Can we reframe the AI discussion

[Stringer1313]

Can we reframe the AI discussion and instead of asking, how can we improve the AI, shift to asking, what external/internal/environmental (or "exogenous" if you want to be fancy) challenges can the game introduce that is *NOT dependent (or much LESS dependent) upon the AI performance of rival civs*?

So for instance, Colonization was fun b/c your main competition was not the rival civs, but your mother country, which has an inherent and completely acceptable different rule set.

Here, I wish they would introduce more exogenous challenges or difficult quests especially in mid- to late-game that either happen randomly or semi-randomly especially if you are in the lead, or that are "triggered" if you reach certain victory milestones. This can involve things like, you have to steal X% of the world's great works or your tourism count gets knocked back by 50%, or you have to deal with sudden religious schism in your own territory whereby when you are close to achieving religious victory a rival religion splits off in your own civilization and spawns X missionaries and apostles that you have to deal with, or when you conquer 50% of the world's capitals, your rivals get +X permanent combat bonus for rest of the game b/c they are terrified (ok, that is somewhat dependent on rival AI performance.

Of course, what I'm saying is not a new idea; Jon Shafer I think is pursuing this and i'm sure others. I would love to see more discussion of this than with how to improve the AI (which is still an important discussion but i think disproportionate and has diminishing returns).

 

[Rosty K]

Well, for a start you could make factories (or power plants) require coal. And add some more resource requirements.

Then there is another thing, which the early game gets right, but it gets lost in the late game - certain game-changing milestones.

Discovering Iron - important, affects your expansion decisions
Political Science - important, you can use all types of policies now
Feudalism - much more tile improvements
Apprenticeship - production explosion

after that... nothing Well, maybe niter, but by that time you've expanded enough to have some of it nearby at least, and there are other ways to get your musketmen anyway - City States, Magnus, etc. Well, caravels on non-pangaea maps are also kind of a huge thing.

Civ 5 (in its final form) had these 'and now we start a different game' moments spread through the timeline almost to the end. I also seem to remember that Civ2 had it right somehow (I'd finish every game!), but it was so long ago, I've forgotten the details.

EDIT: actually, my games effectively end at Apprenticeship, because that's where I know I'm about to outproduce everyone.

 

[Sykes179]

I can't remember which iteration of the game had it but remember when pollution and sickness was a factor? Now you can have mega cities with factories, power plants, roads, airports etc and no downside. You had to build things / research techs etc to try and counteract.

Which game had the random natural disatsers? That could change things up a bit.

Having said that I've played in several games where I've won despite being well down on score and lost games where I've been well on top.

 

[Rosty K]

Oh, one more thing. Civ VI definitely has some unrealized potential in extra resources. Right now they are almost absent from the game, but you could give them a meaning.

Extra iron/niter - strength bonus to units and garrisons;
Extra coal - more production from IZ
Extra uranium - more production from power plants + more science from Campus, etc, etc.

 

[Reckoner]

Really nice suggestions regarding gameplay, I thought about redefining victory conditions:
You could introduce era-based "victories" of each kind, with threshold conditions. For example: winning 2 emergencies in a single era for domination, converting a certain number of cities for religious, gaining a certain number of tourism points for cultural and something(?) for scientific.

To actually win the game, a civilization must gain era-based victories of 3 different kinds. This could make the human players more balanced through the different mechanics, while the AI is already playing in a balanced way most of the time.

 

[Jaybe]

Instead of improving the AI, what you do is nerf the powers of the player where the AI is weak. So (pre-May patch), reduce the tactical control the player has of military units.* Of course, this would drastically change the game and reduce the joy of playing.

* Assign units to “groups” (I can’t say corps or armies for obvious reasons). Give groups orders such as “attack this city” or “approach this city, destroy their units and pillage their lands.”

 

[Trav'ling Canuck]

Here's my current thinking on how the AI should impact the player in single player:

1. It should roleplay each civ based on that leader's agenda / priorities so that games play differently based on who you start next to and how you want to approach your relationship with them. Certain civs (Shaka, Genghis, Alexander) should be more prone to attack. Others should be easier to maintain friendly relationships with. Your relationship with the AI civs should be controllable with enough time and effort, so that (in an ideal world) maintaining friendly relationships is roughly as difficult and rewarding as engaging in combat with them. If you are vulnerable, or close to winning, it should be more difficult to stay on good terms with them, but never impossible EDIT: what I meant was, but there should still be tools you can use to try and maintain good relationships. Changes in diplomatic relations should occur from time to time based on in game events (both player initiated and AI initiated such as changes in their government type or their relationships with other AI civs, i.e. they've come to hate a civ you're allied with), but the reasons for these changes should be knowable by the player with sufficient diplomatic effort. I personally think the AI is pretty good on this front in Civ 6 [EDIT: except for maintaining good relations being weaker than all out conquest, which is really an issue with the value/ease of straight conquest], but I know others want changes to the diplomacy system.

2. It should pursue a victory condition with sufficient competence as to be able to reach victory within an expected range of time in the absence of player intervention. This then sets an expected time frame by which the player knows they must be able to achieve their victory if they are going to win. So, for example, perhaps on Deity, standard settings, you should expect an AI civ to win between turns 250 and 300 in most games, whereas on Prince perhaps the expected time frame is between turns 450 and 500.

3. It should be capable of disrupting the player's plans, so that the game is not completely "Sim Civ". This means the player must spend some time thinking not just about what they are doing, but about how the AI could interfere, and taking risk mitigation steps. Beating you on Wonder races is part of this. So is the AI grabbing territory you want, or using Spies against you. The current AI is not bad at these, in my opinion. Having war be dangerous is another aspect of this that the AI so far in Civ 6 has struggled with.

In addition to these 3 elements, as others have mentioned "playing the map/game" in a broader sense could also be part of the challenge of winning at Civ. For me, what distinguishes the Civ series from other games, however, is an AI that creates the above 3 game elements.

 

[Phoenix1595]

I agree with Stringer— some level of powerlessness within one’s own civ would do wonders to improve gameplay without worrying to much about aia competitors. Historically, civs have dealt with their own internal issues (i.e. factionalism/civil wars, externalties with natural disasters/ecology, etc.). The series has dealt with these issues somewhat, but not in a really in-depth, substantive manner. Civ6 in particular, seems to have backed away a lot from this, although I do think the Ages system and loyalty are a step in the right direction.

I remember Civ 3-4 dealt with internal challenges fairly decently, especially with the random events in Civ4 and breakaway/vassal civs. All this said, I agree that AI competitors need a once-over before this title is through.

 

[Sykes179]

Are you kidding me? I've had civs buy tiles that border on to my cities to grab rsources. Infuriating but thats the game.

In a recent game I was near to getting a religion and then several turns later I noticed I still hadn't got one and when I checked all the religions had been taken. Couldn't understand how because when I'd checked my progress previously 2 were left. By going back to the auto saves I worked out that on the turn before I would have got a great prophet a civ bought one (how is that possible) and Arabia got the last one. I went back several times and every single time the AI did the exact same thing.

 

[gozpel]

You can buy Great People with faith or coin if you have it. If you had some faith or coin, you could of had bought it a few turns earlier. When it's only 5 turns left, you can buy a prophet quite cheap.

 

[NoonePerson]

• Change happiness functioning slightly (auto amenities via 4 largest cities is ehhh). Specific buildings for increasing local happiness yields based on local tiles or features.
  
• Housing should do an awful lot more than cap growth. Slums or urbanisation comes to mind. Really, housing is a perfect mechanism to interlink that and it's kind of wasted. All it is at the moment is an automated prevent growth button

 

[megabearsfan]

I'm not sure if Colonization's "win the revolution against the Motherland" condition is a very good model for future games. Basically, you play a game that's almost entirely about resource management and trade with little-to-no military gameplay, and then suddenly, you trip a switch and it becomes a shallow war game against an overwhelming opponent. The game is basically decided by whether you or not you knew (prior to flipping that switch) that you needed to start spending all your resources on building weapons and soldiers. Plus, historically speaking, not all European colonies fought a revolution for their independence. Some were granted independence without a fight, and some were basically just abandoned by their mother country as the European empires crumbled. So it doesn't really make historical sense that the revolution was the only path to victory in Colonization. I do agree, however, that the core game, up until the revolution against the Motherland was pretty enjoyable. I think there are valid lessons that core Civ franchise could learn from Colonization (and from Beyond Earth, for that matter). I'm just not sure the Revolution mechanic is one of them.

Last year, I wrote a blog post about my frustrations with the way that Civ handles difficulty settings in general. It was mostly focused on how all the A.I. bonuses are front-loaded in the game, which doesn't make for a compelling long-term challenge.
Feel free to read it if you're interested: http://www.megabearsfan.net/post/2017/06/03/Frustrations-with-Civ-difficulty-levels.aspx

In the second part, I also talked about ideas that I had for expanding the sources of in-game challenge in the hopes that it could "offload" some of the burden from the A.I.:
http://www.megabearsfan.net/post/2017/06/05/Expanding-challenge-in-Civ.aspx

First and foremost, I think the player should have more granular control of difficulty settings when creating a game. There should be more options for player versus A.I. handicaps (such as tweaking growth, production, science, etc handicaps individually), more options for tailoring A.I. behaviors, and more options for tweaking things like barbarian activity and aggressiveness, and how frequently ancient ruins / tribal villages / goody huts appear (as opposed to a binary on/off switch). A lot of sports games already have similar options and sliders that allow the player to custom-tailor the game experience to provide a challenge most suitable to both their skill level and their play styles (see the linked blog post for examples). So why not Civ?

Changing the victory conditions so that players aren't bee-lining to a very specific and limited victory might also help even out the difficulty.

As suggested by the OP here, I'm also open to the idea of more "player versus the board" challenges over the course of the game. Things like dealing with climate change (including natural droughts, floods, disasters, ice ages, and anthropomorphic disasters such as deforestation, pollution, and global warming), dealing with internal politics, managing resource-depletion, market shifts, and other entropic or asymmetrical game effects could all be valid options. There's two main issues with this to consider, and we talked about it a lot on this morning's PolyCast (ep 308). One is that you'd also have to program the A.I. to handle these things too, unless they are asymmetrically designed to disproportionately (or exclusively) affect the player. The second problem is with getting over the hurdle of game-changing events feeling "random" and/or unfair. We generally don't want game-changing events to be random, nor do we want to burden the player with a bunch of trivial inconveniences that just add more tedious dismissal of dialogue prompts. If you're constantly being stopped by dialogues asking to chose an option, and players are always picking the same options, then you're just wasting the player's time without adding any value. Beyond Earth was full of those sorts of thing. So any such mechanic needs to be a predictable, system-based mechanic that gives the player meaningful choices.

The Ages system in Rise & Fall is actually a really good example (IMO) of what I'm talking about. Golden Ages aren't random events that affect certain players at a time. They are a goal that you work towards, and there's also value in falling into a Dark Age as long as you can manage the loyalty penalty, so there's a strong element of player choice involved as well. It's not a perfect system. I think Dark Ages aren't punitive enough, the lack of a reward for exceeding the score needed to trigger a Golden Age is annoying, and there are some things that probably deserve Era Points but don't award anything, etc.. But I think the concept could provide a really good framework for providing systems that feel dynamic and add meaningful, substantive choices for the player, while not proving too taxing for the A.I..

 

[NukeAJS]

Putting road blocks in the player's way could be done well if implemented correctly.

However, I think the best way to do it would be some sort of crisis. If a run-away civ appears, it could trigger some sort of internal crisis. Some people mentioned civil war, religious schism, and natural disasters. Those are all good ideas and more could be added. That being said, there's already a game system in place that could be co-opted for this purpose -- the era/ages system.

If a runaway happens; first, make it near impossible to get a normal age. To avoid a dark age, the player will really have to focus on getting as many era points as possible. To explain it, call it complacency. When a dark age happens this way, a civ MUST run a special dark age policy in a special dark age policy slot. This would be different than the dark age policies we currently have in that it would always be negative and it would be a substantial negative. An alternative (and better way to do it imo) would be that a runaway civ in a dark age would make their normal dedication in addition to their "complacency dedication" (the one that's always negative). This way the player can't switch out their complacency card -- they're stuck with their decision for the rest of the era.

Regardless, such complacency cards/dedications would be punishing and their effects would get worse and worse the later the era. A few simple ideas would be -3 amenities all cities, all units cost +5 gold a turn, all diplomacy is blocked for the era -- things like that.

The idea would be to integrate the era system and handicap the runaway civ while presenting interesting challenges that have historical precedent. Such handicaps need to be punishing enough to actually pose a real (if temporary) threat to the runaway's victory. Another alternative is that all the other civs would vote on your handicap. This could possibly help integrate some of the complacency dedications like all diplomacy is terminated and impossible for the rest of the era. But more importantly, your rivals would get to hit you where it hurts the most. After all, what's the big deal of having a religious schism is you're going for a science victory or paying more upkeep for your units if you only have a small self-defense force? If the voting route is taken, perhaps the emergency mechanic is the better vehicle to implement such a crisis.

 

[bbbt]

More internal tension - for example if something like idealogies returned (or even with governments) and spread (akin to religious pressure but stronger) and citizens that weren't ofyour idealogies took happiness.

Heck even making it easier for cities to become unhappy and rebel under the current model (i.e. currently a city won't become a 'free city' without external pressure).

I suggested in another thread but making natural disasters hit the human player harder than the AI on higher difficulties.

 

[Anno]

I think getting rid of victory conditions, or really just the concept of victory at all, would go a ways towards reframing how people perceive the AI, and how the AI is able to be coded. I've liked playing Civ way more ever since I stopped caring about winning or not and just focused on building a cool civilization.

 

[Trav'ling Canuck]

I think getting rid of victory conditions, or really just the concept of victory at all, would go a ways towards reframing how people perceive the AI, and how the AI is able to be coded. I've liked playing Civ way more ever since I stopped caring about winning or not and just focused on building a cool civilization.

This is exactly the direction the development team is trying to go with Civ 6. The marketing for the Switch makes that very clear.

 

[Leathaface]

But the game ends when you achieve your victory condition. How could a victory even get rid of? There'd be no point to the game.

 

[Cagarustus]

This is an interesting thread: it seems that Fireaxis has created the environmental mechanics as a way to pose challenges to the player without focusing too heavily on the AI. Although I’m sure they’ll fix the AI quite a bit in this patch.

 

[shaglio]

But the game ends when you achieve your victory condition. How could a victory even get rid of? There'd be no point to the game.

One thing I’ve yet to do (but this thread is making me consider trying) is turn off all victory conditions except for time victory. Theoretically, that would achieve what was suggested, although the actual calculations for time victory really need an overhaul.

 

[Trav'ling Canuck]

This is an interesting thread: it seems that Fireaxis has created the environmental mechanics as a way to pose challenges to the player without focusing too heavily on the AI. Although I’m sure they’ll fix the AI quite a bit in this patch.

I hope you're right regarding the AI.

Whether the environmental mechanics pose an interesting challenge to the player will really depend on how they're implemented. If it's as simple as figuring out "I can chop 10 tiles each game, so let me pick which ones when" that's still an improvement over the current situation, but less interesting as a speed bump than if you also have to take into account what the AI leaders are doing and figure out what's optimal for this particular game.