SGOTM 15 - Kakumeika

DanF5771 · Feb 15, 2012

Thanks for the pdf, mabraham, nicely allows quantitative comparison.
Could you explain the 2 side columns, left and right of the main column for each turn and what 'Build Queue Movements' means?

I'm still favoring the FPfarm-first variant:
Working it continuously after completing the farm does delay SH by 1 turn (T40 compared to your T39) but Delhi's borders will pop on the same turn (501 :culture:

on T58).
Likewise the Settler would be produced 1 turn later with regular production (T49 compared to your T48), but for a fair comparison we could also assume to 1pop-whip it so that Delhi is at size 4 at the end of both tests. BW fits in very nicely after Hunting so that Eiffel doesn't waste any worker turns on unneeded improvements and can start chopping directly after improving the Deer.
So we get the Settler out at the same moment and have more overflow (16 instead of 13) plus we can produce TGW in Delhi for one turn (before the Warrior when growing to 5) to get 22 :gold:

of fail gold. This and the commerce from the continuously worked FP more than makes up for the lost commerce from working Bombay's Gold mine 2 turns less.

The only disadvantage I see is the minimally increased risk to lose Stonehenge on T39 and the Fast Worker produced by Bombay a bit later (1 turn?).
Imho pushing back BW after TheWheel is an adaption due to the timing of research<->worker_actions not adding up and thus something we should avoid, no?

Note: I don't really want to whip Delhi 5->4, I believe there is no immediate hurry to settle near Toku since we will not beat him to his 2nd city anyway but have time to beat him to his 3rd city.

Manco Capac · Feb 15, 2012

DanF5771 said:
And as mentioned earlier, city maintenance is most likely a bit cheaper in the real game (still -3 with Delhi@5).

I am closing in to embrace mabraham's microing (A pdf is always good).

But WastinTime made a mention I did too much wealth (slider) than it needs.
Anyone has a guess what would the real maintenance with both cities?

Found it. Will calculate a little later. Gotta go.

DanF5771 · Feb 15, 2012

Sorry, the "-3" in the above quote is not the amount of city maintenance, but our total expenses (includes -1 due to civic upkeep). For calculation of city maintenance you mihgt want to try out this xls-spreadsheet.
It's for Emperor level, maxPlotDist can be changed as well as number and population of cities (blue fields). Total city maintenance is given in red field (2 in the example with Delhi size 5 and Bombay size 2 at plotDistance=6 and maxPlotDist=64 not 56 as in test game).

View attachment CityMaintenance_v2.zip

shulec · Feb 15, 2012

mabraham said:
To do this T37 SH, you need to work zero turns on the flood plain after the settler finishes, as well as switch off the corn. Correct

By T38 that has cost 82 in the capital for a return of 5 base and one turn earlier on SH. I'll take your word for that

This will slow all future growths in the capital by at least 2 turns. Not true.
I don't think the gain is worth the cost because I think SH and Oracle are likely to be quick enough at T38 and T61.

Yes, but every settler and worker our capital builds for the rest of the game will be 2 turns slower. Again, not true.

That is very likely to reflect in our victory date if our Oracle is fast enough already. Unfortunately we don't know whether we are fast enough...

The micro I propose will get us SH on T38, a warrior on T41 and a Settler on T48. The difference is your micro has 3 :food:

and 13

overflow. My micro has 0 :food:

and 10

overflow. A sacrifice of 3 :food:

and 3

to get Marble into our cultural borders one turn earlier.

I don't think I laid out my exact micro, but let me do that here:

After the first settler, I worked the Farmed corn and mine at 2 pop building stonehenge with the immediate stone bonus.
At pop three, I worked an additional mine.
On T37, work three mines, no corn.
T38 start warrior working Corn, FP, mine finish in three turns.
On T41 at pop 4, start settler while working corn and 3 mines.
T48 settler finished, off to east.

This is the same as your spreadsheet, but there is a sacrifice of 3 :food:

and 3

.

The difference is that I work the three flood plains later than you. I work one turn without corn and worked a mine instead.

EDIT/Addendum:

Based on previous testing, this gives a border pop on T57.

Manco Capac · Feb 15, 2012

DanF5771 said:
Sorry, the "-3" in the above quote is not the amount of city maintenance, but our total expenses (includes -1 due to civic upkeep). For calculation of city maintenance you mihgt want to try out this xls-spreadsheet.
It's for Emperor level, maxPlotDist can be changed as well as number and population of cities (blue fields). Total city maintenance is given in red field (2 in the example with Delhi size 5 and Bombay size 2 at plotDistance=6 and maxPlotDist=64 not 56 as in test game).

edit: wait .... removing bug ...

~~Which bug?~~ Different values I see.
Thanks for the spreadsheet.
I see our test game has a total maintenance of -4 when city#1 gets 5 pop and city#2 gets 2 pop. In other words, a bit more expensive. Anyways, for my turnset, the maintenance is accurate.

DanF5771 · Feb 15, 2012

yup, had made some 'cosmetic' changes which altered the formula ...
now it is correct.

... we'll know maxPlotDistance for sure once we settle Bombay.

Manco Capac · Feb 15, 2012

I've looked through shulec's path of getting SH one turn earlier and I admit if mabraham really wants to keep the hammers invested in the warrior, then both paths see the settler starting on T41. But do we want to sacrifice 3 :food:

and 3

, equivalent of one turn into a plain cow.
I also verified if any hammer may come to loss by rounding off, nope, the time is all right.
Do we want to ensure even more the Oracle more than it can be, assuming most AI's are Toku's?

We have to slove the Gordian knot before moving on.
BTW, I assume everyone agrees in settling NW as first city.

But WastinTime made a mention I did too much wealth (slider) than it needs.

After all, I don't really get the exact point.
My 0% / 100% comes from the fact we lose one beaker when the commerce output is an odd number during pre-librarian era. But with an even number, then 50 % slider is another option. Early in the game, the percentage of gained beakers is non negligible.

---------------------------------------------------------------------------

Okay. I made a little study through mabraham's saves and unfortunately I couldn't conclude anything more than Hamuragawa and Toku don't have Agriculture in the case they have it BOTH.
Indeed, 5% discount for each known civ having a tech we are teching does not make any difference on 10-11 beakers.
But 10 % from two civs make one beaker difference.
Not conclusive then.

There was a little error in my test game of Toku's behaviour:
I gave him Mysticism (and Meditation).
Still, the first attempt with riverside corn was the most optimal I could see from such AI.

WastinTime · Feb 15, 2012

Just try to pay attention to the 20% research boost which is the most likely multiplier. It's not always true, but generally, as you all probably know: Research = (Beakers+1) * 1.2 (truncate result)

Beaker totals like 9, 14, and 19 are good. try to avoid 8, 13, etc. See table below:

Bkrs--> Research:
8 ---> 10
9 ---> 12 (jumps +1)
10 ---> 13
11 ---> 14
12 ---> 15
13 ---> 16
14 ---> 18 (jumps +1)

Manco Capac · Feb 15, 2012

WastinTime said:
Just try to pay attention to the 20% research boost which is the most likely multiplier. It's not always true, but generally, as you all probably know: Research = (Beakers+1) * 1.2

Beaker totals like 9, 14, and 19 are good. try to avoid 8, 13, etc. See table below:

Bkrs: Research:
8 10
9 12 (jumps +1)
10 13
11 14
12 15
13 16
14 18 (jumps +1)

Oops. I forgot about BW innate 20%. Very good point.
In other words, I should try to combine both the rounding off beaker and the 20%. Let's see if this doable. But the fact the FP is worked on different time between shulec's and mabraham's plan change how I deal the slider .

I know a PPP is needed, but I can't incorporate two theories.

DanF5771 · Feb 15, 2012

Tachywaxon said:
Indeed, 5% discount for each known civ having a tech we are teching does not make any difference on 10-11 beakers.
But 10 % from two civs make one beaker difference.

There are 9 teams in the game (us = team0, 7 single AIs in 7 teams, 2 AIs together in 1 team) - the possible research modifiers for TECH_COST_TOTAL_KNOWN_TEAM_MODIFIER amount to:
AI teams we know that have the tech / Modifier

Code:

0	+ 0%
1	+ 3%
2	+ 6%
3	+ 10%
4	+ 13%
5	+ 16%
6	+ 20%
7	+ 23%
8	+ 26%

With 2 neighbors who know the tech we just get +6% which we will only see if we have 16 raw beakers, otherwise they get rounded down.

mabraham · Feb 15, 2012

DanF5771 said:
Thanks for the pdf, mabraham, nicely allows quantitative comparison.
Could you explain the 2 side columns, left and right of the main column for each turn and what 'Build Queue Movements' means?

Actually they're two columns "on the right" of a main column. They're various intermediate subtotals. Managing multipliers and overflow and food to hammer conversions is pretty complex. There's a bunch of hidden rows

[Did you know... some national wonders don't have the 50% hurry penalty? That the first unhappy population does not eat while building a settler or worker? The first unhealthy population doesn't lose a food while building a settler or worker? These discoveries arose from past game-vs-spreadsheet mismatches

]

The build queue movement values are indices into the then-current build queue, which isn't shown. It allows for the user to queue up a bunch of plausible builds and to select from the queue, store incomplete former builds with their progress hammers, and have items pop off the front when complete. Not as pretty as you'd like, but it works

DanF5771 said:
I'm still favoring the FPfarm-first variant:
Working it continuously after completing the farm does delay SH by 1 turn (T40 compared to your T39) but Delhi's borders will pop on the same turn (501 on T58).
Likewise the Settler would be produced 1 turn later with regular production (T49 compared to your T48), but for a fair comparison we could also assume to 1pop-whip it so that Delhi is at size 4 at the end of both tests. BW fits in very nicely after Hunting so that Eiffel doesn't waste any worker turns on unneeded improvements and can start chopping directly after improving the Deer.
So we get the Settler out at the same moment and have more overflow (16 instead of 13) plus we can produce TGW in Delhi for one turn (before the Warrior when growing to 5) to get 22 of fail gold. This and the commerce from the continuously worked FP more than makes up for the lost commerce from working Bombay's Gold mine 2 turns less.

The only disadvantage I see is the minimally increased risk to lose Stonehenge on T39 and the Fast Worker produced by Bombay a bit later (1 turn?).
Imho pushing back BW after TheWheel is an adaption due to the timing of research<->worker_actions not adding up and thus something we should avoid, no?

Note: I don't really want to whip Delhi 5->4, I believe there is no immediate hurry to settle near Toku since we will not beat him to his 2nd city anyway but have time to beat him to his 3rd city.

That looks very nice, and different from what I'd had in mind based on my earlier interpretation of your plan. I spreadsheeted both cities below - please let me know if I've inferred wrongly. Your plan seems to get an earlier worker in the Deer city (depends a bit on chop timings I guess), and works a bunch of extra turns on the floodplain farm (+1 :commerce:

each) as well as the 22 TGW fail gold to more than compensate for the lost two turns on the gold main.

shulec said:
The micro I propose will get us SH on T38, a warrior on T41 and a Settler on T48. The difference is your micro has 3 and 13 overflow. My micro has 0 and 10 overflow. A sacrifice of 3 and 3 to get Marble into our cultural borders one turn earlier.

I don't think I laid out my exact micro, but let me do that here:

After the first settler, I worked the Farmed corn and mine at 2 pop building stonehenge with the immediate stone bonus.
At pop three, I worked an additional mine.
On T37, work three mines, no corn.
T38 start warrior working Corn, FP, mine finish in three turns.
On T41 at pop 4, start settler while working corn and 3 mines.
T48 settler finished, off to east.

This is the same as your spreadsheet, but there is a sacrifice of 3 and 3.

The difference is that I work the three flood plains later than you. I work one turn without corn and worked a mine instead.

EDIT/Addendum:

Based on previous testing, this gives a border pop on T57.

Correct, spreadsheeted below. Apparently I was comparing a different approach with my earlier analysis, sorry.

Tachywaxon said:
I've looked through shulec's path of getting SH one turn earlier and I admit if mabraham really wants to keep the hammers invested in the warrior, then both paths see the settler starting on T41. But do we want to sacrifice 3 and 3 , equivalent of one turn into a plain cow.
I also verified if any hammer may come to loss by rounding off, nope, the time is all right.

The three food goes into a bin and sits there for quite a while, if our next few builds in the capital might be settler, worker, Oracle. So I'm not concerned by the loss of 3 :food:

3

if people think that's a sound price for insurance on SH and Oracle.

Tachywaxon said:
Do we want to ensure even more the Oracle more than it can be, assuming most AI's are Toku's?

I think the T62 Oracle is fast enough, but that may be more wishful thinking than not. I don't know of a way to gather representative data

Tachywaxon said:
We have to slove the Gordian knot before moving on.

One "cleaves" a Gordian knot in English

On balance, I now like DanF's floodplain farm plan the best but if we want cheap insurance on SH and Oracle then shulec's plan is good.

Tachywaxon said:
BTW, I assume everyone agrees in settling NW as first city.

I really hope so - haven't seen any suggestions to the contrary (apart from the ones you made which I think you've mostly retracted).

shulec · Feb 15, 2012

I think all of the micro plans look really good. The question then becomes, how much risk is it to push back the Oracle 2-3 turns? T62 is 1520 BC. The Oracle can definitely go before that on Emperor games. I have seen it go as late as 700 BC.

I personally feel that since we are playing a game with no tech brokering, the free tech is more valuable than usual. That being said, DanF's micro looks really good and appears to be more productive overall.

Manco Capac · Feb 16, 2012

It seems a consensus has been done around DanF's micro plan and I really like how it times so perfectly. Especially BW timing exactly with almost no OF along last turn of deer improvement. I adopt his plan right away. Good job proving the point food is always the way.
Slider micro determined according to WastinTime's rules. The play will occur tomorrow, unless someone wants me to play around the middle of the night. :mischief:

BW at T36.

Will emit the final PPP along the final PPP video (unless this method isn't liked at all).

mabraham said:
I really hope so - haven't seen any suggestions to the contrary (apart from the ones you made which I think you've mostly retracted).

Yes, I "flinched" from the risk of losing the city site to Toku. Not enough escort and the site is suboptimal short term compared to the site you first proposed, despite having little long term sacrifices like being one tile off coast. Still, we may be lucky no hidden strategic resources are laid on that part of the region, messing with settling values. And, you know, if ever Toku has an affinity for that location first, will Neilmeister short-circuit teams from getting the strat. resource. The luckiest getting it, the others rot in hell.

mabraham said:
[Did you know... some national wonders don't have the 50% hurry penalty? That the first unhappy population does not eat while building a settler or worker? The first unhealthy population doesn't lose a food while building a settler or worker? These discoveries arose from past game-vs-spreadsheet mismatches ]

How many other secrets are you hiding from the rest of us.

I just learnt about the development unit (worker or settler) effect on first unhappy pop from VoiceofUnreason.

@DanF for post #570

Thanks. Makes sense more teams for a given map will lead to smaller portions of discounts.

Manco Capac · Feb 16, 2012

Tech choice:

Masonry => Hunting => BW

Considering WastinTime remark on discount BW gets. I changed how I will deal with the slider.

Finish Masonry (For T32) 100% :science:

slider
4 turns 100% :gold:

slider (41 :gold:

in bank); a tech which won't receive any discount and the number of civs met is insufficient to boost with the meager 6% on 10-11 beakers.

4 turns Hunting 100 % :science:

slider
Remaining turns on BW until T46 (date we get use of the tech), 100 % :science:

slider

Wheel will come right after for next turn-set. This time, the meager 6% has its effect.
Toku has wheel.

Cities:

Delhi:

After settler completion (T30) , SH is the next goal; one shot completion. Masonry is researched next turn, making stone active for doubling production rate before the hammers are banked.
On fringe of city growth to pop three, swaps GH to improved FP. The next pop will assign to GH. After SH completion, one turn into GW. Afterwards, a two-turn-warrior will follow along a growth timing with it. 5 pop. Then settler for 6 turns.

T32 founding turn -Bombay

On foundation, the first build the TGW, working the deer tile.
With perfect timing with first Bombay growth, the gold mine is up.
At this moment, change to worker production for 10 turns.

Workers:

Eiffel:
Finishes the FP. Then two turns moving northward. Then improve gold. One turn lost moving on deer, then improve. Finally start a chop, which sync with finishing BW.

Explorers with clubs:

Amundsen:
Will avoid the lion and bear unless someone wants to take its chance against the lion.
Will stay for fogbusting around the gold or go near fish area to uncover fogged tiles in search of seafoods.
Or perhaps, find a way to avoid the bear and keep an eye on Toku for settling location?

Burke:
Will continue its voyage north-west? once the second city is settled.

Third warrior (Columbus): May explore or stay there against barbs. Your choice?
I think more exploration. I need a vote on this.

Espionage:
Always on Hamuragawa since we started with him.

Stopping points:

Meet a new AI.
A dilemma regarding animals.
An info that breaks the planned game.
New culture pattern.
Near impossible but if it happens loss of SH.
Resource popping on a worked hill (when that RNG starts?)

Miscellaneous:

Take screenshots of the demographics screen,
Look for AI culture in sattelite view,
Save game,
Note (using Alt+S) forest/jungle growth, animals appearing etc.

Visual PPP:

mabraham · Feb 16, 2012

PPP looks great. If we don't know that our later Oracle is early enough to be a sure thing, then slightly increasing the risk of failure for a significant return (growth to size 5) sounds like a good prospect. We need to accept some risk to win if others might be doing similar strategies.

Tachywaxon said:
Explorers with clubs:

Amundsen:
Will avoid the lion and bear unless someone wants to take its chance against the lion.
Will stay for fogbusting around the gold or go near fish area to uncover fogged tiles in search of seafoods.
Or perhaps, find a way to avoid the bear and keep an eye on Toku for settling location?

Don't really mind, prefer to avoid combat generally, but moving to the forest and taking C1 is probably around equivalent to no upgrade in a forest (so ~95% survival). Want C1 in the hope that we can end up fortified on that hill (or inside our third city if we don't get TGW, or don't get it in time).

Tachywaxon said:
Burke:
Will continue its voyage north-west? once the second city is settled.

Third warrior (Columbus): May explore or stay there against barbs. Your choice?
I think more exploration. I need a vote on this.

I think we need a coastal explorer, and someone lurking around the wheat such that we can see all the land tiles (so barb warriors/archers can't spawn and threaten our second city). No need to be in the city, of course.

We'd have to stay really close to Delhi to provide cover there - and it can build a warrior in 2 turns if something threatens.

DanF5771 · Feb 16, 2012

Really glad you like the FP-first micro.

Thanks, mabraham for spreadsheeting it and for the explanations, yeah projecting all the BTS-mechanics into an excel document is quite a challenge. :goodjob:

If we agree on not 1pop-whipping the Settler on T47 but prefer regular production so we can do without overflow from the Warrior, we could squeeze out some more TGW- fail gold (30 :gold:

) with a slightly different micro:
After SH start building Warrior working Cf, FPf, 2*Gm on T40 and then switch 1 Gm to GreenForest on T41, finish Warrior with 3 overflow.
On T42 build TGW for 1 turn working Cf and 3*Gm. This represents a maximization for building TGW for a single turn.

Delhi starts Settler on T43 @ size 5 with 0 :food:

and 0

in the boxes, still finishing it in 6 turns (6*17 :hammers:

= 102

).

Visual PPP looks great, Tachy!

Re Ammundsen's odds:

I'd also like to avoid combat -- keep the lion alive so that it will spawn bust against appearance of barb Warriors/Archers.

DanF5771 · Feb 16, 2012

Wanna save 1 beaker?

During the 4 turns with slider @ 100% :gold:

select Hunting for only / exactly 2 turns, switch to BW for the other 2 turns. That way we will get Hunting online with zero research overflow. Otherwise the 2 overflow beakers will reduce to 1 beaker when the modifier (=106% as both neighbors have Hunting) gets removed.

Manco Capac · Feb 16, 2012

I am still waiting more feedback until 7-8 hours later. By that time, I shall play.
I will consider more deeply all the remaining details you want me to incorporate (like that beaker lost from OF of Hunting :eek:

; nothing escapes from you!).

I also may try to sum up all the beakers we saved from all this micro (in other words, the 5 first turns and my turn) to show how far the impact of perfect micro may bring.
Already, The almost perfect fitting of BW shows us how our micro save one worker turn.

BTW, DanF, continue with all the ideas of super micro that might crop up in your mind; Kakumeika is one of most dedicated team to micro, even it is our motto!; there won't have any: "Bah! That is too maniacal!" or "All is grandstrategy!" (Although the last has some truth...).

BTW, I recommend people to download the visual PPP if they want to see details. I haven't found the way to get better qualities while minimizing the video size.
4shared is the settled platform from now on. In the not so far past, Megaupload was my favorite platform, but was taken down by FBI. (Arghh! I lost all my files.)
Another way to protect those vids from cheap stealing:
I noticed the first part of my posts were revealed. I alarmed myself and concluded where to put my vids.
Yes, I strategically put all my videos at the end of my posts; long ones.
May look quite maniacal, but visuals are the most helping factor to those who wants to cheat. And I don't think visuals with videos were used up to now, so it may be a delicate case.

bcool · Feb 16, 2012

PPP looks good. I also prefer to avoid combat.

mabraham's suggestions of coastal explorer and fogbuster around the 2nd city sound good to me.

Manco Capac · Feb 16, 2012

mabraham said:
We'd have to stay really close to Delhi to provide cover there - and it can build a warrior in 2 turns if something threatens.

Hmm. I forgot the arrival of the 4th warrior. That warrior may not go elsewere; he's a necessary MP I think. At pop 5, I think it is necessary. Will verify later.

SGOTM 15 - Kakumeika

Emperor

Friday,13 June,I Collapse

Emperor

Grrrrr... I AM the force!

Friday,13 June,I Collapse

Emperor

Friday,13 June,I Collapse

Deity

Friday,13 June,I Collapse

Emperor

Deity

Attachments

Grrrrr... I AM the force!

Friday,13 June,I Collapse

Friday,13 June,I Collapse

Deity

Emperor

Emperor

Friday,13 June,I Collapse

Deity

Friday,13 June,I Collapse

Similar threads