(2-02b) Counterproposal: Increase coup success rate based on number of rigged elections (with other changes on formula)

[azum4roll]

Coups are very powerful, but completely rely on RNG. Losing a coup also sends your influence to rock bottom and kills your spy.

This proposal would add a way to guarantee successful coups, at the cost of spy turns.

Current formula of base success rate B = (50 - D / 25)%
Current formula of final success rate R = B * (1 + 0.1 * (L - LE + IB)) * (1 + PM)

Proposed formula of base success rate B = (50 - D / (25 - (L - LE) * 5) + N * 20)%
Proposed formula of final success rate R = 1 - (1 - B) ^ ((1 + 0.2 * IB) * (1 + PM))

where D = influence difference from current ally
L = 0, 1, 2 for Recruit, Agent and Special Agent respectively
LE = Rank of Enemy Spy, if there is one in the CS
IB = Rank bonus from Cultural Influence: 1 if the player's culture is popular over the current ally of the CS, 2 if influential, 3 if dominant
PM = Modifier from Policies (currently only used for "covert action" with a value of 1)
N = number of successful rigged elections done by any of your spies in this CS after the last successful coup

N is limited to "after the last successful coup" in order to
1. prevent easy counter-coup 20 turns after a successful one;
2. create a dilemma where you both want to coup later for higher success chance and earlier than your rival does.

Spy rank difference is moved to the formula for B to make spy rank always helpful in coups (otherwise it doesn't matter if B is 100%).

Rationale of changing the formula for R, as mentioned in comments:

However, for a one-time hit-or-miss chance, direct multiplication is a terrible way to modify the success rate. Instead we should make a 2x modifier equate the success rate of performing two coups in a row with the base rate. The resultant success rate would then always be bounded in the (0,1) range without any explicit min/maxing.

Now it's possible to have a 100% success rate of coup if your spy stays for long enough.

EDIT 1: Changed the proposed formula and title after getting the full current formula from @axatin

 

[Stalker0]

Coups are very powerful, but completely rely on RNG. Losing a coup also sends your influence to rock bottom and kills your spy. There's also no incentive to perform coups with a high level spy.

This proposal would add a way to guarantee successful coups, at the cost of spy turns. Shorter time is required for a higher level spy.

Current formula of success rate = (50 - D / 25)%
Proposed formula = (50 - D / (25 - L* 5) + N * 20 + L * 10)%

where D = influence difference from current ally, L = 0, 1, 2 for Recruit, Agent and Special Agent respectively, and N = number of successful rigged elections done by any of your spies in this CS after the last successful coup.

N is limited to "after the last successful coup" in order to
1. prevent easy counter-coup 20 turns after a successful one;
2. create a dilemma where you both want to coup later for higher success chance and earlier than your rival does.

Just want to check an example to confirm all the () are in the right place.

Lets go with a level 2 spy (L = 1) after 2 successful rigs (N = 2), and with an influence difference of 100 (D = 100)

50 - 100/(25-5*1) + 20 * 2 + 10 * 1 =
50 - 100/20 + 40 + 10 =
50 - 5 + 40 + 10 =
95% chance to coup

Is that right?

 

[AndreyK]

I don't understand. Where are turns (duration)?

 

[azum4roll]

Just want to check an example to confirm all the () are in the right place.

Lets go with a level 2 spy (L = 1) after 2 successful rigs (N = 2), and with an influence difference of 100 (D = 100)

50 - 100/(25-5*1) + 20 * 2 + 10 * 1 =
50 - 100/20 + 40 + 10 =
50 - 5 + 40 + 10 =
95% chance to coup

Is that right?

Yes. It's unlikely the influence difference is so low though. You probably should've just sent a diplomat instead.

 

[Stalker0]

Yes. It's unlikely the influence difference is so low though. You probably should've just sent a diplomat instead.

Actually for early coups this is very common. Person X gets ~100ish influence through some quests, Person Y is the newcomer who has no influence with the civ and wants to coup it to take it. Happens all the time in my games currently.

 

[nekokon]

I think we shouldn't include rigging election into coup formula.

Usually when you try to coup a CS it's mainly because you can't compete influence with the AIs for the long term, thus resolving to more risky option. Rigging election, on the other hand, is the more stable version which you would be using along with diplomat to win the influence race. Having both together is pretty unintuitive, as usually you can already win the influence game after rigging election and diplomat (which would be the requirement for current proposal) thus there's no need to risk a coup (or might not even possible to coup since you're their ally already) with how high your influence already is.

Coup should remain a risky move used when you can't catch up with influence in time for a specific purpose (right before congress vote or dow), thus their chance should depend on something else. Ideally I would prefer a system where coup chance is increased by the amount of resource you invested in, and decrease by the amount of influence of the current ally, but it will require new code.

 

[Solic]

I like making higher level agents having more of a chance to coup and I'm surprised it doesn't work like that already. I guess they only settle into a CS faster. I disagree with making successive election rigs matter so explicitly. It's an opaque mechanic that is already represented in the influence differential and it doesn't need this additional factor.

 

[Stalker0]

I would prefer a system where coup chance is increased by the amount of resource you invested in

Technically that is what this proposal does, the resource is spy time....which means you are not stealing yields elsewhere. Effectively the proposal is saying "the more spy time you are willing to invest, the higher your coup chance is"

 

[nekokon]

Because of the nature of coup (risky move to acquire CS in a short time), you can't simply invest more time when it's needed (else spamming diplomat or rigging election would be safer and more guaranteed)

 

[Stalker0]

Because of the nature of coup (risky move to acquire CS in a short time), you can't simply invest more time when it's needed (else spamming diplomat or rigging election would be safer and more guaranteed)

I'll disagree with that. Coups serve three primary functions:

• Flipping a CS that is peace blocked with you, its the only mechanic that can stop this.
• Flipping a CS where distance makes diplomat influence impractical. Can happen with narrow rift maps where you can see CS but can't get diplomats over until astronomy. Or Cs where warring civs makes crossing the distance too dangerous.
• Gaining a massive influence swing, where the risk of a spy loss is worth the sheer amount of hammers needed to flip it traditionally.

Investing more time for your spy doesn't stop any of these functions from working imo.

 

[nekokon]

Problem is if you decide to invest more time (by leaving the spy there to rig election) then you're better flipping the CS normally (given the success chance at 50% when you have the same amount of influence with their ally, which means roughly less than 30% if you're gonna coup with a significant gap in influence). Unless you want to try your luck with sub 30% chance (currently calculated based on influence difference) and 70% chance of having all of your current influence down the drain (thus also wipe out chance to coup next oppotunity) even just rigging is the better choice.

Having coup chance modified through other means than influence would solve this issue instantly.

 

[Stalker0]

Problem is if you decide to invest more time (by leaving the spy there to rig election) then you're better flipping the CS normally (given the success chance at 50% when you have the same amount of influence with their ally, which means roughly less than 30% if you're gonna coup with a significant gap in influence). Unless you want to try your luck with sub 30% chance (currently calculated based on influence difference) and 70% chance of having all of your current influence down the drain (thus also wipe out chance to coup next oppotunity) even just rigging is the better choice.

Having coup chance modified through other means than influence would solve this issue instantly.

I feel that is the point. I mean in general influence units should be the primary way to flip CS, coups should be a speciality tool need for certain situations.

So the idea that influence would be better than coups…well yeah that’s as it should be

 

[nekokon]

We're trying to establish influence as the common slowly grindy method and coup as the high risk high reward instant decision, making coup much worse except for unavoidable situations (and very small chance to success in those situations, based on the calculation from influence) would just discourage ppl from using it entirely. Using another resource aside from influence would make it much more balanced and not completely op or borderline useless like this, and the boosted coup chance (to make it worth trying) is paid by the high cost, so that coup would remain the inferior choice in most situations, but the better choice in some situations when it's needed.

 

[axatin]

Current formula of success rate = (50 - D / 25)%

This is only the formula for the base success rate. For the calculation of the success rate of a coup, modifiers for spy levels and policy modifiers are applied already in the current version. Sorry for the confusion, I should have made this more clear in the other thread when I mentioned the formula.

The complete formula is the following:

Base success rate = (50 - D / 25)% (but at least 5%)
Total success rate = (Base success rate) * (1 + 0.1 * (L - LE + IB)) * (1 + PM)

with L = Spy Rank (0, 1, 2 for Recruit, Agent and Special Agent respectively, as explained above)
LE = Rank of Enemy Spy, if there is one in the CS
IB = Rank bonus from Cultural Influence: =1 if the player's culture is popular over the current ally of the CS, =2 if influential, =3 if dominant
PM = Modifier from Policies (currently only used for "covert action" with a value of 1)

The total success rate can't be lower than 5% and can't be higher than 99%.

 

[azum4roll]

Thanks for the real formula. I'll clarify it in the OP.

However, for a one-time hit-or-miss chance, direct multiplication is a terrible way to modify the success rate. Instead we should make a 2x modifier equate the success rate of performing two coups in a row with the base rate. The resultant success rate would then always be bounded in the (0,1) range without any explicit min/maxing.

New formula in OP.

 

[Recursive]

Proposal sponsored by axatin.