Qsc19-Ottomans Results & Strategy Discussion

[Txurce]

I continue to score in the 2d division of the QSC rankings, and feel it's primarily due to one glaring deficiency: the inability to manage an optimized settler factory. I did much better in gotm 20, but can't really say that I have cracked the equation that allows a good location to pump out settlers in 4-5 turns. As some of the prior posts attest, it would be very helpful if one of the more adept players would sketch out what is required to build and manage such a factory, given varying terrain.

 

[DaviddesJ]

This has all been explained well elsewhere, but a 4-turn settler factory might go like this:

5 pop, 10 food, 0 shields
5 pop, 15 food, 6 shields (+5/+6)
6 pop, 10 food, 14 shields (+5/+8)
6 pop, 15 food, 21 shields (+5/+7)
5 pop, 0 food, 0 shields (+5/+9)

This requires that you have 5 tiles that together produce 13 food and 5 shields (always assign citizens to work these); a 6th tile that produces 2 food and 1 shield, like an irrigated plains or a mined grassland (assign a citizen to work this at size 6); and a 7th tile that produces 2 shields, like a forest, or mined plains or hill or mountain or desert (never assign a worker here, but you get the forest production on the 2nd and 4th turn when the city grows, assuming you specify "Emphasize production" in the governor menu).

A "normal" tile in despotism (plains or forest or grassland or flood plain or hill) puts out 3 food-plus-shields per turn, when improved. So 5 tiles would normally give you 15. In the illustration above, you need 18, so you need 3 units of "bonus" food and/or shields (and you also need the right mix).

Sometimes you may have to do something more complicated, with changes every turn, because you don't have exactly the right mix of terrains. Or you might have slightly more than you need, so you can "share" the bonus resources with adjacent cities, by working them some but not all turns of the cycle. But I think it's pretty easy to do those things (if you have the patience for the micromanagement) if you understand the above sequence.

P.S. You also get one extra shield when your city grows to size 7, if it's on a terrain that normally produces a shield. I didn't include that in the above calculation, but if you know you have that, you can have one less shield at some point in the cycle.

 

[MadScot]

Txurce

by no means an expert but...

you need bonus food locally to give +5fpt at 2 or 3 consecutive populations

you need to be able to generate ~7spt at the same time

you need a granary

let's say you have a couple of grasslands with cattle; irrigate both to give 4fpt, 1 spt each
your city gives 2fpt and 1spt

so pop2 will be 10fpt and 3spt

add a forest: 1fpt, 2spt, and a mined bonus grassland: 2fpt, 2spt

pop4: 13fpt, (+5 excess) and 7spt

after 2 turns you'll grow to size 5. When you do, the new citizen will work your most productive (shields) tile - a forest, we hope. So after 2 turns you get 7+9 shields.

At the start of the turn, move the new citizen onto a regular grassland. We now have 2 turns at pop5 with 15fpt and 7spt. After 2 more turns we grow to pop6, and build a settler (16+7+7=30; in fact we'd get 9 the last turn, so we could have saved a little along the way). Settler takes you back to size 4, and start again.

You MUST have some kind of food bonus to make it work; just improving regular tiles won't get the job done.

edit: too darned slow. cursed one handed typing

 

[Txurce]

Thanks very much, guys. What I hadn't figured out was the number of citizens required in the "sweet spot." You both seem to indicate that it's in the 4-6 range, depending on whether I just built or am about to build a settler. I've always been impatient, and start building settlers before I grow big enough to optimize.

 

[Ribannah]

Originally posted by MadScot
You MUST have some kind of food bonus to make it work; just improving regular tiles won't get the job done.

That is right. It is why finding those food boni early (either for the capital or for your second city - or both/more!) is so very important.

If you can't get to +5 food there are several possibilities:
- you can build Warriors in between, because you will have more shields than food;
- the Settler factory will simply start out / stay smaller (not recommended if it's your capital);
- the Settler factory will grow a bit while building an improvement, then shrink again while it's pumping out the next batch of Settlers.

The line between Settler factory and a normal city, that occasionally builds one, is thin there.

And sometimes it's wise to be impatient and build one or two Settlers before setting up the true 4-turn factory, because:
- your Workers are still busy improving the terrain and/or
- those early Settlers have a great spot to found another city that will contribute to the growth of the empire.

 

[Green Light]

First of all, thanks to the GotM staff for continuously improving these excellent events (GotM/QSC)

I have to say that i'm extremely pleased with my 16 th position in this QSC, nice climb from qsc18 ~50 th position. Propably will never reach this high position again Would have propably done better if i had known how to properly set up a settler factory. (didnt even cut the game forest ) Unfortunately i discovered the secret of perfect settler factory only from Bamspeedys Deity Settlers article when preparing to Gotm20. (excellent article BtW )

Even though there are some unclear issues to me in QSC, for example, do we get negative points for unhappy people, or why is that number 5 there like this : (5).

If this is so, it seems silly, since that could be avoided by adjusting luxury slider high enough as stated perviously by DaviddesJ.

Adjusting the slider also affects the amount of content and happy people, thus netting you more points. Or is there some twist here that i dont understand ? (wouldnt be the first time )

 

[cracker]

Greenlight,

You are correct that adjusting the luxury slider in the early game when you can appropriately do that will increase your score - - in the Civgame, in the Jason Score, and in the QSC.

The QSC scoring impact is minor but intended to at least get your attention. The early impact in the game and Jason scores can be significant and uses a cummulative average across all turns so you have to begin to pay attention to this feature early in the game to get the most benefit.

Getting good at this also avoids lost production from fallow added citizens as well as disorder.

 

[RowAndLive]

Based on a review of the saved minimaps presented for GOTM-19. It is readily apparent for all of the AI civs that the cities were developed on the same sites in each of the individual games played. Further, from comparison of the maps, you can roughly determine what order they seem to have been founded in relative to each other. Is this consistently true? Assumably it has to do with the random # generator seed being the same for the saved GOTM file, as well as a set of hard coded rules that the AI operates under. If these are true, and the cities are founded in a given order in each game (from a review of retirement timeline films), then that would further imply that the rules extend down to unit level, and that behavior will be consistent. As examples, what terrains are explored in what order, or if terrain is the same in all directions, then what directions are explored (either compass or relative to map equator) in what order; where roads & improvements will be developed in what order; what civs will always be at war with whom; what order your cities will be attacked in, etc. Thus, upon game start and except as effected by the actions of the human player, the game will always progress the same and arrive at the same conclusion. For any given game, once another civ is encountered and their locale is fully mapped, then their given behavior if unmolested should be determinable, and further, if they are countered, then their response should be equally predictable. As another example, if you knew where their next city was to be built, and you built there first, then you could also determine where they would build next instead. You could act as puppet master for them to a certain extent, forcing them to react in expected ways simply by shaping your own actions. Of course, this would need a very thorough study of comparable GOTM files, or a review of the source code.

I feel kind of silly, having this eureka moment, when it is just software after all, and it would logically make sense that it performs this way, but this just hit me as a huge leap. I have been re-reading and analyzing the lessons from GOTM 8 & 9, as well as making my own analysis of the different civs and the mathematical likelihood's of who will have GAs when, and such, over the past few weeks, but seeing the maps just put it all together for me. I'd appreciate any comments. Thanks.

 

[DaveMcW]

Originally posted by RowAndLive
You could act as puppet master for them to a certain extent, forcing them to react in expected ways simply by shaping your own actions. Of course, this would need a very thorough study of comparable GOTM files, or a review of the source code.

Technically it's possible with enough knowledge of the game engine, but you would need more information than you get in normal play. The AI's city locations are based on resources (including "undiscovered" resources). Their unit moves are affected by enemies, so you would need military plans for all civs including the barbarians.

 

[ControlFreak]

Its sort of like dropping a jar full of superballs. Knowing the physics, you could predict where every ball will end up but the complexity make the solving unfeasible.

 

[Bamspeedy]

RowandLive- If you have PTW, experiment with the debug mode.

As soon as you use a RNG number, you affect all the other AI's RNG numbers, even on the other side of the planet that you don't have contact with.

You never know, using an RNG number can cause another AI on the other side of the planet to either capture their opponent's capital, or have their entire stack of warriors fall to a lone spearman.

Where they build cities is based mostly on terrain and resources, so even though there may be a random number involved, the terrain would cause the AI to favor these sites, so even though an RNG number has been used up, they will still heavily favor these city sites and will nearly always build there.

Cracker has studied AI tendencies towards where the AI likes to build and can adjust the terrain to 'help' or encouage the AI build where he wants them to if he chooses or desires to (I'm not saying he has actually done this, though, in this game or any other).

As far as I know RNG numbers are used up when:
1. You founded a city
2. You trade techs with the AI (they use a RNG number to pick their next tech to research)
3. Worker automation, and unit 'auto-explore'
4. Battles
5. Culture flips. Not sure how often this is checked. I know it would be checked every turn in trouble spots, but don't know if a city with 0% chance of culture flipping would use up an RNG number or not.
6. Every unit completion and tech completion uses an RNG number, when the governor recommends what you should build/research next.
7. Popping goody huts
8. Every civ you have contact with uses an RNG number to decide whether or not to make a trade with you, declare war, demand tribute, etc.
9. If you see a civ, but haven't contacted them, they may use up an RNG number to decide whether or not to contact you (I have had civs ignore me).
10. Suicide galleys

I'm sure there are more....
But you can get the general idea that for 2 games to become identical, it would require alot of attention to every single detail.

 

[Qitai]

Originally posted by RowAndLive
Based on a review of the saved minimaps presented for GOTM-19. It is readily apparent for all of the AI civs that the cities were developed on the same sites in each of the individual games played. Further, from comparison of the maps, you can roughly determine what order they seem to have been founded in relative to each other. Is this consistently true? Assumably it has to do with the random # generator seed being the same for the saved GOTM file, as well as a set of hard coded rules that the AI operates under.

Some of the game mechanics are pretty much deterministic whereas others will be heavily influence by the random number generator(RNG). The city placement is one which is not really affected by the RNG. AI generally places the city at about equal intervals and it is highly suspected (if not confirmed) that it KNOWS where all the strategic resources are right from the beginning of the game and will always seek to settle in those spots. The deterministic mechanics has been well use by many players. For example, once you play long enough, you pretty much know what the AI is going to research. Others like the exploration seems pretty much random (determine by RNG).

 

[ControlFreak]

Originally posted by Qitai

Others like the exploration seems pretty much random (determine by RNG).

The mechanistic side of exploration is that the AI will not explore in areas it has mapped. Some have said you can hurt an AI by selling him a map of the other civs because he won't walk a unit overthere to meet them.

 

[DaviddesJ]

Originally posted by cracker

The QSC scoring impact is minor but intended to at least get your attention.

But the QSC scoring doesn't have anything to do with the real use of the slider. It just rewards people who play the whole game up to 1000BC in whatever way they normally do, and then move the slider to 100% entertainment just before they submit the QSC save file.

The early impact in the game and Jason scores can be significant and uses a cummulative average across all turns so you have to begin to pay attention to this feature early in the game to get the most benefit.

The in-game score is based on the total number of people, not weighted in any way, so the contribution from the early turns is insignificant. Raising entertainment higher in order to produce more happy people just to increase your score (without affecting civil disorder or WLTKD) is completely counterproductive; you're better off with a lower entertainment level and more income which you can leverage into other advantages. (Until you have a relatively dominant position, and don't need the money, and also have colonized most of your land.)

 

[ltcoljt]

Originally posted by DaviddesJ


The in-game score is based on the total number of people, not weighted in any way, so the contribution from the early turns is insignificant. Raising entertainment higher in order to produce more happy people just to increase your score (without affecting civil disorder or WLTKD) is completely counterproductive; you're better off with a lower entertainment level and more income which you can leverage into other advantages. (Until you have a relatively dominant position, and don't need the money, and also have colonized most of your land.)

Is this correct? If the scores are averaged it would seem that it would indeed matter. Can someone provide a mathematical example that shows more clearly how this works.

 

[DaviddesJ]

Originally posted by ltcoljt
Is this correct? If the scores are averaged it would seem that it would indeed matter. Can someone provide a mathematical example that shows more clearly how this works.

Sure. Every person you make happy, instead of content, is worth 1 point that turn, which means it increases your average for the whole game by 1/540 point.

So suppose you have 40 citizens in 1000BC. You use the slider to make an extra 10 happy rather than content, which costs you 12 gold. (Slightly more than the number of happy faces you gain, because as you increase the slider some small cities may end up "wasting" happy faces.) So spending 12 gold gains you 10/540 = 0.019 points to your endgame score (which the Jason score is supposed to track), or 0.0015 points per gold spent.

Suppose you did this for many turns, and spent 400 gold. Then you'd increase your score at the end of the game by a trivial 0.6 points. Don't you think that spending 400 gold on research or trading or improvements, in the early era, would let you increase the strength of your position significantly, and gain, in the long run, significantly more than 0.6 points in your endgame score?

 

[RowAndLive]

I am in PTW, so I'll have to try playing with the RNG, and working through the same game a few times, and watching how they develop on the macro level - tedious, but science advances slowly sometimes, even when trying to discover man-made laws of software. I pretty well know the AI pattern on tech development, but I just wish they wouldn't be ahead of me more often than not... ;-) And concerning unit moves for exploring, again I'm thinking on the macro level - toward the equator first, toward the rivers and coasts first, etc. that would then lead them to build cities in the same order from game start. This would then imply that they know where the resources are before they've even explored the fog - meaning that their exploration,while perhaps RNG controlled on a tile level, is not random on a macro level. They explore macro areas of fog in a given order so as to build cities in certain order - "exploration" tile exposure for them at that point becomes just a cosmetic / map trading control, and limit to how fast they can develop certain city sites. This could be useful if you trade maps frequently, allowing you to get a better idea of what benefits lie "beneath the tiles" beofre they are exposed. That would also explain the great importance that the AI places on maps. Similarily, it will trade ROP and maps for different prices, paying more or asking more based on the relative strength in tiles. So again, it knows how many tiles I have that it hasn't seen vs how many tiles they have that I haven't seen, and it demands or pays more accordingly.

 

[Ribannah]

Originally posted by Bamspeedy
As far as I know RNG numbers are used up when:
1. You founded a city
2. You trade techs with the AI (they use a RNG number to pick their next tech to research)
3. Worker automation, and unit 'auto-explore'
4. Battles
5. Culture flips. Not sure how often this is checked. I know it would be checked every turn in trouble spots, but don't know if a city with 0% chance of culture flipping would use up an RNG number or not.
6. Every unit completion and tech completion uses an RNG number, when the governor recommends what you should build/research next.
7. Popping goody huts
8. Every civ you have contact with uses an RNG number to decide whether or not to make a trade with you, declare war, demand tribute, etc.
9. If you see a civ, but haven't contacted them, they may use up an RNG number to decide whether or not to contact you (I have had civs ignore me).
10. Suicide galleys

I'm sure there are more....

Some possibles:

11. Trading a contact
12. The AI trades for contact with you (or not)
13. You build an embassy
14. The AI builds an embassy (or not)
15. Disbanding a unit
16. Abandoning a city
17. Razing a city
18. Trading a Worker
19. Capturing a unit
20. Disease
21. Privateers in AI territory / sight
22. Birth of Barbarians
23. Barbarians entering AI territory / sight
24. Barbarians getting killed
25. The AI goes into Anarchy (or not)
26. The AI chooses a new Government
27. The AI declares war / signs a treaty
28. Rebellion
29. Nuclear Power Plants
30. The AI does Espionage
31. The AI uses a Leader
32. The AI starts /switches to a Wonder

Things that seem to use different rules: doing espionage yourself, generating leaders.

 

[rabies]

The RNG is very hard to predict.
I have spent a couple hours trying to re-create some of the top QSC timlines (sir pleb, Creepster, Ribbanah)....and though some of them are very detailed, and I follow them to the letter, I don't get the exact same results. Often, this results in contacts made 1 turn later, or settlers coming out one turn later...In Sir Plebs case, I simply don't understand how he did not have massive barbarian problems up to the 2030 mark..by that time, he only had 3 warriors...and in my mirror game, 2 of those warriors were sacked by barbs..then pressed onto cities that caused production fall-backs or loss of gold (not allowing me to continue deficit tech spending). Yet in his notes, he does not even mention a barb enconter till many turns later! It is kind of annoying that you cannot recreate somebody elses game to learn...

 

[SirPleb]

Originally posted by rabies
It is kind of annoying that you cannot recreate somebody elses game

It sure is! Some discussion near the end of the GOTM19 spoiler1 thread got me trying to recreate my timeline and I couldn't do it. I gave up after realizing that discussions with other Civs use the RNG.

In trying to replay it I found that the barbarian behavior (seen at 3500 around Entremont) and Celt exploring behavior could go in numerous different ways.

It sure would be nice if Firaxis could change it to use a separate RNG seed for all player-initiated non-move/combat related actions. By changing a few of those (diplomacy dialogs and suggested next builds/research for the human only) to use a separate seed, timelines would become reproduceable I think. Though it would matter a lot to us I guess it is too much to hope for, reproduceable timelines are probably not a top priority for them