Patch update on the Firaxis livestream

[Ryika]

The AI never understood what they can and cannot give up anyway, so that's pretty much whatever. The important part is pricing stuff right, and at the moment it feels like you get way too much way too easily. You can basically get all your affinity for leveling up your units from opponents whose capital you took while you are on your way towards a turn 90-or-so domination victory. In longer games you can get SO many of the cheaper Affinity-Technologies that aren't your main Affinity that hybridization has become even easier than it already was.

Though to be fair, the fact that you can declare war on an AI, break its neck and move on in a total of around 5 turns is a big part of the problem.

 

[poom3619]

Did the AI ever reject a deal you offered, if it was within the war score? Are they forced to accept terms as long as they're entertaining peace offers? I can't seem to get them to ever actually entertain offers in the first place so I haven't been able to check.

I think that is how it is suppose to work. Problem is the war score needed to get city from them is too cheap in practical Player vs AI war. In EU4, AI willingness to fight war do weighted with war score when negotiation. Those factors mean something like, their army still can put up a fight against your, they feel you're trying to fight war for few months and peace out, or their nation is largely unscathed from war.

Someone will have to balance those "price" of city or rebalance the warscore system. (wish it is Firaxis) It is not the system that broken. It is the value assigned in the system. I imagine a system where "AI willingness to continue war" x "true warscore prize of thing" = "prize you will have to pay". AI willingness is 1.00 at lowest and go higher due to number of factors.

 

[omniclast]

It might seem to be a bad implementation but I think it actually makes sense now that the game has an actual system (war score + war spoils) that quantifies the peace terms. If the AI is losing badly, thanks to the war score, it knows the value of the stuff it will have to give up. So if the AI determines that it is ok with giving up that quantity of stuff then it makes sense that it would agree to the player's demands even though it does not know the specifics of the peace terms. And if the AI is winning, it gets to offer peace terms and the player is free to accept or decline. So that works. Lastly, if the war score differential is 0 or near zero then neither side can give up anything of value anyway. So, I think the system works well from this point of view.

Except in that respect it's not different than War Score at all. The AI determines whether to make peace based on a fixed set of terms that it has to give up for peace -- before it was just cities, now it's techs and stuff, but you still can't actually vary the VALUE of what you ask for just what specific items you get.

What if I want to make a lowball offer, because I care more about ending the war than the spoils? I can't, there's no way to even make that kind of offer. Instead I'm stuck trying to guess what I need to do to make them accept the terms I don't want to ask them for. Which is exactly where I was with War Score, i.e. endless wars all the time.

 

[SupremacyKing2]

Except in that respect it's not different than War Score at all. The AI determines whether to make peace based on a fixed set of terms that it has to give up for peace -- before it was just cities, now it's techs and stuff, but you still can't actually vary the VALUE of what you ask for just what specific items you get.

I'm confused. You can draft any peace term from 0 up to your war score differential. So, you can change the value of your peace terms.

What if I want to make a lowball offer, because I care more about ending the war than the spoils? I can't, there's no way to even make that kind of offer.

Yes, you can. You don't have to spend all your war score differential. You can choose to draft peace terms that are worth less than your war score differential. For example, my war score differential might be 400 but I can offer peace terms where the AI only gives me 10 energy (I only spend 20 war score differential).

Instead I'm stuck trying to guess what I need to do to make them accept the terms I don't want to ask them for. Which is exactly where I was with War Score, i.e. endless wars all the time.

Honestly, I don't think you understand how the system works because you are so used to the old civ5 system where the specifics of the peace term determined whether or not the AI would accept. In this new system, you don't have to guess what the AI will need in order to make peace. The AI tells you in advance if they are willing to accept peace, you only need to draft peace terms that you are happy with. If you are winning, you can ask a lot or you can be generous and ask nothing from the AI. And the AI will accept.

I've never found myself in endless wars in any of my games. I'd be curious to know what you are doing to cause what you perceive to be endless wars.

 

[SupremacyKing2]

Let me just add that in both BE and civ5, the player has to wait for the AI to be interested in peace. There are plenty of times in civ5, where you can ask for peace and the AI will tell you to take a hike. The real difference between the two systems, as far as I can see, is that civ5 was more free form. The negotiation table allowed an infinite number of combos. The player could be winning and could even offer to give the AI something to make the AI more likely to accept peace. In RT, the war score/war spoils makes things a lot more structured. The winner always takes from the loser. And the system quantifies what you can ask for so that you can't ask for more than you deserve to get.

 

[Aldor]

AI
o Increased AI desire for expansion via founding cities.

From all I've seen, there has been no change here at all. I'm mostly playing on Vostok, sometimes Gemini. The AIs usually build 2-4 cities, then never more. Very late in the games, whole continents are completely unsettled. As it is now, the game is just way too easy, as the AIs never expand, never go for any victory and instead only focus on warring between themselves (yeah yeah, I guess I should go to a higher difficulty).

This was better in vanilla. Not sure if its the new diplomacy (respect/fear) but after building their first cities, ALL the AI permanently war with each other. I suppose this is a factor that prevents them from building colonists.

IMO this would be THE best point to improve the AI: Expansion should not be a flavor thing, instead it must be hard coded and a priority. Then the war-likeliness can be toned down, and aggression can result more from settling too close to each other. Take a look back at Civ 4 for comparison: Even on Settler difficulty, the AI will continue expanding until the entire map is filled.

What's your experience with this?

 

[Gorbles]

From what I've seen, the AI expands like rabid bunnies on some form of illicit substance.

The problem here is given the amount of active posters there's never going to be a large-enough data set from Firaxis to reliably draw conclusions from.

 

[hevans900]

From what I've seen, the AI expands like rabid bunnies on some form of illicit substance.

The problem here is given the amount of active posters there's never going to be a large-enough data set from Firaxis to reliably draw conclusions from.

Just played a game on Apollo versus Elodie - she literally settled 15 cities in the first 120 turns or so. Madness.

Other sponsors seemed to be acting as usual though.

EDIT: can I just add that combining Chungsu/ARC with the new warscore system, you can get feasibly get all 42 techs in a ~200 turn game (which is silly).

 

[Aldor]

Other sponsors seemed to be acting as usual though

Which means?

Anyway, so on higher difficulties the AI builds more cities then?

I've started dozens of games on Vostok (a few on Gemini) and finished a fair amount of them. In literally every game the AI always have three cities, one usually builds 4 or 5, one only 2 cities.

And I'm sure this wasn't so bad in the base game. It could take a while there too there before they started expanding, but eventually they did (most of them at least). Now they never seem to.

 

[Gorbles]

If it helps, I regularly play on the bottom difficulties because I'm farming Artifacts and the like. I'll get back up to Soyuz eventually, just still exploring the expansion content really.

Artistic touches never cease to amaze me. The ocean / coast renderer grants minor detail differences on certain biomes (Arid has underwater plants where the coast turns to ocean, for example).

 

[Westwall]

Which means?

Anyway, so on higher difficulties the AI builds more cities then?

I've started dozens of games on Vostok (a few on Gemini) and finished a fair amount of them. In literally every game the AI always have three cities, one usually builds 4 or 5, one only 2 cities.

And I'm sure this wasn't so bad in the base game. It could take a while there too there before they started expanding, but eventually they did (most of them at least). Now they never seem to.

This was my experience too. Elodie settles but everyone else rarely has more than 3-4 cities.

I don't think I've ever seen Al Falah with more than 3.

 

[hevans900]

Which means?

As in, no noticeable change in the amount of cities settled in BERT release compared to BERT patch 1 (now).

I only play on Apollo (which imo is not even close to hard enough right now) so I can't speak for the other difficulties - but the AI on Apollo will always aggressively settle up to ~ 6 cities regardless of sponsor.

Elodie seems to behave differently than the rest currently, and will go wide as all hell.

I've also had one bizarre game with AI Hutama going completely crazy runaway - and I mean CRAZY - the AI seemed to be making full use of the additional trade-routes. He settled over 10 cities that game and conquered another 10 or so. This game was pre-patch.

I have only played 3 games since the latest patch - not enough to do a real analysis yet. I'll play 2-3 more games tonight and post back here tomorrow.

Games since BERT patch:

(Apollo games)

1. Chungsu/quick speed/dwarf map - turn 140 promised land victory (THIS IS SO EASY NOW WITH EMBARKED SETTLERS - 2 MOVESPEED + 6 SETTLERS PER OUTPOST
2. PAC/standard speed/standard size - turn 222 domination victory
3. Al Falah/quick speed/massive size - ongoing

 

[SupremacyKing2]

This was my experience too. Elodie settles but everyone else rarely has more than 3-4 cities.

I don't think I've ever seen Al Falah with more than 3.

I concur. The only AI that seems to expand a lot is Hutama. He's often the runaway AI, probably because even the AI can do well when you give it that many trade routes.

 

[Olodune]

I concur. The only AI that seems to expand a lot is Hutama.

I haven't noticed much of a difference in the AI expansion behaviour either. I have noticed that my (2) games have been about 10 turns slower, but this could just be because I've been messing around with the Warscore system.

I've seen Barre with 12 cities at turn 100 (Standard) pre-patch. That was unusual, but he often does well. The AU and Polystralia are the most reliable AI leaders.

I have never seen Al Falah do very well, but I have seen them build ~10 cities in two separate games.

 

[hevans900]

The AU and Polystralia are the most reliable AI leaders.

Elodie always seems to do well in my games, just had one game where I was forced to DoW because she was just getting such a ridiculous science lead

 

[Ryika]

In my games it seems like Elodie spends more time drinking le vin and eating le fromage than she actually spends working on her empire. She's always one of the worst Sponsors, which is to be expected given that she has no actual bonuses that really help her and high flavors for rather useless stuff.

 

[hevans900]

In my games it seems like Elodie spends more time drinking le vin and eating le fromage than she actually spends working on her empire. She's always one of the worst Sponsors, which is to be expected given that she has no actual bonuses that really help her and high flavors for rather useless stuff.

Maybe she's just getting lucky, or perhaps she is drinking crack-infused vino in my games

 

[Olodune]

One thing I do find annoying is that the AI keeps switching its traits. With the new patch this automatically cancels any agreements that were reliant on the old Trait. Annoying. Also very questionable decision making on the AI's part.

 

[Alkaid]

In my games it seems like Elodie spends more time drinking le vin and eating le fromage than she actually spends working on her empire. She's always one of the worst Sponsors, which is to be expected given that she has no actual bonuses that really help her and high flavors for rather useless stuff.

I don't feel like the abilities help the AI that much, since none of them really have any idea how to utilize them. The flavors definitely seem to be the main factor, if anything. It's usually pretty random who does really well or terrible in most games. However, even though her ability sucks and her flavors are meh, Elodie usually does among the best for me, but then once in a while she gets pretty much crushed by a neighbor that got bigger than her. Same with Reginaldo. Hutama used to be one of the constantly top AIs in my base BE games. Now he's usually just in the middle, despite having his insane ability and TRs taking no effort to use well.

 

[Ryika]

Even though her ability sucks, Elodie usually does among the best

That's an awkwardly generalized statement given the fact that in the last 6 Posts people have given very different impressions. ^^

I don't feel like the abilities help the AI that much

That depends on the Ability. Polystralia for example profits a lot from their Ability, as it's for the most part passive and will always provide relatively good results, even in the hands of an AI.

Same is somewhat true for Femin***-Germany, even if she picks garbage deals she still gets "more garbage deals" that somewhat help her out during the earlier phase of the game, no matter what she does. But it's certainly a lot weaker.

Then comes Franco-Iberia who basically gets nothing early on. An AI that doesn't snowball early on usually won't get better during the game. Of course that doesn't mean that she's "always bad", she can still have a good starting position, roll synergistic Character-Values or just be the first one to land if delayed starts are on to compensate. But on average it will drag her down.