[BTS] SGOTM 25 - Home Slices

[WastinTime]

Looks very much to me like the halving of MasterPower if no land target happens only when not at war (= peace vassal)

correct. That's only peace vassals.

This must be iVassalPowerModifier, which for Roosevelt is 0. I was looking at this a lot, and reading capitulation code, but couldn't figure out what this 0 means in practice. Where does 1.4 come from?

    case TRADE_SURRENDER:
        return GET_TEAM(getTeam()).AI_surrenderTrade(GET_PLAYER(eWhoTo).getTeam(), 140);
        break;

140 is hard coded as the default and modified by iVassalPowerModifier. So Church and Roos are 1.4 since iVassalPowerModifier=0.
Mansa, for example is -20. So he's 1.2.

 

[WastinTime]

So requirement for Sitting Bull is that average power needs to be at least twice his power?

SB is 50, so that 1.9x

 

[Kaitzilla]

And another related question (quoting from our thread):
This must be iVassalPowerModifier, which for Roosevelt is 0. I was looking at this a lot, and reading capitulation code, but couldn't figure out what this 0 means in practice. Where does 1.4 come from? I noticed this modifier corresponds to "Resists Capitulation" in Know Your Enemy, all Ai with 0 in XML says 4/10 in KYE.Should I look at the number in the latter and add that much to their average power demand? So requirement for Sitting Bull is that average power needs to be at least twice his power?

We dug into this issue a while back starting here
Then again later here

I believe WT discovered the 1.4 thing from reading code.

Historically, it also popped up here in great detail.

Ok - let's crunch some digits

Current powers:
You: 1,088,000 ( = "MasterPower" for negotiations)
Mehmed: you/1.9 = ca. 573,000
Hatty: Rival Worst = 211,000
Rival Average: 552,142
Total Average: "AveragePower" = (7*RivalAverage+You)/8 = 619,124
Mehmed + Hatty: "Power_MehmedHatty" = ca. 784,000
Mehmed's VassalPowerModifier (xml): 0
Mehmed calculates his power for negotiations: "VassalPower" = (Power_Mehmed+Power_Hatty)*(140+VassalPowerModifier)/100 = 1,097,600 (the default power multiplier 140 is hard coded!)


Check 1 - Mehmed vs Invader (=you):

+ Adjustment of MasterPower for war success (WS) ratio: MasterPower = MasterPower * [0 ... 2]
An example: You conquer 5 cities (WS = 50) and kill 50 units (WS = 200) without losing a single unit (WS is at least 10 for everybody!)
WS_Master = 250; WS_Vassal = 10 --> MasterPower = MasterPower * 2 * WS_Master / (WS_Master + WS_Vassal) = MasterPower * 500 / 260 = MasterPower * 1.92
{If a 3rd player has a higher WS vs the target than you then the target gets the difference: WS_Vassal = WS_Vassal + (WS_3rdPlayer - WS_Master)}

+ Conditional adjustment of MasterPower for worst enemy: MasterPower = MasterPower * 3/4

Power check: VassalPower > 2/3 * MasterPower ? --> "We are doing fine on our own!"

Conclusion: If you redeclare on Mehmed you will be his Worst Enemy again (right now it's Augustus). Without consideration of any war successes the check goes:
VassalPower = 1,097,600 vs. 2/3 * 3/4 * MasterPower = 1/2 * MasterPower = 544,000 --> DENIAL!
But with a bit of war success decreasing Mehmed's and increasing your power you should not have any problem to pass this test.


Check 2 - Mehmed vs Average:

+ Conditional adjustment of AveragePower for Mehmed being a LandTarget for other players (= additional threats "raise the average"; LandTarget = at least 1 city in other player's primary area + at least 8 landplots adjacent to other player's land territory anywhere on the map):

- If Mehmed is a valid LandTarget for a 3rd player who's power is greater than Mehmed's power (vassal power included for both):
AveragePower = AveragePower * 2 * Power_3rdPlayer / (Power_3rdPlayer + Power_MehmedHatty)
- If Mehmed is a valid LandTarget for a 3rd player who is your ally in your war vs. Mehmed:
AveragePower = AveragePower * (MasterPower + Power_MehmedHatty) / Power_MehmedHatty
​

Power check: VassalPower > AveragePower ? --> "We are doing fine on our own!" (This is where they made capitulation easier in 3.17, check was VassalPower > 3/4 * AveragePower in 3.13.)

Conclusion: Currently Mehmed is a valid LandTarget for Saladin and Darius, but both are weaker than MehmedHatty so the AveragePower would not change unless they join you in your war vs. Mehmed.
VassalPower = 1,097,600 vs. AveragePower = 619,124 --> DENIAL!

• Killing Hatty would increase the AveragePower to 677,427 and decrease Mehmed's VassalPower to 816,200 (remember the hard coded 140!) --> still not enough to pass the test.
• Reducing Mehmed to Hatty's anemic power level but letting Hatty live would change AveragePower to 573,874 and Mehmed's VassalPower to 590,800 --> still no capitulation.
• But the LandTarget factor is quite significant. Say you ask Darius to join your war, this will increase AveragePower to 1,478,316!!! --> capitulation.
• Alternatively you could gift Konya to Augustus and thus turn Mehmed into a valid LandTarget for him. Since Augustus is stronger than MehmedHatty this would increase the AveragePower even without a war between them but only marginally to 639,338. Thus having a suitable war ally is the way to go here. Also note, that the power of this war ally is irrelevant, so a very weak ally should be optimal in order to avoid capitulation to the wrong player.

All in all I got to agree with TMIT in that the way these mechanics are coded is very intuitive and well documented.


Your game:
I think you don't need to fear any German AP cheese right now, because as you mentioned, Saladin lacks a Confu city and he is running his favorite civic Theocracy - so no spread possible (the AI is not familiar with the Missionary gifting exploit).
Ze Germans hev zeir hends full, ja!

Edit: just checked again and found that Mehmed is not a valid LandTarget for Saladin due to lacking 1 adjacent land plot.

The key point is this line:

Mehmed calculates his power for negotiations: "VassalPower" = (Power_Mehmed+Power_Hatty)*(140+VassalPowerModifier)/100 = 1,097,600 (the default power multiplier 140 is hard coded!)

Since none of the AI we were dealing with had vassals , all of them calculated their power when considering the World Average Power as Power = (Power_AI)*(140+iVassalPowerModifier)/100
So if England had 160,000 Soldiers, they calculated their power as 160,000*(140+0)/100 = 224,000 Soldiers and would capitulate if this number was lower than World Average Power.
If De Gaulle had 160,000 Soldiers, he calculated his power as 160,000*(140-20)/100 = 192,000 Soldiers
If Sitting Bull had 160,000 Soldiers, he calculated his power as 160,000*(140+50)/100 = 304,000 Soldiers

Since Average World Power is so hard for the human player to raise, this makes civs like Sitting Bull and Ghenghis Khan quite hard to capitulate in practice.
Sitting Bull also takes the longest before willing to talk peace iRefuseToTalkWarThreshold=12.
He should be 11/10 for Resists Capitulation.

We also found out that if 1 AI was vastly more powerful than the human player, the 1.4 goes out the window.

Wait, something still feels wrong.
What if in the 2 AI case, Roosevelt became all powerful?
10000 nukes, infinite Soldiers.
No matter what, both the Human and Churchill are far below the World's Average Power.
How low can the Human power go and still be able to cap Churchill?

Human 217,000 Soldiers / Churchill 140,000 Soldiers / Roosevelt 2,205,000 Soldiers
World Average Power = 854k Soldiers
Churchill will Capitulate.

Human 215,000 Soldiers / Churchill 140,000 Soldiers / Roosevelt 2,205,000 Soldiers
World Average Power = 853.33k Soldiers
Churchill won't Capitulate.

This whole business is littered with damn mysteries.
Did anyone figure out Freedom Appreciation?

 

[elitetroops]

Thank you very much for clarifying those!

Btw. Did I come to the correct conclusion that the details about the mechanics and execution of your super secret EP liberation trick were not discussed in this thread? Tried to find the plan laid out, but couldn't.

 

[Kaitzilla]

Thank you very much for clarifying those!

Btw. Did I come to the correct conclusion that the details about the mechanics and execution of your super secret EP liberation trick were not discussed in this thread? Tried to find the plan laid out, but couldn't.

It was all hashed out, schemed, and plotted in a private discussion.
We are currently trying to have it merged into this thread, but we might just have 1 long post quoting everything.

 

[elitetroops]

We also found out that if 1 AI was vastly more powerful than the human player, this whole 1.4 business goes out the window.
This whole business is littered with damn mysteries.
Did anyone figure out Freedom Appreciation?

Freedom Appreciation is how many +diplo they have towards master when liberated as colony.

Regarding the vastly more powerful AI, what I found in my tests was that you still always had to have at least 1.5x their power. You are pretty close in your test:

Human 217,000 Soldiers / Churchill 140,000 Soldiers / Roosevelt 2,205,000 Soldiers
World Average Power = 854k Soldiers
Churchill will Capitulate.

Human 215,000 Soldiers / Churchill 140,000 Soldiers / Roosevelt 2,205,000 Soldiers
World Average Power = 853.33k Soldiers
Churchill won't Capitulate.

But for some reason it seems you need slightly more than 1.5x their power.

We are currently trying to have it merged into this thread, but we might just have 1 long post quoting everything.

Thanks, this would be much appreciated! The great thing with SGOTMs is that the threads leave behind such a wealth of information and strategies to explore. Would be a shame if some of it was tucked away in a private conversation forever.

 

[LowtherCastle]

We teched HBR T52... Once we decided to win in 20 turns, it was the first thing we did. Early on it was all Swords with some Chariots, then Chokos added to the mix.

I was referring to settling your two GPs. That's about as early as you can accelerate.

 

[WastinTime]

Would be a shame if some of it was tucked away in a private conversation forever.

I found a XenForo addon that might be able to export it for us. It's my fault for going private. If we decided not to use the tactic, I want to protect the secret a while longer.

Culture: I don't see the need to use 80 spies and get the culture all in one turn. 40 spies could each do 2 missions (over 6 turns.)
And I may have a solution to get 150,000 culture with less than 10,000 EPs!

You going to reveal your trick before or after the game is over...

Problem with saying it now is that a lurker can read it and go off on a espy-culture binge in the HoF or elsewhere--which may prompt a public discussion of how such early dates were being posted. Other teams could get wind of our trickery.

 

[Fippy]

Espionage and AP are 2 added expansion features, and imo not as well balanced and thought out as the base game.
Proof being delivered often in sgotm or HoF games, or even just regular ones.

I always was happy with sgotm, much credit to for example Alanh for making this competition very professional.
But this time will be as good as any other, for saying..certain aspects are really questionable for professional and competitive play.

Imo, and really just imo, this stops us from getting more players and an overall bigger "tournament".
There are so many peoples who are not interested in features that overthrow basic requirements of an excellent Civ4 game, maybe the best game ever created (certainly for us ).

Exploit? Wrong word imo, it's usually standing for negative actions and abusing games.
I would word this differently..we are not excluding some parts of this game that should not appear in tourney play.
I follow multiplayer threads for fun reading, and those guys simply exclude stuff like EP missions, they know it's ment for casual players just enjoying game features, but was never balanced enuf for competitive play.
i would encourage everybody involved in sgotm to think about that.

 

[elitetroops]

Regarding the vastly more powerful AI, what I found in my tests was that you still always had to have at least 1.5x their power.

Forget about that. Here I can cap Roosevelt with only 1.2x his power:

Spoiler :

Average power exactly 1.4x Roosevelt's power.

That's weird, because in a previous test with wimpy De Gaulle I always needed 1.5x his power.

This subject would probably deserve it's own thread in S&T to figure out the exact mechanics.

Edit: Solved

Next I managed to cap him with 1.09x his power. Here is the relevant part of the code:

       int iMasterPower = GET_TEAM(eTeam).getPower(false);
       int iVassalPower = (getPower(true) * (iPowerMultiplier + iPersonalityModifier / std::max(1, iMembers))) / 100;

.....

           iMasterPower = (2 * iMasterPower * iTheirSuccess) / (iTheirSuccess + iOurSuccess);

           if (AI_getWorstEnemy() == eTeam)
           {
               iMasterPower *= 3;
               iMasterPower /= 4;
           }

.....

       if (iVassalPower > iAveragePower || 3 * iVassalPower > 2 * iMasterPower)
       {
           return DENIAL_POWER_US;
       }

Apart from the average power check, there's also a check that MasterPower is high enough. Master powered is influenced by war success and it gets quite a big hit if you are worst enemy. So if you are weak in relation to the subject you try to capitulate, your best bet is to not be their worst enemy. War success also has an effect, but it is very minor.

VassalPower is multiplied by iPowerMultiplier and iPersonalityModifier, whatever those are. I think the latter is related to iVassalPowerModifier.

 

[Gumbolt]

I think by time of capping we had declared on most civs at least 1-2 times. Apart from Romans and Ottoman. Near end we were pretty much worst enemy with most AI.

Congrats on the gold. Didn't upset the party this time but we stole the wooden spoon instead.

 

[WastinTime]

iPowerMultiplier and iPersonalityModifier, whatever those are. I think the latter is related to iVassalPowerModifier.

iPowerMultiplier is the hardcoded 140.
and yea, iPersonalityModifier is iVassalPowerModifier.

 

[Gumbolt]

Have to say Elite's knowledge of game functions has been amazing this game. Always seemed to be able to figure out problems using knowledge of the game. When your solving stuff that Home Slices didn't solve you are doing well. We do need someone who actually reads the forums more for new tricks and old tricks that are around. Least we keep your team honest on dates.

 

[Noble Zarkon]

SGOTM 26 - Lady Stoneheart, Queen of the Ice Age, has been announced.

 

[babybluepants]

That was quick!
Sounds like a fun variant.

 

[Mitchum]

I promise to double my participation in this next one!! If you'll have me, of course.

 

[LowtherCastle]

I promise to double my participation in this next one!! If you'll have me, of course.

From the looks of it, we'll all have to quintuple our participation. Wonder what the screenshot will show.

 

[Mitchum]

From the looks of it, we'll all have to quintuple our participation. Wonder what the screenshot will show.

Deal. I'll quintuple my participation then.

This does sound like quite a challenge and it will go deep into the game. Sounds interesting too.

 

[WastinTime]

Whew, that is a lot of tech in SG26. I guess it's been a while since we had to research much. Although... I may have a way to get some of those buildings without the required tech

 

[LowtherCastle]

Whew, that is a lot of tech in SG26. I guess it's been a while since we had to research much. Although... I may have a way to get some of those buildings without the required tech

shhh...we don't want to scare Fippy and elite off.

Let me guess...ice sculptures?