI would never make because it's not part of the script. Many of them have been mentioned in this thread, but let me try to compile a more comrehensive list:



* The AI ignores anything outside (georgraphically and temporaly) of the specfic combat instance.

** A human player might take into account when this army will fight next and may alter strategy. For example, if an army is going to fight again the next turn, maybe the player prioritizes healing and makes units retreat earlier than if there is no impending combat.

** A human player may play defensively or kite if he wants to re-engage in a later turn to get a geographic advantage or stall until reinforecements arrive.



* The AI will always let all combatants into the fight.

** A human player might keep militia, settler units, or armies he wants to fight in a separate battle out of combat.

** A human might rearrange armies on the fly to put complimentary units together.

** A human might subdivide an army to attack a settler multiple times to prevent it from running away.

** A human player might protect units that are necessary to fulfill a quest.



* The AI always accepts the default unit placement at the beginning.

** A human player should almost always tailor the unit placement to get tanks up front, cavalry and flying units flanking, ranged in the rear. A human player might try to protect weaker support units or heroes or position AOE healers to maximize army healing.



* The AI ignores relative army strengths (evaluated by any metric). It fights every battle as if it's going to win easily.

** A human player may, if outnumbered, play defensively, prioritize taking out the units with lowest hit points, or stall to allow his hero to survive. He may also do these things if things are evenly matched under some circumstances.

** A human player may stall a battle which he is likely to win to allow his healers to maximize health upon exiting the battle.

** A human player might prioritze high profile units (heroes, support units).



* The AI only uses the aggressive stance, except perhaps for settlers.

** A human player might use non-aggressive stances if he wants to hold a strategic position or chokehold, protect his ranged units, allow for injured units to escape, claim forest or high ground, or position his units to avoid or encourage special attack types (like chain lightning or sweep back strike).

** A human player may attempt to position his units in such a way to block the arrival of enemy reinforcements.

** A human player might take advantage of this by using hold poition stance and allowing calvalry or other fast units to approach first and thus split the battle up into more manageable sections.

** A human player using the Necrophages may attempt to cause disease in most or all enemy units as early as possible.



* If a unit can't reach its target, the AI won't consider secondary targets. This results in units sometimes moving backwards or sideways to positions where it is taken out of the battle.

** A human may select reachable targets each round. Sometimes this may not be easy in the aggressive stance because if the initial target is destroyed, the unit reverts to the AI script. A human may choose to operate in 'hold poision' mode.

** [edit] Faction specific, a human may switch a Proliferaor's targets each round and prioritze targets likely to die soon to maximize the number of battle born.



* The AI does not give deference to lower initiative units. The units basically trip over each other trying to get to the same units leaving some units unable to attack.

** A human may give instructions that space units out to get maximum and advantageous adjacencies.



* The AI doesn't finish off units.

** A human player might kill off a unit to prevent it from taking a turn.



* The AI doesn't pay attention to battle length. It will send units to attack units they can't reach before the end of combat while ignoring units that are reachable.

Sze, you may wanna make points outside of combat. Most of those things have already been mentioned and are being addressed by the devs.

Part 2

To continue from my earlier post about combat, all of the game's AI systems seem to be based on a fairly simple script. No one is expecting true AI, but the way it is now it is too predictable. The AI tends to behave the same way, some examples: the starting hero explores and attacks minor factions and explores ruins, some unit are sent to map the area, after x turns the first settler is built, if the AI gets too far ahead or behind a player and he has an army that can take one of that player's cities, he declares war and then sends the army.



Let's take another example, expansion. A human identifies an area to settle and builds a settler. The AI does the reverse, it has a script that says "these conditions have been met, time to make a settler". The conditions being chiefly time since last settler and probably empire happiness. Once the settler is built, it looks for the strongest FIDSI and may also look at adjacency, nearby factions, and maybe some other factors, I'm not sure. A human looks at a lot more: starting FIDSI, built out FIDSI, ablity to fit the city within the geography, anomalies, rivers, coast, height/defendability, minor factions (count and type), resources (count and type), trade routes, and proximity to other factions.



---



Here are the three things I think need the most attention:



1) Factions using their abilities. This has been brought up in this thread before, but Roving Clans don't setseke or use privateers, Drakken don't force peace even at the brink of defeat, Vaulters don't teleport, Broken Lords tend to not exapand, etc. It's inconsistent though because some faction abilities are used by the AI like market bans (liberally) and AM spells.



2) Military strength being purely a function of active units. The AI doesn't consider industry, tech level, or dust and resource stockpiles. I may seem weak now, but if you declare war on me and I can double my military in one round by buying units and mercs and retrofitting, then maybe war isn't the best idea. The converse is also true, if I declare war on the AI, I expect it to retrofit, consolidate units, but marcs, and rush build military units. I don't think it does any of that except maybe rush build units (when it's already under seige).



3) AI progress tracking. The AI needs to be more goal oriented instead of following a predetermined path. There are plenty of examples of the AI lagging because it hasn't done some basic things. It tends to research military techs first and then science. It seems to lag behind in dust, food, resource, and utility techs. I can't tell if it grabs happiness techs because it probably gets a big happiness boost anyway. You can tell how much each tech is valued when you try to trade with the AI. It uses these same values to select its next research. For example, trading with Cultists, they value Borough government (military) about 8x more than other era III techs like reaping station or cargo docks.



This leads to some of the following: Broken Lords stalling due to not enough dust, no AI cities building empire mint (not researched), factions having trouble up front with minor factions because they can't parley (not researched), Drakken not being able to make alliances (not researched). Maybe the problem with the Roving Clans not using privateers comes down to that tech not being on the preferred research list.



I suggest a new approach where each faction tries to keep itself from lagging in any area (those on the score page + resources + trade routes) AND tries to keep at the front of the pack in at least one area. It can do this by polling all known enemy players. This is pretty much what human players do, "I'm last in science, I need to work on that". So instead of following a script, they'll research, build buildings, settle new regions, allocate workers, and select governors in such a way that they don't lag. It's complicated and the algorithm needs to take into account that factions are predisposed to have more of certain things. If a Drakken is not way ahead in influence, that should be a red flag.

Hi all!



We shut shop for a week for Bastille Day but now we're back in action - time to catch up

Nosferatiel wrote:
I've got the opposite case. One very nasty thing the AI just doesn't seem to discern or care about: When I'm playing the cultists and they start razing my converted villages. The AI does not seem to recognize that as a hostile act of any sort.

Susucr brought this up here so it's on the wishlist

abmpicoli wrote:
I've just used an exploit regarding the AI that made me destroy easily two delvers units with my starter vaulter's army , at turn 1. The AI targets *always* the weakest units in battle.

Likewise Nos brought up this issue I'm doing some book-keeping today, hence the Brainy Smurf act.

HappyHead wrote:
AH, yes.

I've noticed some things, all of them related to war, especially war on multiple fronts.



Situation 1:

This one is a rare occurrence, but I ran into it and it annoyed me.

If I am at war with empire A, and empire B is between me and empire A, empire B should not freak out and break peace with me when my army marches through their land, especially if empire B is also at war with empire A.



Situation 2:

If empire A is at war with empire B and is having a successful campaign, they should not continue to capture empire B's near defenseless cities when my army comes in and starts capturing Empire A's cities on the other side of the battlefront.



Situation 3: If I roll my entire army out to go to war with Empire A and Empire B has vision on my territory, Empire B should take note of this and recognize that my cities are inadequately defended. If my army is far away and busy with empire A, and relations between me and Empire B aren't all that good, and empire B is NOT currently involved in an active war, they should try to conquer my land.



Situation 4:

If I roll my entire army into someone's territory, and the layout of the land is such that certain cities are more safe than others, armies should be moved out of garrisons and sent towards cities close to where my army is to bolster defense. This should be done, ideally, before the attack actually begins.



Situation 5: If an empire has vision on my units and sees that they are headed towards their territory, assuming they have no need to defend multiple fronts, they should respond as recommended in situation 4, regardless of my relationship with them.



General Rule regarding 4 and 5:

Empires need to anticipate where attacks will come from and understand that its ok not to try and evenly defend every city during war. Go for where the action is, stop the conquest in its tracks and then retaliate if they are successful and the enemy army is decimated.

All good points, I have taken note! The request was actually for times when the AI doesn't send a messages (ie. "we do not appreciate that!") or sends the wrong message - sorry if I was unclear about that - but more general AI issues are what this thread is all about so fire away

natev wrote:
Some kind of communication would be appropriate if the player settles a province the AI intended to settle within the next turn (settler already in the province, probably). A situation like that would generally prompt some words in a multiplayer game

Noted, thanks

HappyHead wrote:
As complicated as that could be, adding human elements to the ai like that would be very interesting... heck, it would revolutionize diplomacy in a manner no game has ever really attempted before.

Negotiating future borders, opportunities for backstabbing and misdirection... all the stuff that goes on behind the scenes in a multiplayer game brought to a single player game... amazing

This is what excites me most about game AI ^^ Unfortunately we can only do so much with the resources we have, and this kind of stuff would basically require "True AI"... "Synthetic Intelligence" rather than "Artificial Intelligence"

hera35 wrote:
I don't know if it's been mentioned before, but sometimes the AI seems to value Dust way too much compared to other resources when it comes to diplomatic deals. I attached a few screenshots of Hard difficulty when dealing with the Vaulter AI and a save file just in case.



First picture is the initial deal where I'm trying to get them to attack another empire.

http://forums.amplitude-studios.com/attachment.php?attachmentid=17647&stc=1&d=1437050240

Second picture is the same deal with all my non-dust resources on the table instead of Dust.

http://forums.amplitude-studios.com/attachment.php?attachmentid=17648&stc=1&d=1437050245

Third picture is another deal where I'm basically stealing the AI out of all their precious resources.

http://forums.amplitude-studios.com/attachment.php?attachmentid=17649&stc=1&d=1437050250



Even if the AI didn't need any of the strategic or luxury resources they'd be getting much more Dust just by selling it all to the market or trading it to others. The other AIs seem to be much less willing to trade away their resources now, but in early game too it was way too easy to trade their luxury resources in exchange for measly amounts of dust. I would have actually snatched the Wealth Harvester deed from the Wild Walkers if I would have just had a little more Influence to seal the deal back then.

Caotico09 mentioned this here. The issue is that the market-value of resources isn't sufficiently taken into account: the AI focuses on what the resources is directly worth to him/her/it (it probably, yes?), not on what the resource could potential be exchanged for. We're on it (the problem, not the AI)!

Sze wrote:
The "AI" in EL is a fairly simple script.

The AI used for tactical battles is simplistic, yes. The focus has always been on the strategic AI, since when this fails there are no armies to command in the first place. Also since the tactical battles are wholly new in EL and since they kept being iterated on design-wise until the game was released there wasn't a lot of time for the AI team to "catch up".

Sze wrote:
Military strength being purely a function of active units. The AI doesn't consider industry, tech level, or dust and resource stockpiles. I may seem weak now, but if you declare war on me and I can double my military in one round by buying units and mercs and retrofitting, then maybe war isn't the best idea. The converse is also true, if I declare war on the AI, I expect it to retrofit, consolidate units, but marcs, and rush build military units. I don't think it does any of that except maybe rush build units (when it's already under seige).

Interesting point - I don't believe the AI currently considers stockpiles and dust reserves when deciding whether to attack someone. Then again we've tried, as much as possible, to build an AI that doesn't cheat (ie. see stockpiles a player has).

Sze wrote:
AI progress tracking. The AI needs to be more goal oriented instead of following a predetermined path. There are plenty of examples of the AI lagging because it hasn't done some basic things. It tends to research military techs first and then science. It seems to lag behind in dust, food, resource, and utility techs. I can't tell if it grabs happiness techs because it probably gets a big happiness boost anyway. You can tell how much each tech is valued when you try to trade with the AI. It uses these same values to select its next research. For example, trading with Cultists, they value Borough government (military) about 8x more than other era III techs like reaping station or cargo docks.

The AI doesn't follow a predetermined path when it comes to research, though given the same conditions (as evaluated by its heuristics) it will make similar choices, and certain choices may be overvalued. If we push the AI to be more balanced and economy-focused it may be that it doesn't provide an interesting challenge early-on and is simply obliterated. Balance is hard to achieve. If there are any technologies or improvements you feel the AI is choosing too often in most/all circumstances please let us know!

Every AI player will consistently trade me all of their strategic and luxury resources in exchange for something like one technology, a peace treaty, or just sharing our maps.



It should identify that if I request a resource, I have demand for it, and it should weight the value more heavily. It should also factor its remaining amount of that resource into the supply/demand formulation. I think what is happening is it says I have 150 titanium, so it values each unit low. It doesn't consider that if I accept this deal it will have 0 titanium.



Right now I can trade every AI the same useless tech, like diplomatic manse, in exchange for enough strat resources to create a monstrous army. I can also look at their empire and specifically give them tech they can't use. They love it. "Cargo Docks for my land locked cities? Sign me up! Have about 300 paladium and all my dust."



Every game I feel like I'm selling the Brooklyn bridge.

AI (red Wild Walkers) is not accepting a truce. They have only one city and no units left. They have no chance to survive another turn but they do not propose or accept a truce.
Wild Walkers - Turn 131_AI Problems.zip

I realize the following suggestion may well be outside the scope of possibility and the scope of this discussion... but this would improve the AI a bit in my opinion



Diplomatic coexistence, something that happens in multiplayer, never in singleplayer... If it were possible for things like this to happen in single player, it would be awesome.

In the end, it would behave like: under X,Y,and Z, conditions, do this!.... oversimplified, but, ya.





In practice, during the start of a game, players spend the first few turns figuring out who their closest neighbor is. If the neighbor is close enough, and no other players are in the immediate area, they may decide to form a mutually beneficial alliance until they inevitably decide to back-stab one another later on because its preferable to wasting energy trying to kill each other early on.



Happening most often with the cultists and roving clans, the clans will settle all the land around the cultist capital. The clans get the land and resources, but the cult gets the minor factions, while both benefit from trade. Both then work together in guarding that territory. The clans will help support the cult in war defensively, and the cult will help support the clans- but only if attacks are being made in the area around their main city.

For a periodic service fee, the clans may offer to surround the cultist city with clan units, meaning no army can really siege the cult without going to war with the clan. This also means the cult may be in immediate danger if it should ever decide to go to war with the clan and it gives the clan the opportunity to stifle city growth if they want to force the cult to declare war on them...



Then, we have the case where cultists may become vassals of the necrophages. Faced with going to war with the necrophages, a cultist player may offer vassalship, sending almost all generated minor faction units into necro territory, steadily and slowly, as a sacrifice.

Necrophages benefit from the fodder feeding their war effort as they go on campaigns elsewhere while the cultists gain time to shore up city defences or perhaps do some shenanigans on the side to wipe off the necros.

The pact lasts, by default, as long as the cultists keep feeding the necros. In the end, its a competition to see whether it will be the greed of the necrophages or the fear of the cultists that will be the undoing of either kingdom. Inevitably, either the necros will wipe out the cultists before they can really build themselves up, or the cultists will keep them pacified up to a point where they can conquer them. Usually, the cult will become buddy buddy with the necro's current target in secret, and share vision so they know if and when the necro army is weak. This usually goes unnoticed as the necros rarely visit the diplomacy screen.





Similar treaties are created in game between other factions, most often one faction will generate protection and an army while the other provides trade. These alliances are formed from the beginning, with the intent of both factions benefitting by uniting to conquer everyone else. Other empires team up with similar plans to protect against the plans of the other team. All of this is entirely informal, leaving room wide open for betrayal and double crossing....



Diplomatic Coexistence can be a state of diplomacy much deeper that just a standard alliance, or the truest expression of cold war as both parties try to silently kill each other. Usually, it becomes both.

And, given the upcoming espionage update, adding stuff like this to the AI may be appropriate to some degree...

Jaynestown wrote:


Right now I can trade every AI the same useless tech, like diplomatic manse, in exchange for enough strat resources to create a monstrous army. I can also look at their empire and specifically give them tech they can't use. They love it. "Cargo Docks for my land locked cities? Sign me up! Have about 300 paladium and all my dust."



Every game I feel like I'm selling the Brooklyn bridge.

Yup, which is why I tend to ignore the AI in my games. That's why I want the diplomatic response of F'off loser, or get bent ya tool, or F' your mom. I want the AI to interpret my response as such, and I wanted it to be insulted as a result and stop bothering me.

Jaynestown wrote:
Every AI player will consistently trade me all of their strategic and luxury resources in exchange for something like one technology, a peace treaty, or just sharing our maps.



It should identify that if I request a resource, I have demand for it, and it should weight the value more heavily. It should also factor its remaining amount of that resource into the supply/demand formulation. I think what is happening is it says I have 150 titanium, so it values each unit low. It doesn't consider that if I accept this deal it will have 0 titanium.



Right now I can trade every AI the same useless tech, like diplomatic manse, in exchange for enough strat resources to create a monstrous army. I can also look at their empire and specifically give them tech they can't use. They love it. "Cargo Docks for my land locked cities? Sign me up! Have about 300 paladium and all my dust."



Every game I feel like I'm selling the Brooklyn bridge.

This issue has been brought up here and here I believe. I'd suggested earlier that we have the AI consider the market-value of resources, since this is a good best-guess of how much they're actually worth. Currently the AIs judge resource values solely based on how much they personally own. What you're bringing up here is an interesting additional flaw in this estimation: if I have a lot of dust then dust is worthless so I can just give it all away. What the AI should do: re-evaluate the value of each dust unit as it is added to the diplomatic agreement. So if I have 2000 dust the first 1000 is worth less than the remaining 1000.

xypocrite wrote:
AI (red Wild Walkers) is not accepting a truce. They have only one city and no units left. They have no chance to survive another turn but they do not propose or accept a truce.

Thanks for the save, I'll check it out!