New Stability Rules

[Leoreth]

As mentioned in the recent update, I have revised the stability rules. Initially I only wanted to adjust some factors, but over the course I decided to make more extensive changes. The resulting system is both simpler and should provide better results

Rule changes:
- I have completely removed crises, so far without replacement.
- If you are on collapsing and your stability has not improved since the last check, you collapse. The AI gets a collapse to core instead if possible.
- Stability required for stability levels is now constant for all players in all situations (less than -10 is collapsing, less than 0 is unstable, less than 10 is shaky, less than 20 is stable, more than 20 is solid). Only AI civs beyond their historical fall date have a different threshold (10 higher for all levels).
- On each stability check, if your stability is higher or lower than the boundary of your current level, you lose or gain a level. You can only change by one in either direction per check. AIs beyond their historical fall date can lose multiple levels per check.
- Stability factors that depend on long term trends (economy, happiness) reset when collapsing to core or being vassalised to stabilise civs after a setback.

Changes to stability factors
- Core population multiplier grows only half as much per era (i.e. more limited empire sizes in the late game)
- Added temporary positive expansion stability for recently founded / conquered cities in historical territory (higher for conquered cities when running Conquest)
- Increased the effect of the Portuguese UP
- Economic growth of less than 2% now also counts as recession
- Only requires 0% growth to avoid recession with Central Planning
- Increased (positive and negative) economy stability with Free Enterprise
- Reduced negative economy stability with Public Welfare
- Happiness stability was too all or nothing (often either -10 or +10), now it is bounded by -5 and +5
- Religious unity is now weighted by the population of cities, instead of simply counting them
- Minor religions do not contribute to religious disunity
- Religious disunity impact of non-state religions reduced when state religion is also present
- Positive stability from religious unity halved when running Tolerance
- Completely removed negative stability from unstable neighbours
- Completely removed stability from open borders / instability from number of contacts (penalised the AI too much, while easy to achieve for the player)
- Added new relations stability factor that is based on temporary AI memory of your actions (refusing/agreeing to their proposals, gifts etc.)
- Vassals receive additional stability when their masters are more stable than them
- War success stability is less sensitive to minor events that happened in the past
- Slightly increased war weariness stability

The only thing I am hesitant about is the stricter limits for expansion stability. It is now easier to get overexpansion, but this is counterbalanced by the fact that many other (often unpredictable) sources for negative stability have been removed or softened. Therefore it should be possible to balance these negatives with positives in other areas. But I am open to feedback.

 

[Fresol]

- Economic growth of less than 2% now also counts as recession

This spells disaster.

 

[Tigranes]

- Economic growth of less than 2% now also counts as recession

This spells disaster.

Totally agree. We all have to simply agree that game does not have a good way to reflect the true economic nature of the civs controlled by AI in the scripted environment of RFC-style rising and collapsing of civs. We need penalties in some very simple cases only, as when bleeding gold, starving cities, or more simple checking. Like penalize civ only if GDP ratio to population (GDP per capita) kept decreasing for last 3 checks.

And we absolutely need to inforse the "collapse to core" rule. Just now rolled Canada with 600 AD start. Sure enough: England, Vikings, China and Dutch are all collapsed. Without England and Netherlands Australia will stay as a wasteland, nobody even cared to pop the good huts there. Civ like England should disappear from the game only if her 1-2 core cities are captured.

 

[Fresol]

Two observations on the current stability system:
1. Harsh penalty for economic stagnation. -10 penalty is common, and especially wrongfully inevitable for early civs like Egypt, China and Babylonia who have to train workers and improve tiles from scratch.
2. Religious disunity penalty can be broken. Say Orthodoxy spreads to 2 pop Addis Ababa and Ethiopia converts to Orthodoxy. But it never spreads to Aksum, and it takes forever to build a monastery and train a missionary, so the 8 pop Aksum remains pagan. Now there'll be a -10 religious disunity penalty which cannot be removed quickly, which leads to Addis Ababa ceding, and the religious disunity penalty only gets worse. At the end, Ethiopia collapses. This is what happened in a game I just ran.

 

[Leoreth]

- Economic growth of less than 2% now also counts as recession

This spells disaster.

Totally agree. We all have to simply agree that game does not have a good way to reflect the true economic nature of the civs controlled by AI in the scripted environment of RFC-style rising and collapsing of civs. We need penalties in some very simple cases only, as when bleeding gold, starving cities, or more simple checking. Like penalize civ only if GDP ratio to population (GDP per capita) kept decreasing for last 3 checks.

Two observations on the current stability system:
1. Harsh penalty for economic stagnation. -10 penalty is common, and especially wrongfully inevitable for early civs like Egypt, China and Babylonia who have to train workers and improve tiles from scratch.

Something to keep in mind here is that one check (i.e. one three turn interval) does not automatically cause bad stability. Instead the code determines a "trend" of the last 10 checks. Checks with negative growth can be countered by checks with positive growth, or even checks with neutral growth if there are enough. 3 negative + 7 neutral checks for example results in 0 stability.

The reasoning behind this is that I rarely encountered negative growth in the game, and I wanted to make economic stability a more hard to balance category. The way the game works is that things are mostly increasing, especially for the player. My test games confirm that this is not a damning change, even for the AI, but concrete counterexamples are welcome. I admit that I did not review the effects of this change on the early game, I'll give that aspect another look.

And we absolutely need to inforse the "collapse to core" rule. Just now rolled Canada with 600 AD start. Sure enough: England, Vikings, China and Dutch are all collapsed. Without England and Netherlands Australia will stay as a wasteland, nobody even cared to pop the good huts there. Civ like England should disappear from the game only if her 1-2 core cities are captured.

The rule is enforced in so far as that collapse to core always happens instead of a full collapse as long as they have at least one city outside their core (and a collapse to core triggers the reset of many long term stability effects), so unless their stability stays bad and they face another collapse immediately afterwards, they will stay alive. But imo if the bad stability is the result of internal factors instead of overexpansion, they should collapse. The only exception is that civs beyond their fall date will always completely collapse, but that does not apply to civs like England and the Netherlands.

There are two points where I still want to address this problem in other ways:
- civs like England still suffer disproportionally from domestic instability mostly because they run the wrong civics. For example, it's usually possible to improve their stability by a lot by switching out of Redistribution and into Tolerance. I plan to improve the AI in that regard to make them prefer more appropriate civics for the time period. This should also help with the immersion problem of ahistorical civics still being run.
- in my opinion it is fine for civs to collapse, as long as they come back quickly. Currently that is not guaranteed because respawns are random and can affect any civ that is capable of respawning. Instead I would like to give some civs higher priority to respawn than others, depending on how long they have been dead and who they are.

2. Religious disunity penalty can be broken. Say Orthodoxy spreads to 2 pop Addis Ababa and Ethiopia converts to Orthodoxy. But it never spreads to Aksum, and it takes forever to build a monastery and train a missionary, so the 8 pop Aksum remains pagan. Now there'll be a -10 religious disunity penalty which cannot be removed quickly, which leads to Addis Ababa ceding, and the religious disunity penalty only gets worse. At the end, Ethiopia collapses. This is what happened in a game I just ran.

That's a good point, I'll see what I can do to prevent this.

 

[calad]

Some feedback before latest Git update

 

[calad]

Feedback: I dont know what you did but started as Romans and Chinese, Babylonians and Egypt had all collapsed. China had only two cities, Egypt one as well Babylonia.

Edit: run it twice and looks like all ancient civs just collapse on their starting position. Strongly not recommended to play this revision.

 

[Leoreth]

Can you upload that Rome save?

I didn't think the ancient era would be so heavily impacted, I will review that more this evening. But post 600 AD starts should be playable without issues.

 

[calad]

Sure, here.

Regent/normal.

 

[Leoreth]

Thanks!

 

[Jarlaxe Baenre]

one check (i.e. one three turn interval)

Is that the same for all game speeds? It's harder to maintain constant growth from turn to turn on Marathon than on Normal.

 

[Ogi123]

Run two games as Babylon, collapse on turn 21-23. Reason -18 economy penalty for recession, training worker takes 15 turns from pop 1...

 

[calad]

Oh, and also got -6 recession penalty on Rome game. I think lack of workers, buildings and civs are major issues on early game.

 

[Leoreth]

Is that the same for all game speeds? It's harder to maintain constant growth from turn to turn on Marathon than on Normal.

Yes, it's three turns on normal and scales appropriately with game speed.

 

[Genghis Khaiser]

I played as Moors in the 3000 BC scenario. Everything was going great until I collapsed right in 1200 AD. I also noticed that both Rome and Byzantium were dead when I spawned.

Most of my instability was because of overexpansion, even if several times before it wasn't that much of a problem even if I had ran larger empires.

 

[Tigranes]

I rolled 3000 BC Canada start. No England, no Dutch, no Vikings, no China again. Germany captured Amsterdam, so at least Dutch collapse was natural. But English collapsing without loosing a single core city is very irritating. As well as not crushing few Celtic troops in Ireland. Maybe we should help AI England and pass Ireland to British control after some civic change? Then Dublin will be that magic "one outside of the core city" which will prevent British from collapsing completely.

America and Russia had very limited expansions on respective continent. Number 1 civ was India! Tech spread was more or less satisfactory. So our main problem remains stability and rules for collapsing. I suggest - lets turn off economic stability completely for a moment and see how everything else goes. Than we can iron out economic stability rules!

 

[calad]

Or atleast prevent completely collapse due to bad economy. It anyways does not make sense on modern era.

 

[Ogi123]

For a moment I would prefer return of previous economic stability rules, because there should be something here instead of completely turning off single section.

 

[Leoreth]

New update:
- negative growth threshold scales with era
- increase base core population modifier
- display stability level change on next check in tooltip
- make the AI avoid adopting Judaism as state religion

The first means that the threshold is 0 again for the ancient and classical eras, avoiding the issue of economic instability when you only have one city and little opportunity to grow.

The second should help Rome and other early empires, in combination with the last which often hurt Rome with religious disunity because they loved adopting Judaism after conquering Jerusalem. Some of my Maya games had almost a completely historical Roman Empire and they were break even thanks to recent conquests combined with the Conquest civic. If the temporary stability runs out or they get more barbarian pressure, they should head towards collapse, which is pretty historical. The human player can obviously do better in terms of economy and managing expansion.

 

[Tigranes]

Ok things look better with this latest update. England is back in the game! But Russia feels too much barbarian pressure from Siberia. There were never any Siberian Cuirassier in real life.