Granary or settler?

[ecuwins]

I can illustrate a "real world" dilemma here with a couple screenshots. This one is a no-brainer, it is an extreme example but compare and contrast between the ideal and real application.

Ok, starting location, move worker, decide to settle in place. Does not look good. Map settings: Std, 80% water, pangea, civs all random, wet, warm, 4mil.

Spoiler :

Now after some exploration things change big time.

Spoiler :

A settler must be built first to settle on the hills so irrigation can be connected to the cattle on the other side. Look at all the mountains and hills, better chance of Iron and I have Immortals on a map that should be cramped .

Edit: Update!

France and Maya are close but I should be able to get at least 5 citys and crank Immortals. Yes Iron is a lynchpin here but archers and cats will do in a pinch. Not sure who the first victim is yet but time will tell. Settler #1 almost done

Spoiler :

 

[eldar]

A fascinating thread... but if it's example time, here's a couple I made earlier (both C3C 1.22):

1. No granary (at first):
Sumeria, Monarch, 470BC, Standard Continents, (60%/Warm/Wet/4 Billion?)

Notice the copious amounts of food around the starting location! I knocked out two settlers before a granary: the first to take advantage of both game forests (which I still haven't chopped, and don't plan on doing so), the second to head to the flood plains. (The whole granary question became moot the second I popped an SGL of course.)

2. Granary:
Hittites, Monarch, 330BC, Standard Pangaea, 70%, Warm, Wet, 4 Billion, Sedentary barbs

Apart from the starting site, no decent food anywhere in the first ring. Yes I now have a city sharing the cows, but to start with it was used for military. (In hindsight I maybe could've gone settler-granary, the sharing of the cows didn't really enter my head until later! Would've worked too, growing the capital on a 5-7 cycle at +4pft (1 cow) then +6fpt (both cows), giving the other city +2/+4 (for an average of +3fpt)).

 

[ThinkTank]

have you already noticed settler 6 missing in your excel sheet ?

Uhm, no, I think it's right, what do you mean?

1) I see some guys are arguing for the theoretical max expansion -- it is important to know this.
2) Others are arguing for the applied max expansion. (applied = in game, don't know map, AI is out there)
I believe it is important to understand the first in order to apply it to the second.

It is equally important to recognize how your particular game start differs from this theoretical ideal.

I agree to this picture. I think understanding the theoretically optimal scenarios helps improving practical play.
At least for me it does. There may be another kind of player that intuitively "knows" what is right without knowing why; this is normally called talent and I'm afraid I don't have that. I need to understand what I'm doing. Sorry!

 

[Marsden]

No need to be sorry, this is very helpful for me.

Also, what I'm learning sounds like on tiny to small maps the granery isn't as useful because less total cities will be needed and on larger to huge maps a granery is even better because of the huge amount of cities.

Please correct this assumption if incorrect, but that's what I've gleaned from this. I think map size is important because one certainly doesn't approach a tiny map like a huge or vice versa.

 

[WackenOpenAir]

Uhm, no, I think it's right, what do you mean?

I agree to this picture. I think understanding the theoretically optimal scenarios helps improving practical play.
At least for me it does. There may be another kind of player that intuitively "knows" what is right without knowing why; this is normally called talent and I'm afraid I don't have that. I need to understand what I'm doing. Sorry!

You have counted your settler 1 2 3 4 5 7 8 so the comparison goes wrong the builds are even at the 3th settler, not at 7-8.

Don't give up on your analythical approach to the game. I am a big fan of the analitical approach and if you get good at it, it can beat anyones intuition by miles.

Edit, i now see you may have not forgotten the settler but somehow your calculations caused 2 of them to get there together. I don't know how you got to the numbers though.

 

[ThinkTank]

You have counted your settler 1 2 3 4 5 7 8 so the comparison goes wrong the builds are even at the 3th settler, not at 7-8.

In this picture I summarize the total number of cities created. So the jump from 5 to 7 just means that between turn 65 and 70 two additional towns where founded (in the no granary scenario).

Also, what I'm learning sounds like on tiny to small maps the granery isn't as useful because less total cities will be needed and on larger to huge maps a granery is even better because of the huge amount of cities.

Please correct this assumption if incorrect, but that's what I've gleaned from this. I think map size is important because one certainly doesn't approach a tiny map like a huge or vice versa.

This is indeed where my analysis points to. But that conclusion should be provisional, because as many people in this thread have been pointing out, I made some idealizations. So it remains to be seen if that conclusion is still valid if we do account for other factors (other assumptions on food/shield ratios, commerce, building military, workers, factor in the value of the granary after REX, and so on).

 

[Ringo Kid]

Interesting. Does the cart take into account that settlers born in the center of your empire take longer to reach a new city site than those born at the edge?

And how about workers building roads to speed the settler to the new city sites.

Also another aspect is to at some point switch the core to military so you can expand by taking the AI cities away. Then you need to consider how to weigh that since you are both increasing your empire and decreasing your opponents ability to expand at the same time.

 

[WackenOpenAir]

Could you explain where you got those numbers ?

in GOTM's, QCS scores are calculated at turn 80 (1000BC). The number of cities and citizens are widely accepted to be the most important factor by then, and thus the gotm players concentrate on this number.

Depending on the map, the number of cities at turn 80 generally vary between 12 and 20. (mostly between 12 and 16, over 16 on very rich starting positions)

 

[anaxagoras]

Could you explain where you got those numbers ?

in GOTM's, QCS scores are calculated at turn 80 (1000BC). Depending on the map, the number of cities at turn 80 generally vary between 12 and 20. (mostly between 12 and 16, over 16 on very rich starting positions)

The spreadsheet used an extremely minimal starting assumption of just 2 sfpt at the opening city and at every city thereafter. That's much lower than the typical GOTM starting position will have available in at least one of the first two settlements. As a result, actual game settlement rates are likely to be much higher if they have anything like a favorable opening position.

 

[MAS]

Could you explain where you got those numbers ?

in GOTM's, QCS scores are calculated at turn 80 (1000BC). The number of cities and citizens are widely accepted to be the most important factor by then, and thus the gotm players concentrate on this number.

Depending on the map, the number of cities at turn 80 generally vary between 12 and 20. (mostly between 12 and 16, over 16 on very rich starting positions)

The XL sheet (can be opened in OpenOffice if you don't have XL.) is linked to at the bottom of the first post. I think the overview is pretty clear. (there are 2 sheets in the document)

The question was, "is it worth building a granary in capital if only 2 surplus food is available (like a sea of grass without bonuses)"

 

[RickFGS]

Granary or settler? Granary for sure and i dont think your calculations are correct, you overlook a great deal of details, choping forests for faster granies, industrial workers roading faster, granary also allows for a worker pump in capital, not just a settler one and so on, making in the long run a much quicker builder of cities.

 

[ThinkTank]

Could you explain where you got those numbers?

As MAS indicated they are in the spreadsheet, but here is an example.

This assumes only 2sfpt (surplus food per turn) cities without a granary, pumping out settlers as soon as possible,
which is every 20 turns after they come into existence. I assume a settler takes 5 turns to found a city.

The city numbers reflect not city founding order but rather generations (and, as well, the way in which the diagram is constructed). City 1 to 5 are the first generation, they are founded by settlers from the capitol (city 0). City 6-15 are second generation cities, they are founded by settlers created by first generation cities. And so on.

. . . you overlook a great deal of details, . . .

I know, as I have indicated in other replies in this thread. I want to understand a complex problem by first getting a simple version of it sorted out. Then add in other factors. So this is really just the start of the research, not the end.

 

[ThinkTank]

@ ecuwins: your example suggests that sometimes the map will pose conditions that force a decision, like the irrigation problem which forces you to a settler first. The question is whether the irrigation problem itself is sufficient for that. If in fact you had more room than for 5 cities only, say for 20, maybe a granary first still would have been the better choice? I'm inclined to think that the restricted space is the main reason why not building a granary is good here.

@ eldar: nice examples. In both cases you seem to have come out of the REX phase as the strongest tribe and I suppose you won both games. So you did well. Question is could you have done better. Both games have the capitol as a possible 4 turn settler factory. In the Sumeria game you did not use that opportunity. I think you should have; if my calculations are correct the break even point for building a granary in a +5 food town is 3 settlers, and you evidently had far more space. Do you have a 4000BC save of the Sumeria game? Then we can just try it out.

 

[eldar]

I've not completed either game - both would've resulted in very easy wins, so I probably won't go back to them.

I'm fairly sure in the Sumeria game that I did run the capital as a 4-turn factory, chances are I've switched it off by the point the game is at (there's another city about to take over the settler-producing role, and plenty of food about in other places to get workers). And there's tons of space left to expand into. If I were to continue it, I'd be looking at a SS/Diplo win.

As for the Hittite game - well the continent is as good as mine and I can probably get a regulation domination/conquest out of it.

 

[ThinkTank]

I have some addtional calculations.

I have added three cases in which the first town does not have 2sfpt but 3, 4 or 5 sfpt, and calculated the effects of having a granary first or not in the first city. All other cities except the first are still assumed to have only 2sfpt. The break even point is indicated in red; the QSC turn in green.

The general trends:

• The effectiveness of a granary rises with increasing surplus food, which has been observed before.
• The break even point for choosing a granary first decreases with increasing surplus food.
• In all cases, the break even point is so low that the choice for a granary first will be valid on the majority of maps.

What is somehow surprising is that the effect of a granary in the 3sfpt case is relatively small. I have reluctantly set the break even point at 5/6 cities at turn 55, but look at the chart, the result stays about even until 18 cities at turn 90. This can be explained by the fact that in this case the granary actually wastes a lot of food. A 3sfpt city without a granary grows 2 citizens in 14 turns wasting 2 food of 42 (4.7 %). A 3sfpt city with a granary grows 2 citizens in 8 turns wasting 4 food of 24 (16.7%).

As a next step I want to look at situations where the second city has better sfpt than the first and calculate the case for a granary in the second city rather than the first. I'll report the results here when I have them.

 

[MAS]

• In all cases, the break even point is so low that the choice for a granary first will be valid on the majority of maps.

Where do you base that claim on?

 

[ThinkTank]

If the break even point is three cities - I think on the majority of maps the room you have to expand during REX in larger than three cities.

Again, these conclusions are provisional and subject to the disclaimer that I have only been looking at settler production. However, the possibility that other factors may change these conclusions does not prove that they will.

 

[MAS]

But in your chart (2sfpt), both strategies still have 8 cities at turn 75. That is more than 3 cities.

EDIT: Also, in the no-granary scenario, how come in a 5sfpt and 4 sfpt city your settler is out at turn 25, just like in a 2sfpt city? I expect that first settler to come earlier. Just like what happened in the 3sfpt city.

 

[ThinkTank]

But in your chart (2sfpt), both strategies still have 8 cities at turn 75. That is more than 3 cities.

Yes in the 2sfpt both strategies are about equal in the range 4-8 cities. After 4, a granary first is not worse; after 8, it is better. Since a granary is also useful after REX I would in fact think that it is already better after 4 cities. What I said is that it is a "valid choice" on most maps - which I think still stands.

(Edit: when I said 3 cities in post 37 I was in fact thinking of the 5sfpt case.)

EDIT: Also, in the no-granary scenario, how come in a 5sfpt and 4 sfpt city your settler is out at turn 25, just like in a 2sfpt city? I expect that first settler to come earlier. Just like what happened in the 3sfpt city.

That is a bit strange indeed and I will check my calculations. The explanation is probably as follows. What I did is look at MM scenarios that enable you to build 3 warriors before starting on a granary or settler. For 3 sfpt I assumed a mined cow; for 4sfpt an irrigated cow; and for 5 sfpt an irrigated cow and a city center with 3 food (agri). This means that the 4sfpt and 5sfpt cases are less rich in shields.

 

[MAS]

For 3sfpt I think a cow on plains is a better scenario, as you'd normally not mine anything that can break through the despotism penalty. (the effect is the same though)