Overextension warnings

[Seravy]

I'm changing them.

Old :

If player has 4*(land size setting+1) cities or more then there is a 25% chance to activate.
If it activates, generate a ((Land Size+1)*4-Cities)/(8+Land Size-Difficulty)*random(4)/3*4 strength diplomatic warning event from all other wizards. This can't be stronger than -15 and is halved if the player receiving it is an AI.
Allies and uncontancted wizards are not giving a warning.
(land size is 0-4, difficulty is 0-6)

New :

If player has 7*(land size setting+1) cities or more then there is a 1/12 chance to activate.
If it activates, generate a -(6+random(4))*(Difficulty level+2)/4 strength diplomatic warning from all other wizards. Double the amount if the player also has at least 10*(land size setting+1) cities. Halve the amount if it's an AI.
Allies and uncontancted wizards are not giving a warning.
(land size is 0-4, difficulty is 0-6)

Which means a loss of this much per year :

Fair land size, Fair Difficulty : 0-20 cities, no warning. 21-29 cities lose 6-10 REL, 30+ cities lose 12-20 REL.
Fair land size, Lunatic Difficulty : 0-20 cities, no warning. 21-29 cities lose 12-20 REL, 30+ cities lose 24-40 REL.
Huge land size,Fair Difficulty : 0-34 cities, no warning. 35-49 cities lose 6-10 REL, 50+ cities lose 12-20 REL.
Huge land size,Lunatic Difficulty : 0-34 cities, no warning. 35-49 cities lose 12-20 REL, 50+ cities lose 24-40 REL.
Tiny land size,Fair Difficulty : 0-6 cities, no warning. 7-9 cities lose 6-10 REL, 10+ cities lose 12-20 REL.
Tiny land size,Lunatic Difficulty : 0-6 cities, no warning. 7-9 cities lose 12-20 REL, 10+ cities lose 24-40 REL.


Number of cities sounds about right? Maybe the penalty amount is too much? The concept is to have less frequent warnings with higher effects, to avoid triggering a "repeated warning war" too easily. Maybe difficulty shouldn't raise it so much that Lunatic is double of Fair? With (difficulty level+4)/6 in the formula instead, it would only increase by 66%, to 20-33 instead of 24-40. (Which is still a hell of a lot...idk. )
...if the higher amount of cities would do +6 instead of "double" then...

Fair land size, Fair Difficulty : 0-20 cities, no warning. 21-29 cities lose 6-10 REL, 30+ cities lose 12-18 REL.
Fair land size, Lunatic Difficulty : 0-20 cities, no warning. 21-29 cities lose 10-16 REL, 30+ cities lose 20-30 REL.
Huge land size,Fair Difficulty : 0-34 cities, no warning. 35-49 cities lose 6-10 REL, 50+ cities lose 12-18 REL.
Huge land size,Lunatic Difficulty : 0-34 cities, no warning. 35-49 cities lose 10-16 REL, 50+ cities lose 20-30 REL.
Tiny land size,Fair Difficulty : 0-6 cities, no warning. 7-9 cities lose 6-10 REL, 10+ cities lose 12-18 REL.
Tiny land size,Lunatic Difficulty : 0-6 cities, no warning. 7-9 cities lose 10-16 REL, 10+ cities lose 20-30 REL.

Yes, this sounds better. Which would mean this rule :


If player has 7*(land size setting+1) cities or more then there is a 1/12 chance to activate.
If it activates, generate a -(6+random(4))*(Difficulty level+4)/6 strength diplomatic warning from all other wizards. Add 6 to the random value if the player also has at least 10*(land size setting+1) cities. Halve the amount if it's an AI.
Allies and uncontancted wizards are not giving a warning.
(land size is 0-4, difficulty is 0-6)

[Nelphine]

I think that's too many cities.

On fair, I usually end up around 35-40 cities. However, I'm better at placing the cities than the AI, so the ai would probably end closer to 32-35

With the old method, that meant the warning kicked in when you reached 1/3 of the plane. That meant it kicked in late enough that your first arcanus opponent never got it, and your second arcanus opponent only got it if you were already at least as strong as both remaining wizards.

That feels like the right amount.

The suggested method means it never kicks in until you are stronger than all remaining wizards combined on your plane. I think that's too late.

I wouldn't go higher than than 5 times land size +1. That would be 15 cities on fair, which would requirevyou to have more than your share, but not yet be more than half the plane.

But given you're also making them less frequent, I'd probably actually leave it at 4 times.


I'd make it 8 times for the higher amount (7 times if you leave it at 4 times for base).

[Seravy]

Do note the old formula used (owned cities-threshold) as the baseline for the penalty amount.
Until you exceeded the threshold by a significant number of cities, the amount of the penalty was minimal.

For example, with a 12 threshold (4/land size),  if you owned 14 cities, you got (assuming all settings on Fair)

((Land Size+1)*4-Cities)/(8+Land Size-Difficulty)*random(4)/3*4 = -2/(8+2-2)*random(4)*4/3 =
-2/8*random(4)*4/3=-random(4)/3 = 0 to -1 but the division is done first and rounds down so always 0.

It's dividing by 8, so you need to exceed the 12 threshold by at least 8 cities to get a penalty that's nonzero. Unless, could it be that on negative numbers, it rounds "down" towards -1, not zero?
ehh, this whole formula is so wrong...

anyway, the old intention was for the warning to come with an insignificant amount of penalty if the threshold was only slightly breached, the player did need a much higher city count for it to matter.

The difference is, now the cutoff is higher to remove all the annoyance of having to deal with 0-1 strength warnings which don't really do anything but sometimes still mess your game up if you roll that 1 in 100 chance of losing a treaty for it.

If I ignore rounding, the old formula was 1/6 to 4/6 REL penalty for each additional city exceeding the limit, which if I multiply by 3 to consider the difference in frequency, would be 1/2 to 2 REL.

12 cities : nothing
13 cities : 0.5-2
14 cities : 1-4
15 cities : 1.5-6
16 cities : 2-8
17 cities : 2.5-10
18 cities : 3-12
19 cities : 3.5-14
20 cities : 4-16
21 cities : 4.5-18

So at 21 cities, the old was 4.5-18, the new is 6-10, but the old had all sorts of rounding down so the effective penalty was much less. In fact looking at how we calculated, I'd say below 12 would get rounded to 0... let's see...

(12-21)/8*random(4)/3*4=(-9)/8*random(4)/3*4=-1*random(4)/3*4 ... you needed to roll a 3 or 4 to get a penalty of -4, otherwise you got nothing. So you had an average of -2 per activation, so -6 in an entire year. Now it'll be -6 to -10, which is ...more, not less.

About the number of cities, Arcanus has 20% extra land now due to the Myrran change. So people will have 20% more cities than before.

Not sure what's the design goal of your numbers? Mine is to have the warning kick in around the time the player controls about half the territory on Arcanus, after they eliminated one (or maybe two) wizards. Any earlier than that and we are just pushing a 4 direction early war on the player for no reason.

[Seravy]

Think I'll remove the random part from the formula entirely. The frequency is already random so there is no point.

So 6*(Difficulty+4)/6 ...wait then there is no point having the 6* and /6 parts...

Difficulty+4, and an additional 6 if exceeding the second threshold. That's simple and easy to understand.

6 or 12 for Fair, 10 or 16 for Lunatic.

Let's think about it a little. Natural REL gravitation is about 2 points per year for each 9 REL difference you have.
So with the "weak" warning, you would stabilize your relation at 27 points below the starting amount, but the frequency is random so it sometimes dips even lower. That sounds about right, low chance of war with people who are neutral or above starting relation, high chance otherwise. (REL needs to dip below -40 for a war declaration based on REL only.)
At stronger overextension, REL stabilizes 54 points below starting. That's guaranteed war with pretty much anyone unless you have a treaty to improve REL faster or something like that.
Both of these for Fair difficulty, which I think is too harsh? Would be ok for Advanced or Expert.

Basically, each 1 REL loss from this makes your REL stabilize 4.5 lower than usual. If we want Expert to do the -27 thing, then we need Difficulty+2 in the formula, which means :
Easy : -9
Normal : -13.5
Fair : -18
Advanced : -22.5
Expert : -27
Master : -31.5
Lunatic : -36

For the strong REL loss, we want, let's think about it... I guess the -54 sounds about right for Expert as well, so we can keep the +6 for this, meaning :

Easy : -36
Normal : -40.5
Fair : -45
Advanced : -49.5
Expert : -54
Master : -59.5
Lunatic : -63

All these are long term numbers, the change is only 1/4.5 of that per year, minus the effect of gravitation, so it will likely take about 6-8 years to actually reach those numbers. (and ofc, it's all random...)

That's too slow, and REL gravitation rounds its effects up so it's also stronger than assumed. So we probably should push the penalty up by another +2... then we're back to the "Difficulty+4" formula we already had, nice. This would stabilize about 9 REL lower than the listed numbers, but that much is needed if we want the war to happen in a timely manner at least on average RNG rolls.

So, REL loss of (Difficulty+4) for (Land size*7) or more cities, REL loss of (Difficulty+10) for (Land size*10) or more cities.

This btw will make lawful wizards a pain, as they double these penalties... assuming you didn't make a treaty to offset losses, it'll escalate into war quite easily.

[Seravy]

I noticed this entire block is only executed once every 2 turns so in reality it was 1/8 turns for the old, not 1/4, making it half as powerful as calculated...

(this also means we'll need to make it 1/6 chance instead of 1/12 for the desired outcome)

...overall it means this change will make these penalties a lot worse than they used to be, I'm not so sure about this anymore.

Maybe it should be Difficulty+3 instead of 4...

[teelaurila]

I seem to be always getting these when banishing the first AI. Is there a connection (score?)?

Could at least the severity also be reduced for Wizard Pact, perhaps also the limit increased? Or relation rather than pact? What I don't like here is that there seems to be a hard cap in city # after which relations start to sour with everyone who are not yet allied. So if you were making progress for an alliance, suddenly it slips out. To make it worse, the cap is not visible.

Otherwise perhaps the right stage is when you have defeated one opponent and are making progress in a war against the second, the third should start to look at the writing on the wall and make a decision: Am I friend of foe? Depending on personality and shared books the 3rd would either start to offer pact/alliance, or sour up and gear for war (seeking also to ally with the 2nd). If the player refuses the pact/alliance, the 3rd AI switches to foe strategy.

That would imply at about half a plane on "usual" settings? BUT it depends on number of AI wizards in the game? Perhaps "overextended" should also depend on whether towers are broken, so that it only looks at home plane when towers are sealed, otherwise both? And basically if you divide the land evenly between wizards (either on this plane or both), the wizard who controls more than 2 "shares" is overextended? This kind of logic should allow for an Arcanus wizards's alliance to gang up on a Myrran, too?

I guess I went too complicated...

[Seravy]

I seem to be always getting these when banishing the first AI. Is there a connection (score?)?

The warning you get for banishing a wizard uses the same text and ID as this, but it's a separate source.

Pacts increase your relation over time, so while you still get the penalty, it will not lower relation overall. Having a pact adds 1.5 REL each turn, so 18 over an entire year which is more than even the highest amount of penalty (which on lunatic is 4+6+6 = 16, so you're still gaining relation instead of losing any).
It might be harder to upgrade the pact to an alliance as penalties plus natural gravitation together might result in your REL stabilizing below alliance requirements, but you won't be losing relation for it. You usually need some tributes to fill the last 10-20 REL needed, with this in effect you might need 1 more tribute overall.

You might lose the pact however if two of these happen in a row before the AI rolls a "forget previous warning" which is a 1/7 chance a turn vs 1/12 for the warning, but only if they also rolled the "proceed with warning" roll both times, which can have a 100% chance of not proceeding if you kept the relation and treaty interest high. Assuming you raised REL to +50 through tributes (or the treaty itself) and didn't bother the AI with trade/treaty offers, even a maniacal wizard can have a 100% chance to never complain.

Basically if he gave you a first warning, but you raise REL by tributing in reaction, you should be safe 99% of the time.

Number of players, yes, that kinda complicates things. Not too much though, if the player at least had two opponents, the game will be about the same (except you don't kill 2 wizards first to reach having half of the plane) and if it was only one then we want a war to happen, having one opponent who is an ally the whole time is anything but interesting gameplay. So I think we can safely ignore the number of players.

The AI has no long term strategies, so that sort of thing would be difficult to implement. I guess I could make it so that wizard's pacts disable the overextension warnings altogether, but that kinda defeats the purpose of having it. If you don't have a pact, they will attack you no matter what (unless peaceful or lawful) through random aggression. So I guess that makes the purpose of this feature to break the pacts? It doesn't do that unless the pact is weak and/or you also do something else bad to them though...

[Nelphine]

Ok, I hadn't Mathed out the old so I see where you're coming from. I still think it should come earlier .

For me, the goal is to have the larger number of cities aim to break pacts that have nothing else supporting them. This should happen early enough that when the pact is broken, the ai isn't already so far behind that they lose.

Thus, I'd want that to happen around the time the human controls 3/5 of a plane. Assuming 40-45 cities (due to increased arcanus size), that means around 25 cities.  Since that's fair land size, that means 8 times as many cities as land size +1.

In turn, the base level of overextension should be happening after the first wizard is eliminated, and the human has more than their share of the remaining plane - more than 1/3. Since 4 times is 1/3, we go for the next lowest, which is 5 times.

The penalty at this stage shouldn't be high enough to break the pact.



How does difficulty play into this? I would say on advanced (or maybe one below advanced?), the large penalty shouldn't be enough to break a pact. On lunatic, the base version should be enough to break the pact. Master should not quite Break the pact on base version. Expert should break pact on the large version.

[Seravy]

I realized we are doing this wrong.
Instead of treating this feature as an existing thing that needs some numerical adjustments, we should first define the intended role for it.

I see a few possibilities for that...

1. Prevent the AI from letting the human player outgrown them and win without ever having to fight.
First of all, we shouldn't want this to happen all the time. The game should be playable through a peaceful strategy, with focus on diplomacy and economy. Meaning it should only really be inevitable if the player isn't playing for that strategy and only with some of the enemy wizards, not everyone.

However we already have a feature that does exactly this : The generic war declaration that says if the player has near the AI's military strength and relation isn't good, declare war.

So I think we can conclude this is NOT one of the goals.

2. Ensure the last one or two AIs declare war on the human before the human wins the game through getting an unstoppable strategy.


For this, city count is not really a relevant measure. A player with only 10-15 cities can still summon a stack of 9 regenerating Colossus and win the game. Especially if the node power is high.

The conclusion here is, this is one of our goals, but this feature isn't the way to achieve it.

3. To punish a player for overextending and holding more cities than what they can defend.
The player needs to be able to do that if we want the game to be playable with a peaceful strategy - the only way to keep up with the AI resource advantage is to grab more land early than anyone else. So the penalty should be in a form that diplomacy and tools of peaceful games can counter - specifically Aura of Majesty and Charismatic. A recurring warning with a REL loss penalty is exactly that sort of thing, so we are good, and I believe we can agree that real overextending should not be ignored by the AI.

So this is the true goal for this feature.

So now that we see the goals, I'll try to design a solution.

[Seravy]

1. We have nothing to do here, as we concluded we don't want this.

2. I gave more thoughts about this and we really don't want this either. If the AI jumps the gun and picks a fight with a human when they could be racing to SoM or could be conjuring up an unstoppable very rare army, that's bad for the game. We'd need this to be elaborate enough to have the last wizard know whether they are in the position to beat the human immediately, or if time actually benefits them more than the human. Chances are, the AI will cast something nasty that will make the human want to start the war himself.
Also, peace rewards the experienced player, who can grow faster or plan better than the AI. Forcing a potentially unwinnable war on the human as soon as possible doesn't improve the game.

What made my game yesterday worth playing was that I was able to find a way to win, using wraith form wyverns and whatnot. It was still very hard to win, but this is what makes the difference between a player who knows how to win, and one that does not and loses. Had the AI declared war, I would have lost the same way as someone who is a complete beginner.

3. For this, we most likely should use a formula involving both turn count and land size to determine how many cities counts as "overextension".
10 cities in 1403 should, 20 cities in 1420 should not count overextended on Fair land size for example.