Challenge-XVII-10

WastinTime said:

P.S. in my brief test game the AIs had only 1 defender for a while. Probably not specific to Roos.

Click to expand...

Sounds to me like your most important strategy right now is to capture both capitals asap.

LowtherCastle said:

Sounds to me like your most important strategy right now is to capture both capitals asap.

Click to expand...

Yea, I think so. This game is so much fun. So many choices.

marines+tank can capture on DoW+1, but...

2 tanks could capture on DoW+0 if they are not blocked with a hill or forest. Roos is blocked in my game...unless they chop which is doubtful.

My capital ideally should be able to do:
T0 anarchy, then : worker, worker, rax, factory, coalplant, tank, tank.
So that's 2 tanks moving on T7

Or skip a worker and do T6...or skip the barracks...or both.

analysis paralysis. I better take a break.

WastinTime said:

marines+tank can capture on DoW+1, but...

2 tanks could capture on DoW+0 if they are not blocked with a hill or forest. Roos is blocked in my game...

Click to expand...

Assuming you can verify what he has in his capital, sounds like a reason to attack Roos first, since you only have one tank, you want the marines on T+1 as a back-up anyway. Then go for Stalin on T+0 for 2 tanks.

One idea anyway.

WastinTime said:

My capital ideally should be able to do:
T0 anarchy, then : worker, worker, rax, factory, coalplant, tank, tank.
So that's 2 tanks moving on T7

Or skip a worker and do T6...or skip the barracks...or both.

Click to expand...

Those are definitely critical questions for the Gnejs Principle. It's really just a question of what can work. You only need it to work once. My basic idea on how to win faster is the opposite of the game I played -- start surrounded by AIs. How fast can you rush-capture 2 capitals? 3 capitals? 4 capitals? This is counterbalanced by the benefits of a settler instantly creating an operational city. Whatever the optimum is, that's probably the winner.

Another idea is to be centered amongst the AIs rather than at the far end. Maybe there's a way to combine capturing 2 capitals with building 2 settlers a bit later on, expanding in all four cardinal directions. This partly depends on the mechanics of vassalization, if it helps or not.

Ha! We start with 3 settlers. So much for building one.

Frkitall. Guess I was really lucky last time. Get a good map, cherish it. I must have regened 200 times and couldn't get squat. Next year...maybe.

With micro pangaea and some abuse of (peace?) vassal mechanics, finish dates should be much faster than in challenge 1.

Speaking of vassals. I play 90% of my games without vassals, so I'm not really the expert on capitulation mechanics, but here is what I recall:

- We need twice their land or population (not counting our vassals' land and pop), as already noted by elitetroops. That's the main issue I've been running into in OCC conquest/domination games or when playing non ancient era starts. ("We're doing fine on our own")
- We need 1.5 times their power. (A peace vassal deal requires us to have 3 times their power, unless they are our land target). Our power is reduced if another AI has more war success than us, but that shouldn't be an issue here. Edits: Our power is multiplied by 0.75 if we are their worst enemy. Their power is modified by the leader's iVassalPowerModifier (see leader XML, resulting multiplier ranges from 0.8 (Gandhi) to 1.5 (Genghis)). (Thx, WT) ("We're doing fine on our own")
- We need 40 more war success than they have. (Easy) ("Surely you must be joking")
- Peace treaties that keep them or us from declaring war can be an issue.
- They won't agree to become a vassal if they have more than average power. ("We're doing fine on our own") A glance at the code shows that average power is total_power/civs_alive, so that includes us. It gets modified when they have a land target that is more powerful than them or when they are at war with a land target, which would make them more likely to vassal.
- They won't vassal if we are at war with one of their land targets that has a higher power than them. ("We are afraid of your enemies")
- They need to like us enough to peace vassal. ("We just don't like you enough")
- They won't peace vassal if we are at war, and they would have to declare war on their friends ("We couldn't betray our close friends") or if they don't like us enough for a war bribe. (Gandhi & Freddy will normally will peace vassal at cautious, but need to be friendly and pleased, respectively, while we are at war with another civ.) ("We just don't like you enough")
- They won't peace vassal if we are too close to a domination victory (If their land and pop gets us over 75% of the domination threshold.) ("You've grown too powerful for us.")

An early GA bomb could be useful to get some early peace vassals or faster capitulation, maybe. How long does it take to get to the power requirement? Too long?

I've had a crazy busy few weeks, but I should have some time this weekend to finally get started with this Challenge Series. I don't think I'll start with this game. Too much unit micro. I still haven't quite recovered from my last SGOTM turnset.

Thx Pollina, for spelling it out. That's why I love having you on my SG team.
I thought there is one more thing. In the leader's XML, there is a multiplier that you apply to the 1.5 power. So someone like Mansa only needs you to be 1.2 or 1.3x their power and others are harder to capitulate than 1.5.

Pollina.

Free tonight?

Is there any advantage to natural or pressed? I've been selecting solid because I figured it would be more together, less sprawling. Is that right?

I use solid

If you are looking for a micro pangaea you have to use solid. Toroidal/high sealevel/pressed goes down to about 550 tiles, while solid can go as low as 370. (The winning map from Challenge Series I was 660 tiles.) The smallest maps are quite rare, so can be a pain to find one though. (Tip for finding them: generate a HUGE amount of maps with map finder, then sort the files by file size to find the maps with the least land.) Maps with 440-460 tiles are more common, don't know if those extra 60-70 tiles make much of a difference here. Maybe 370 tiles is even too small for this as you might not have any space to plop down your other 2 settlers.

On a 370 tile map it can be quite easy to box in an AI with only one city. This could possibly lead to peace vassaling. I haven't done any math, but everyone starts with quite high power rating due to all the techs, so it probably takes a while to get your own power rating high enough to peace vassal someone.

I manually generate, but I look at tile count first, then oil/coal.
My map, unfortunately was about 650. I went with it because the resources looked so good. I'm seeing a large jungle gap north of my starting group:
Stalin -- Me -- Roos

If something goes wrong with this game or if each AI gets 4 or more cities settled, I'll play again with less tiles. I figured this would be a good warm-up game.

Mapsize and layout are interesting factors. In my map, only Washington was nearby and I eliminated him with his capital. That meant I had to reach to get to all the others. Four were cramped together, limiting their expansion, which was a boon for me. Only Zara (iirc) had space and she dragged my conquest out several turns.

If the AIs are spread around, then it's easy for them to have 4 cities or more before you reach them and that can put a sort of hard cap on how fast you can win.

iVassalPowerModifier is 80 for Gandhi, 100 for the others in this scenario.

VassalRefuseAttitudeThresholds are:

Annoyed: Fred, Max
Cautious: Joe, Zara
Pleased: FDR, Willy

Help me understand this now. To get Max to peace vassal, we basically need:
1. 2x pop or land
2. Max has 0.8 * 2 power (?) or 0.8 power and land target.
3. Cautious
4. Other stuff probably not required.
Right?

To get Max to war vassal:
1. 2x land or pop
2. Max has 0.8 our power
3. 40 more WarSuccess
4. Other stuff probably satisfied.
Right?

Someone please explain land target again, precisely. We have 8 land tiles touching his land in any of all 8 directions? He has 8 land tiles touching our land in any of all 8 directions? Or what?

I regened about 10 times, always between 647 and 670. How many regens to get something significantly lower? That really sounds like an advantage to me.

LowtherCastle said:

Someone please explain land target again, precisely. We have 8 land tiles touching his land in any of all 8 directions? He has 8 land tiles touching our land in any of all 8 directions? Or what?

Click to expand...

Go here for the best explanation: http://forums.civfanatics.com/showthread.php?t=561148

Summary: You touch 8 of his land tiles in any direction, or as you say, he has 8 land tiles touching our land in any of all 8 directions.

LowtherCastle said:

I regened about 10 times, always between 647 and 670. How many regens to get something significantly lower? That really sounds like an advantage to me.

Click to expand...

Are you sure you used the right settings? High sea level, Toroidal world wrap, Solid shoreline. I just generated 10 maps with those settings (modern start, BUFFY 3.19.003), got one 37X tile map, two 45X maps, four in the 540-560 range and only three over 600 tiles.

The 37X maps aren't quite 1 in 10 though, I think more like 1 in 20 or something over a large sample. (Or maybe it was even 1 in 40... I looked at a very large sample a year ago or so, but can't remember the result anymore.) This is why I prefer mapfinder and the file sorting method. You can run it with quite lose settings to get a ton of maps, then look through the smallest maps for anything playable, knowing they all have small amounts of land.

Ty, elite.

No, I wasn't using toroidal. Evidently, that's legal for HOF Challenge games?

LowtherCastle said:

Ty, elite.

No, I wasn't using toroidal. Evidently, that's legal for HOF Challenge games?

Click to expand...

It depends on the map script - see here.

I've submitted a quick run at this. Very fun.

Mirco-pangaea is a big advantage. Mine was 45x with an island. It makes getting a quick-kill on an AI easier, and the cramped conditions make 2-mover sneak attacks easier.

It seems to me, that getting the AIs to talk is the slowest bit of the endgame conquest. Once they talk, additionally satisifying the capitulation conditions doesn't seem too hard. Roosy is the longest non-talker in this game.