How to respond to AI denunciation?

[Thalassicus]

It's a few dozen small number tweaks. I just did a pass through the AI personality files, flavors, diplomacy defines, and war tactics and updated things in the never-ending quest to get the AI to make smarter decisions. For example, one change I made is I slightly reduced overall AI willingness to build military units (-1), yet increased that willingness if the city has hills or forests nearby (+1 each).

 

[Dunkah]

Actually, the AI here appears to be very simple: it just counts the tiles to see if the target is closer to your capital or theirs. So if there's 20 tiles between your capitals and you settle 11 tiles from yours that causes anger, but 9 tiles from your capital is okay. It's a very hard line. I noticed this pattern from watching the effect placing cities has in various games.

Then perhaps it should be easy to give a warning to the player before they actually found a city too close then.

Perhaps a yellow cone of Danger eminating from the closest city when you have the settler selected that means stay out of this area if you don't want a diplo hit...

Is this calulation to the closest city or to the capitol? Does the AI all of a sudden start getting greedy once they settle city #2 halfway between your city and thiers, to where now the distance is halved again?

 

[Thalassicus]

It's to the capital.

I think it should be feasible to give a warning in advance, but it does depend on other factors, so only displaying the halfway point to capitals might be misleading. There are things affected by the "closeness" rating... this categorizes civs into Neighbors, Close, Far, or Distant. If I recall, Neighbor is defined as cultural borders within 7 tiles of one another, modified by map size. This affects things like targeting priorities for warfare.

 

[Jorlem]

Does the "Don't settle new cities near me" choice under discuss actually impact the AI's settling pattern in some way?

 

[Txurce]

Have you ever seen an acquiescent response? I don't think I have. I pretty much use it to provoke an AI into war. That (and other similar actions) do work.

 

[black213]

They will only answer positively if they are afraid of you or know that you could easily crush them. I don't think it really works in any kind of manner.

 

[Ahriman]

I think it should be feasible to give a warning in advance

A warning would be very helpful.
It would also be nice to be able to see the duration for some of these things; if I say I won't settle near you or that my troops are just passing through and not attacking you, how many turns do I have to wait until I can settle nearby or attack you without getting "dishonorable"-type penalties? Where can I see if these have run out or not?

[A warning of diplomatic penalties from declarations of war would be nice too.]

 

[Txurce]

To me, a warning about potentially building too close to the AI belongs in the Advisors' prompts, not a game warning. But based on what Thal wrote, it sounds like an accurate warning isn't feasible. And just counting tiles from the capital in cases where it can't be eyeballed isn't too much micro-management.

I would also like to know how long denunciations and breaking a "not expand so close" or "just passing through" agreement last. But if the info were going to be available in-game, I think it belongs in a mod like InfoAddict, because these are "understandings" that in a perfect game would vary in length with leaders' personalities and unforeseen circumstances.

In the case of a DoF, as long as it's still a treaty that can be terminated early, then I think it should be listed as a "current deal."

 

[Ahriman]

And just counting tiles from the capital in cases where it can't be eyeballed isn't too much micro-management.

Yes it is. Most players will have no idea that diplomacy is based on number of tiles from the capital. It is totally unreasonable to expect to know this. It would be very reasonable to give a prompt to the player when about to build a city that doing so will be seen as a violation.

But if the info were going to be available in-game, I think it belongs in a mod like InfoAddict, because these are "understandings" that in a perfect game would vary in length with leaders' personalities and unforeseen circumstances.

I don't really understand what you're saying here. The game has an artificial length of time where attacking would be seen as a breach of trust. Everyone in the game world knows what that is - they know that attacking on turn X is a breach of trust but that it wouldn't be on turn X+2.
So I don't see why it shouldn't be readily available to the player.
The right place to put the notification is when you are about to undertake an action that will trigger the penalty, not buried in the background somewhere.
The game should be designed so that players understand the consequences of their decisions; that is the only path to making decisions meaningful.

 

[Txurce]

Yes it is. Most players will have no idea that diplomacy is based on number of tiles from the capital. It is totally unreasonable to expect to know this. It would be very reasonable to give a prompt to the player when about to build a city that doing so will be seen as a violation.

I did say an Advisor prompt would be a good thing. But again, it sounds like a moot point, since it's apparently a complicated equation.

I don't really understand what you're saying here. The game has an artificial length of time where attacking would be seen as a breach of trust. Everyone in the game world knows what that is - they know that attacking on turn X is a breach of trust but that it wouldn't be on turn X+2.
So I don't see why it shouldn't be readily available to the player.
The right place to put the notification is when you are about to undertake an action that will trigger the penalty, not buried in the background somewhere.
The game should be designed so that players understand the consequences of their decisions; that is the only path to making decisions meaningful.

I'm saying that an understanding (as opposed to a formal agreement) ought to be a fuzzy mechanic. Most games aren't capable of programming this, but may choose to hide the mechanic in an effort to approximate it. This is how I choose to interpret Civ 5 with regard to all of the points we're discussing (except DoF's).

Again, I can see why someone may prefer to know it in-game, and I wouldn't argue against it, now that I've stated my pov on it.

 

[Ahriman]

I did say an Advisor prompt would be a good thing. But again, it sounds like a moot point, since it's apparently a complicated equation.

The fact that it is a complicated equation should increase the need for making a simple way to communicate the data to the player.
Unless you mean it is more complicated than we understand (because we don't have access to the relevant code)?
[I'm not sure what you mean by advisor prompt; I would want this to be information you got even if you had turned the tutorial advisors off.]

I'm saying that an understanding (as opposed to a formal agreement) ought to be a fuzzy mechanic.

Why? The mechanic isn't fuzzy; you either get a penalty or you don't. It is binary. And so it should be communicated to the player in that way.
If you're saying that you don't think the penalty should be simple then I would disagree as well, that would seem like unnecessary complexity. I don't see any reason for Catherine to care more or less about you taking her land than Washington would.

This is how I choose to interpret Civ 5 with regard to all of the points we're discussing (except DoF's).

Why would you prefer mechanics to be fuzzy? Civ works best in areas where the mechanics are really clear. If you work this tile, you get exactly X food/hammers/gold. Technology Y costs exactly Z beakers to research. A factory will increase city production by A. Imagine how frustrating the game would be if these mechanics were fuzzy.
It works less well in areas where we don't know what is going on (like in combat calculations, where we don't actually get to observe the probability of various outcomes).

I'm also unclear as to why you think that "I won't attack you right now" is necessarily informal, or should be. It seems terrible design to me for a player to not know if an action is viewed as a breach or not.

 

[Txurce]

The fact that it is a complicated equation should increase the need for making a simple way to communicate the data to the player.
Unless you mean it is more complicated than we understand (because we don't have access to the relevant code)?
[I'm not sure what you mean by advisor prompt; I would want this to be information you got even if you had turned the tutorial advisors off.]

I interpreted Thal's post as meaning that an accurate warning isn't feasible.

Why? The mechanic isn't fuzzy; you either get a penalty or you don't. It is binary. And so it should be communicated to the player in that way.
If you're saying that you don't think the penalty should be simple then I would disagree as well, that would seem like unnecessary complexity. I don't see any reason for Catherine to care more or less about you taking her land than Washington would.

I said the mechanic isn't fuzzy, but that my view of the underlying concept is. In my opinion Catherine may not just care more than Washington - she may care longer. This could be the case due to any number of circumstances (which Civ happily provides). To be clear, I know that the game isn't likely to do this any time soon.

Why would you prefer mechanics to be fuzzy? Civ works best in areas where the mechanics are really clear. If you work this tile, you get exactly X food/hammers/gold. Technology Y costs exactly Z beakers to research. A factory will increase city production by A. Imagine how frustrating the game would be if these mechanics were fuzzy.
It works less well in areas where we don't know what is going on (like in combat calculations, where we don't actually get to observe the probability of various outcomes).

As per above, I would want a fuzzy mechanic in service of an ideally fuzzy concept such as "attitude." I would not want a fuzzy mechanic for production or research, unless the game were to become so complex as to allow for weather, etc. The latter would also be difficult to balance, whereas "attitude" is already accepted as unbalanced - Alexander responds differently than Harun.

I'm also unclear as to why you think that "I won't attack you right now" is necessarily informal, or should be. It seems terrible design to me for a player to not know if an action is viewed as a breach or not.

I view it as informal because it's not a treaty. Again, I would like to know how long these understandings last, so I have a rough sense of when it expires without micro-management. My subjective preference is to not get a warning about it, but would have no problem if the game did warn us.

 

[Ahriman]

As per above, I would want a fuzzy mechanic in service of an ideally fuzzy concept such as "attitude."

It would seem reasonable to have intensity attitude modifiers, that one person cared more than others; it might be that Oda Nobunaga gets enraged if you are dishonorable while Montezuma doesn't care so much.
I wouldn't mind the player not know exactly how pissed off the AI would be.
But the player should definitely know *if* the AI will be affected or not. They don't necessarily need to know "how serious a breach will this be" but they do need to know "does this action count as a breach or not".

So having variation in duration would be much worse than variation in intensity.

I view it as informal because it's not a treaty.

Couldn't this be viewed as replicating what might really be a treaty in-game? The mechanic is an abstraction.
Real world leaders would have a pretty good idea about what actions would be seen as dishonorable or not by other countries.

Again, I would like to know how long these understandings last, so I have a rough sense of when it expires without micro-management

And you wouldn't be frustrated if you thought you had waited long enough for the warning to not apply any more, and then got hit with a penalty?

Having variable duration wouldn't be so bad if the intensity ramped down over time (eg: you get -10 penalty if you breach it on turn 1, with the size of the penalty reducing 0.5 per turn so that after 20 turns no-one cares anymore), but variable duration (and uncertainty over that duration) is very bad in a system where the outcome is binary (ie: you breached the agreement or you didn't breach the agreement).
I don't understand what value is served by making the player unaware of whether their action counts as a breach or not, and hitting them with the full penalty if they accidentally get it wrong.
If you're hitting players with a penalty, they need to know that so that they can decide if it is worth it or not.

 

[Thalassicus]

To clarify a bit: an accurate warning of the 50% distance between capitals midpoint should be easy (in theory) since it's on the human-only user interface, and there's functions available to us to calculate the distance between tiles.

The part that would be difficult to represent is "closeness." It primarily affects AI military considerations like who to declare war on, and what cities to target. The variables that sort a person into the four categories (neighbors/close/far/distant) have a range of semi-random values. I have not done any experimentation with these to see exactly how they work:

07 PROXIMITY_NEIGHBORS_CLOSEST_CITY_REQUIREMENT
11 PROXIMITY_CLOSE_CLOSEST_CITY_POSSIBILITY
25 PROXIMITY_CLOSE_DISTANCE_MAP_MULTIPLIER
20 PROXIMITY_CLOSE_DISTANCE_MAX
10 PROXIMITY_CLOSE_DISTANCE_MIN
45 PROXIMITY_FAR_DISTANCE_MAP_MULTIPLIER
50 PROXIMITY_FAR_DISTANCE_MAX
20 PROXIMITY_FAR_DISTANCE_MIN

 

[Txurce]

It would seem reasonable to have intensity attitude modifiers, that one person cared more than others; it might be that Oda Nobunaga gets enraged if you are dishonorable while Montezuma doesn't care so much.
I wouldn't mind the player not know exactly how pissed off the AI would be.
But the player should definitely know *if* the AI will be affected or not. They don't necessarily need to know "how serious a breach will this be" but they do need to know "does this action count as a breach or not".

So having variation in duration would be much worse than variation in intensity.

Agreed.

Couldn't this be viewed as replicating what might really be a treaty in-game? The mechanic is an abstraction.
Real world leaders would have a pretty good idea about what actions would be seen as dishonorable or not by other countries.


And you wouldn't be frustrated if you thought you had waited long enough for the

Having variable duration wouldn't be so bad if the intensity ramped down over time (eg: you get -10 penalty if you breach it on turn 1, with the size of the penalty reducing 0.5 per turn so that after 20 turns no-one cares anymore), but variable duration (and uncertainty over that duration) is very bad in a system where the outcome is binary (ie: you breached the agreement or you didn't breach the agreement).
I don't understand what value is served by making the player unaware of whether their action counts as a breach or not, and hitting them with the full penalty if they accidentally get it wrong.
If you're hitting players with a penalty, they need to know that so that they can decide if it is worth it or not.

I agree that players should know there is a penalty. But my ideal would be keeping the AI attitudinal specifics opaque. (I realize this is unlikely to happen.)

 

[Dunkah]

It would seem reasonable to have intensity attitude modifiers, that one person cared more than others; it might be that Oda Nobunaga gets enraged if you are dishonorable while Montezuma doesn't care so much.
I wouldn't mind the player not know exactly how pissed off the AI would be.
But the player should definitely know *if* the AI will be affected or not. They don't necessarily need to know "how serious a breach will this be" but they do need to know "does this action count as a breach or not".

So having variation in duration would be much worse than variation in intensity.

I love this... it seems reasonable to assume that if you settle too close to Monty he's going to be less happy than settling next to Ghandi but how close is too close?

But also the distance a civ may consider "reasonable closesness" to thier borders could change based on current attitude and leader. Freindly Ghandi may have a formula that looks like (D/2)-2 hexes (times map size multiple)for an acceptable distance. Where as an Unfriendly Monty may be (D/2)+4 hexes. "Don't even try to settle in my sacred lands Mortal!"

A shaded area of hexes that shows up when a settler is selected and about to settle could indicate the area to stay away from. Then there is no Leader screen interaction...

If you have an advisor give you the warning. Would that shut off if you turned off advisor notifications? If it were me i would still want that warning, even long after shutting down the advisor.

 

[Ahriman]

but how close is too close?

There should be a simple defining line that determines what too close is, and you should get a warning "this city will be too close, it will violate your promise, are you sure you want to build here?" if you try to build such a city.

the distance a civ may consider "reasonable closesness" to thier borders could change based on current attitude and leade

I would stay well away from this. It is ok to have *one* intensity variable that varies across leaders, not multiple ones. It is ok to have a leader-varying intensity variable, but it is not ok to have variation in whether a reaction is triggered or not. That just becomes far too complicated for the player for too little design/flavor gain.

A shaded area of hexes that shows up when a settler is selected and about to settle could indicate the area to stay away from

Would be great, but probably hard to code.

If it were me i would still want that warning, even long after shutting down the advisor.

Right, which is why I would keep this separate from advisors.

 

[Thalassicus]

Citystates operate in a similar ways.

The "interfering with my sphere of influence" warning comes from citystate allies closer to the AI capital than our capital. This is not the same as making CS allies others have influence with, the seperate "I noticed your attention with Singapore" warning. The second situation is very obvious now since I added the "Rival" display to the citystate tooltips. These two independent modifiers appear to combine for the "competing for citystates" diplomatic penalty, which is affected by each individual leader's "MinorCivCompetitiveness" personalty characteristic.

I'm honestly not very familiar with programming dialogue boxes or windows in this game, so the halfway-between-capitals warning probably won't get done until I learn this UI programming from integrating the basics of some other UI mods into CivUP. It's on the todo list now.

 

[Dunkah]

There should be a simple defining line that determines what too close is, and you should get a warning "this city will be too close, it will violate your promise, are you sure you want to build here?" if you try to build such a city.

Agreed... but the formula to determine where this simple line is, can very easily be complicated. (Based on current attitude). As long as a simple line and a warning is shown when you go to actually place the city then what does it matter how complicated the formula is that determines where the line is? You know if I settle on this side of the line I am facing consequences, the other side is safe.

I think that this is great way to flavor personalities. Ghandi may not care how close you get to his city as he is probably not going to expand much anyway, and even if you do expand close he may not "do" anything about it. Whereas Sulemain may really care about "Just how close" you get and give you an unfavorable diplo hit, and lastly Monty may declare war based on your usurping his sacred lands.

 

[Ahriman]

The "interfering with my sphere of influence" warning comes from citystate allies closer to the AI capital than our capital.

Interesting! I didn't know that.

Agreed... but the formula to determine where this simple line is, can very easily be complicated. (Based on current attitude). As long as a simple line and a warning is shown when you go to actually place the city then what does it matter how complicated the formula is that determines where the line is? You know if I settle on this side of the line I am facing consequences, the other side is safe.

Because it should be possible for advanced players to know what causes the distance without getting the prompt. I can easily imagine a situation where there are two good city spots, one to the east, one to the west, and I have a new settler. I send it to the one to the east, but then unexpectedly find that the AI over there is super-sensitive and would consider it a violation. I then have to spend 10 turns moving the settler over to the spot to the west if I don't want to trigger a violation.
So, basically: even with a notification, it can still be "too late" by the time the notification is triggered.

I think that this is great way to flavor personalities.

I disagree, I don't really think it adds much flavor. Its kindof possible to think of different leaders as caring more or less about following agreements or not, but its a bit hard to think of clear differences in how much civs care about you being near them or not, and those are mostly caused by geography, not by culture or personality.