From what I know of Civ players, everyone likes it when the AI tries to achieve a victory condition but many people hate it when the AI dumps a game-long alliance to attack you just because you’re approaching the winning line.
Also in Civ (and other 4Xs but not in Old World), there are often diplomatic victory conditions that would simply be impossible with the AI always playing to win. Why would they ever vote for another player?
I think it comes down to whether you regard the game as a game that the players are all playing or a simulation of a world that the human player is experiencing. Many players in the latter case want the AI to be trying to do as well as it can in the context of the world, i.e. get to a victory condition as quickly as it can, without playing to be explicitly the first to win.
Possibly the ideal option here (in my opinion) comes from considering how a human who isn’t doing well tends to play a boardgame. Now of course there are many possible responses here that aren’t appropriate, but a standard one might be to help another player who is doing well and has been nice to them during the game. Or at least to work against the player that caused them to lose.
So any AI who is still in contention, or close to it, should “play to win” with a malus to diplomacy, while any AI that is way behind will tend to ally to its closest friend and support their cause. And once no AI nation / alliance feels it has any chance… well that’s a good time to concede the game and not drag things out.
@Solver: I’ve found a nice exploit in my current game. I have a large army that has just cleared a city site but isn’t occupying it. Every turn, my Roman enemies walk a single unit up to take the site, doing a little damage to my adjacent units and then dying to massed archer fire. This has happened for perhaps 3 turns in a row now.
If I instead occupy the city site, they send forward a much larger force (3 to 5 units) to try to kill the occupying unit. Still small enough to take heavy losses though.
Pagan holy cities let you build religious wonders like the Hagia Sophia, but (AFAIK) you cant build pagan cathedrals. can you build a pagan holy site and 6 religious cathedrals for the “6 cathedrals and a holy site” ambition? if so is that cool or weird?
Not sure how tribes behave compared to nations but the (lone) Assyrian archers spent two maybe three turns firing at the Vandals camp while the two melee units just sat there. Needless to say, the Vandals control it no more!
Isn’t the unit in the camp a javelineer? I can never get the unit icons straight. Anyway, I think the camp always has a ranged unit, which will never leave the site. The failure of the other unit to head over and help is disappointing though. The tribes are deliberately coded with a weaker combat AI, but it should be better than that.
I listened to the Soundcloud hosted version of the 3MA episode, which has the correct audio file. Seems like the audio file hosted on Idle Thumbs 3MA page itself, at the time of writing, is incorrect though.
That’s unusual in my experience. Perhaps the Vandals were too busy spray-painting walls with graffiti to care. I usually find that tribes and barbarians are active in their own defence. And if I try to sneak a scout past a barbarian camp at what seems a safe distance I often take unexpected damage as one leaps out at me.
Regarding the combat AI generally, there is obviously room for improvement but it already passes my crucial test: do I feel apprehensive if a nation declares war on me? The answer is usually a resounding ‘yes’ unless I am clearly stronger.
More robust character sheets ala CK3 would be welcome.
Yes, yes, yes!
You can non-obviously make a state religion from pagan faith. whaaa? I didn’t realize that either.
Some of these guys haven’t won a game yet?
Seriously?
Map distribution/seeding can be very unfair.
Agree, although in SP this probably isn’t a big deal.
The resource market - being able to easily buy/sell resources - is genius.
Agree.
Scout harvesting is a solid addition that gives you something to do with scouts in post exploration phase, and can make scouting a little more “decisiony”, as George Bush, the younger, might say.
Agree
Speaking of maps @SorenJohnson - can we get the eqv of CivIV small continents? IRC, you would end up with relatively equal sized mini-continents with players evenly distributed.
It was the best long game MP map, it would be a rush to dominate your continent/island before turning to confront the victors of the other continents.
So far I’ve been through 6 games. There are so many stats and interdependent decisions that it’s overwheling but in a good way(Shadow Empire for example). I enjoy figuring things out as I go and I don’t enjoy min/maxing as much as many other players and it feels realistic that a big decision impacts many everything going on.
Of course this makes it harder to win but I don’t play to win even though I do “try” to win if that makes sense. My first game ended in an out of control empire with all the factions hating me and this was my downfall. I spent too much time keeping the Romans and Babylonians happy while the Merchants and Artisans fumed. That was a loss, but a fun loss for me.
I find the early land grab to be pretty fair. GalCiv is an example of a game that feels like if you don’t get the planets right away you are screwed. I really like how you can sit your scouts on an area and without a war keep that land to yourself at the cost. One time I had a scout sitting on land and I needed food and went out to harvest some food and the next turn the Persians swooped in on it!!
I play it like a roleplayer. If I’m cunning I do sneaky things etc. If my son’s insane I get him a monkey and marry him off to some tribe to keep them placated. I don’t mind making decisions that piss somebody off. It’s cool that if my King makes the Romans hate me my heir’s at least have a chance of turning that around.
The time frame and scope of the game is very good at making every single decision matter and at same time gives flexibility to change your big picture strategy. The wonders system is way better and it seems like a fair trade off to spend all that stone vs build some more.
I’ve noticed more good AI than bad. They don’t execute long wars if I have a strong army. If I don’t defend my outer cities they declare war and try to take it. If they can’t they usually accept a truce. If they are winning the truce can be too expensive. It’s very effective at picking off my units and then backing off while bringing in replacements. Just like a good general would. I had a two city war with Greece and when they clearly overwhelmed my southern front(but still didnt have the city), their units peeled off and headed north. If felt like AI knew the city was going to be captured no matter what I did and took the chance to move more units to a closer battle. I have seem the units go running all around in the background seemingly with no purpose. I’m guessing they are looking around or maybe just using up all the orders??
I don’t know anything about the actual code, but having played a TON of games, I can tell you that the AI knows the basics. It knows which terrain is most important and it makes for those terrain tiles reliably.
It also knows enough to dogpile units so that it scores kills, rather than just wounding.
As to how optimized it is…that, I don’t have an easy answer for, but my guess would be no. It’s not optimized to make best use of its promotions. At least not currently… =D
So you’re saying the AI knows how to use terrain but it doesn’t know how to use traits? I’m not trying to be a dick, but that seems like an odd distinction to make. And to be clear, these are assumptions and not based on any first-hand knowledge?
This is one of the problems with discussing AI. We see behaviors and we make assumptions. Humans love to infer patterns. We see a unit moving to the left instead of straight ahead and we want to think, “Hey, it’s flanking us!”. We see units sitting in a city and we want to think, “Ahh, the AI is preserving its forces!”. We see a unit fortify and we want to think, “Whoa, it’s anticipating my attack!”. And I’m not trying to call anyone out! I do this all the time myself. It’s just how the human mind works.
But in this situation, I was hoping for a confirmation from someone who knows. Which at this point seems to be just @SorenJohnson. Soren, can you confirm whether the AI takes into account terrain and traits? For instance, will the AI move a unit with Highlander onto a hill when it would have otherwise not moved the unit if it didn’t have Highlander? We’ve got one person here saying it does and another person saying it doesn’t.
What does this even mean? One of the differences between Crusader Kings and Old World is relationships in Crusader Kings are a latticework. But relationships in Old World are a hub, with your ruler at the center and everyone radiating outwards from him, unconnected to each other. You generally don’t care what Bob the Coutier thinks of Billy the Prince or Gertrude the Ambassador, and even if you wanted to know, it’s not in there, because it simply doesn’t matter. There’s a lot less to track. Old World focuses on elegance (i.e. gameplay) whereas Crusader Kings is built on a sprawling spreadsheet (i.e. detail). So in Old World, all the information I need about a character is in his or her tooltip. That’s all the characters sheet I need.
That said, I would be interested in a log of everything that’s happened to a character. Maybe that’s what’s meant by “character sheet”. I’d also like a better way to sort and filter my characters besides scrolling down the second tab on the character panel. So maybe that’s what’s meant by “character sheet”. In which case, I agree.
It’s an option, but it’s kind of a bastardized option. Pagan religions don’t have disciples and they therefore don’t get the lucrative monastery/temple/cathedral improvements (these can be massive, especially if you’re pushing culture). Pagan state religions are, however, a great way to leverage your nation’s four shrines. There’s a lovely bit of asymmetry in the shrine distribution across the six nations. It can be hard to see, but it’s in there and it’s pretty clever. I can imagine a polytheistic pagan power has a lot of fun and semi-unique options.
In fact, I’ve just talked myself into a playthrough with Rome as a pagan power!
Normally, I would not be surprised. When you know you’re losing a 4X, most people just start a new game. But the ambition system in Old World is supposed to address that. It seems odd that none of the guys doing an Old World podcast have seen a game through to the end. The default difficulty level is kind of a gimme, especially if you didn’t toggle Play to Win.
I hope they weren’t claiming this is unique to Old World. Age of Empires III did a great job keeping scouts relevant throughout the game (Rise of Nations before that, which also kept scouts relevant by making them the counter to spies). Old World is similar, but it has its own unique idea. Scouts give you resources for free when you discover resource nodes, but on top of that, they can harvest resource nodes, which remains relevant throughout the game (although less and less as resource nodes get annexed into cities). I especially love how they can boost culture by harvesting luxuries. It’s an excellent assist to getting your Weak cities on their way to Developing. But later in the game, after you’ve researched Portcullis, your scouts can give you permanent line of sight into any city and they unlock the option to install an agent, who I like to think of as a local ambassador. This, to my mind, is Old World’s real twist to scouting: keeping scouts relevant throughout the game as more than just a micro-intensive way to get a resource drip.
I’m not a coder and not a member of the design team, so correct: My conclusions are based solely on observations made by testing. Lots, and lots of testing and watching the AI’s behavior.
When threatened, the AI will absolutely make for best available defensive terrain consistently.
They will use any units nearby to “dogpile” an enemy unit and they consistently seek to kill, rather than simply wound.
I don’t have any information on how well or effectively they make use of promotions, however.
This seems a little unfair to me. The designers shouldn’t reveal what’s behind the curtain. Maybe what works the best is a random number generator making the decisions. They shouldn’t tell us that. They should preserve the magic. It’s up to us as the player to observe and interpret the actions.
It’s tough because it’s not cut and dried. Should an AI move a unit with Highlander into the hills? What if the AI can score a kill on an injured unit if it moves that highlander into the plains instead? There’s a bunch of decisions like that that a player has to weigh, but an AI isn’t playing a game. It’s just calculating numbers based on various weights. What is the terrain bonus weight vs the opportunity of killing a unit? Which is “right”? It’s going to really depend on the situation.
Vel isn’t part of the design team and can’t read code, but I can answer at least part of that. The AI’s aware of those factors when picking promotions / traits, and I know the AI considers urban, trees, etc at least defensively (not sure about offense, I’ll look it up if I can). So when the AI is about to move a unit to a defensive position, it will favor urban/trees.
I can see both sides of it. But…I got flagged in the original question, so…I answered it with the best info I had available. :)
I can also say with some confidence that the AI does indeed understand flanking. I say this again, based on observation, thusly:
If the AI didn’t understand flanking, then it would move a unit up, attack. Move another unit up and attack again, etc. Repeat until the target unit was dead (or the AI ran itself out of available units and orders).
But that’s not what the AI does. The AI will move all its units up into position (as many as it assesses, via some formula or another, it will take to kill the unit in question). It gets its guys into flank position first, and then attacks, getting the flank bonus.
Sometimes it doesn’t or can’t. I can only guess at the reasons there (not enough orders?), but when possible, that’s its preferred behavioral path.
(And i’ve heard from the team that the guy who does the AI watches my Twitch streams and has tweaked the AI’s behavior based on some of the s**t I try to pull…lol)
Well sure for us mere mortal that’s reasonable, but we are talking Tom “The Explainer” Chick. The man of the dreaded one star reviews. He didn’t earn that crogman, but not asking a ton of
questions. :-). All kidding aside, I’m genuinely interested, but you are absolutely right the developers don’t owe us answer.
Anyway, I’m sure all of the answers to all question about the game are revealed in the source code. It is just a matter of digging through it. (Of course I’ve not even located the source code yet, god my tech skills are shriveling)
