SGOTM 05 - Murky Waters

[LowtherCastle]

What we can test in advance

• Given: We cannot test for a better place to settle.
• Knowing all the AIs, our 50% research penalty and our barb beaker bonus, we can get a very good idea whether any tech path(s) are significantly cheaper.
  
  My hunch is that Writing does not get much cheaper between T40 and T80, but that BW does.
  
  This could mean that beeling Writing is like using Alphabet to get a lot of cheap techs--we research Writing while others research everything else for us.
  
  To generalize the above principle: Our optimal research path beelines to our tech goal following the least-travelled path. This is similar to the Alphabet tech-trading strategy, except that we don't prioritize techs for trade value (e.g., Paper), only techs on our Critical Path (either >Construction or >Astro).
  
  So the testing question is: What is optimum?
  • The shortest path to Writing (= Agri>AH>Wtg), which also enables what we know we have (farmland and pigs).
  • The shortest Fishing path to Writing (= Fish>Wheel>Pottery>Wtg), enables WBs for nets and exploration and opens up Sailing.
  • The shortest path to warring (= Fish>Mng>BW>Sail; maybe >Wheel)
  • The agricultural path to warring (= Agri>Mng>BW>Fish>Sail; maybe Wheel)

 

[klarius]

Ok, try to klarify barb research:

The barb team (in other words we) gain

floor((c * floor ( ( 3*nw ) / na))/ 100)

c = cost of tech
nw = number of teams with tech
na = number teams alive (not counting us)

beakers per turn on every tech.

That means when na=7
1% if 3 teams have the tech
2% if 5 teams have the tech
3% if 7 teams have the tech

all percentages are rounded down. So e.g. 1 beaker/turn for fishing in the beginning (3 civs know it = 1%; cost around 120)

It hasn't anything to do with what we or the barbs research.
As monarch AI start all with archery we get 3% of the cost of archery every turn and will learn it after 35 turns (because of rounding down). We will learn it even w/o having the prerequisite hunting and by that unable to research it.

BTW killing one AI would be nice, as already then it would be only 2 AIs with the tech necessary to get beakers.

The longer we don't research techs known to the AIs the more beakers we get for free. As the starting techs get researched pretty quickly, we should restrict ourselves to only research those we absolutely need.

The second benefit of waiting is that when we finally research a tech (we will not wait for all to come for free) we get a bonus on our research if more civs know it already.

You can see the beakers the team gains by just going into the tech adviser and hover over the techs. Even after one turn you will see that we have gained beakers in fishing, mysticism and archery, even if we research something else.
BTW, that's the team beakers (also in normal games for PA or MP teams) what you see here. When the barbs start to research, their measly beakers will also get added there immediately.

 

[Big Pig]

If scout goes NW NW, then we might immediately know if settling in place or 1 tile W is good. Exactly!If Settler goes NW (not river tile I'm guessing) then he has to go S and settle on T3 to be on the river or SW or SE to settle on T2.

There are a few good choices. Kind of depends on whether we're willing to settle on T3 or not.

I don't think settling on turn 3 would be a disaster on epic if we can find a good site. But the NW NW initial scout move will help determine if settling on site is a good option (altho' it won't see the coast tiles to the N)

PS. Our workers can build improvements in barb city squares. And for that matter, our GEs can build wonders in barb cities if we so desire (can't imagine we would tho'....)

 

[LowtherCastle]

How does the 50% team penalty affect c = cost of tech? (In other words, what does the equation look like including that penalty?)

Are you saying that any research the Barb is doing is accumulating beakers for us?

 

[Big Pig]

I suggest the following order of play. Any objections?

LowtherCastle
klarius
Mîtiu Ioan
Murky
jesusin
Erkon
Gnejs
Big Pig

I think traditionally the captain goes last to clear up everyone's messes

Unless, you feel you need Gnejs and I to clean up your mistakes that is......

 

[klarius]

I think we are in a hurry not only to kill one AI, but to kill all . It's only monarch and the AIs are definitely hampered by this special setup. Going slow and building a nice empire will not cut it. Rather burn down bad cities and let the barbs spawn new cities.

 

[LowtherCastle]

I think we are in a hurry not only to kill one AI, but to kill all . It's only monarch and the AIs are definitely hampered by this special setup. Going slow and building a nice empire will not cut it. Rather burn down bad cities and let the barbs spawn new cities.

I definitely agree. No builder mentality allowed. My main concern is not city maintenance so much as military maintenance and being able to reach enough AIs in time to conquer or dominate with what we do research.

One thing for sure: the earlier we war, the easier to conquer.

(I'd just hate to beeline BW + Sailing and find out the Gold winners beelined Writing and got BW for almost free ALMOST AS FAST AS WE DID.)

 

[Big Pig]

I think we are in a hurry not only to kill one AI, but to kill all . It's only monarch and the AIs are definitely hampered by this special setup. Going slow and building a nice empire will not cut it. Rather burn down bad cities and let the barbs spawn new cities.

Forgive my ignorance, but could you spell out the ways the AI is hampered by the set-up. Some I can work out for myself (faster appearance of barb axemen, us not being pillaged allowing greater initial growth than the AI etc) but I want to make sure I am not missing anything.

Understanding the extra advantages we have over the AI in this set-up will help us in coming up with winning strategies

 

[klarius]

Beaker cost:
You will just see when you look into the tech adviser. E.g. fishing is shown as 117 beakers necessary. Normally it would be 78 for a monarch, epic, standard game.

About the AI:
The AI doesn't know that they are at war with us. So their build preferences are as if they were at peace. They will see no reason to build any attackers besides their standard reserves (which is something like 1-2 units per area).

They don't know that they are in contact with the human. Contact with human does increase the required garrison by 1 in the normal game and the AI always builds a defender if a city doesn't have the required garrison.
As long as they have other things to build (and there are no threats at the gates of city), they will be happy to have one defender in every city below size 9 (2 in the capital).

If we make peace with them, they will not plan a new war against us. By that they will also not declare (unless provoked by us rejecting a demand). They will also not ask other AIs to help if they are attacked by us.

A lot of small things, which should make the game easier than standard monarch, so I think one has to go really fast if one wants to get near any laurels.

 

[Erkon]

Please check out the changes in the patch. I've tried to highlight the most interesting stuff, but check the readme for all details.

Spoiler :

• First strike combat bug fix - is this of importance?
• Fixed pop-rush rounding bug - is this of importance?

Changes

• less hammers from chops the farther away from a city...
• reduced hammer yield from forest chops to 20…
• Mathematics increases forest chop yields by +50%...
• Representation: +2 happy in X biggest cities
• Hereditary Rule: Low Upkeep
• Slaver: Low Upkeep
• firepower is now average of curr and max strength...
• Great person point generation from Wonders continues after Wonder has become obsolete

Fixes

• Fixed rush production per population for game speeds other than normal

Additions

• unit cycling optimization...

 

[klarius]

None of any concern.

The stuff below "changes" are 1.61 patch changes (it's a cumulative patch)..

The red stuff was anyways already fixed in the HoF mod.

 

[LowtherCastle]

For what it's worth...

Using klarius' formula:

Tech----------3/7 AIs------5/7-------all
----------------------------------------
Mng/Myst-----1/t-------3/t-----5/t
Agri/Wheel----2/t-------4/t-----6/t
Fish/Htg------1/t-------2/t-----4/t

1. Waiting for the 4th and 5th AIs pays off more with Mining than with Fishing.
2. Another way of looking at it is that waiting for 3 AIs to know Mining is undervalued.

Any idea what the Barbs start out researching and their research path in general?

 

[LowtherCastle]

Ok, maybe here's an example of what I'm talking about about:

Here you see that at 1960BC they've evidently researched Hunting>AH>Fishing>BW. Note the cost of Sailing. It's still full price. Now researching BW before Sailing gives you time to settle your city and build the copper mine, IF the copper is on your island. If it's not, BW enables you to poprush the sailboat when you finish Sailing. Is that faster than Sailing first and then reearching a discounted BW? Who knows (without some testing)???

 

[klarius]

What the barbs research is pretty irrelevant. They will have their research slider @ 0%, because they cannot pay their maintenance (at least this was so every time I looked; look into the city screen of a barb city to see their slider settings). So they gain exactly 1 beaker/turn.
But generally every AI team member will help the research of their team mates whenever they chose a new research. So unless we change horses in the midst of a research they will research the same we research.

 

[klarius]

If you're talking about where the barbs receive the most beakers for free, that's complicated.
There are literally hundreds of weight factors the different AIs apply to decide what to research. And on top of this is a big random number. But with three AI having mysticism there is a big chance that these three go for a religion first and by that don't add anything useful.
The worker techs are largely influenced by the land around their capital, so there's no way to predict what they do first.
BW may take a long time, if they are isolated or at least don't plan wars. Or it may be pretty quick, if all are together and happily plan backstabbing each other.

I still think that we shouldn't try to rely to much on this stuff. My take is:
First find out if we need fishing or agriculture for the capital. If yes do this, if no do mining for now.
When we get the animal intelligence reevaluate.
Take into account neighbours (->BW), if any.
Better city locations than capital (I would then still go for BW to get out settlers quickly).
Small island w/o any good locations (fishing->sailing).

BTW, I don't think writing w/o BW helps us a lot. It will take forever to build a library by hand in the hammer poor region we see up to now. And in the one test I did BW took more than 110 turns to come for free.

 

[Erkon]

LC, are your directives of how to make a test game up to date?

http://forums.civfanatics.com/showpost.php?p=5626695&postcount=101

I also noticed that there's one extra AI compared to the usual setting for standard map size. This means that, unless we're isolated, the closest AI is not that far away. And that we need to build a warrior instead of worker, yes?

 

[Big Pig]

I also noticed that there's one extra AI compared to the usual setting for standard map size. This means that, unless we're isolated, the closest AI is not that far away. And that we need to build a warrior instead of worker, yes?

Not if the AI doesn't know it is at war with us.

 

[klarius]

No we don't need to build a warrior. Our animal friends will distract our neighbours for a long time. And even if they knew they are at war with us , they wouldn't know where we are, while we will soon know where their units are.

 

[klarius]

One thing that there are no misunderstandings:
While the civ AI doesn't know that they are at war with us (and cannot plan war by that), the unit AI knows that we are an enemy. So if a unit stumbles upon us by accident they will attack, if they have good enough chances.
And they will wait and build a stack with other roaming units (like the AI does with barb cities), at bad chances. So we are not completely save, but there will be no coordinated attack (though a bit of bad RNG can still get us in trouble).

 

[LowtherCastle]

LC, are your directives of how to make a test game up to date?

I haven't tried them with the new mod, but incorporating the new mod would be the only addition I know of. Now one thing I don't know is whether the new mod also requires that new patch. I supppose it does.