Why granaries don't work

[realpolitic]

Granaries are almost never worth it. Here's a comparison of choices in the early game:

#1
Cost of granaries - 60 shields 1 gold per turn
with +2 food city grows in 5 turns

#2
Cost of new city- settler + spearman 50 shields
3 free units (after spearman)
With +2 food 2 cities on average grow in 5 turns,
less unhappiness than #1

#3
3 archers or 2 swordsmen or 2 horsemen = 60 shields
should be able to conquer 1 city.
Increases the pop of empire
4 free units.

[Bluefrog]

While granaries aren't at the top of my build list, I think I value them slightly more than you do. In the extremely early game, I'll ignore them completely, but they can be handy once you have 4-5 cities under your belt. Drop one in a city that produces 4-5 shields regularly and has a bonus food tile, and use that city as a settler factory. I'm at work, trying to whittle away the minutes, so I don't have any hard numbers handy, but I've had a size 3 city, with a granary and the conditions set above, that could pop out a settler every six or so turns, when it hit pop 5. Have another city pumping out a steady stream of spearmen escorts, and this gives a massive, consistent boon to your REX phase.

[DaveMcW]

Check out nbarclay's granary analysis.

[Tuberski]

Granarie, to me, are a luxury, if I am going to use a settler factory, then yes it needs one, but otherwise, I'll try for the Pyramids or take it over after another Civ builds it.

ACK!

[Zachriel]

Originally posted by realpolitic
Granaries are almost never worth it. Here's a comparison of choices in the early game:

Well you did leave in the "almost never" qualifier. In some positions, the Granary can make the early game. In this game, completion of the Granary in 3000 BC assured Zululand's domination of the Limpopo River basin and surrounding areas. Indeed, historians view the completion of the Zimbabwean Granary as the key accomplishment leading to eventual victory centuries later.

http://www.zachriel.com/gotm12/Exploration.htm



Addendum: As always, it depends on the position. In this case, there are no obvious villains to attack, but there was plenty of available land to settle, much of it enriched by rivers. Zimbabwe also had a food bonus. Rapid expansion was the key to this position. The rest of the game of was somewhat anti-climactic.

[realpolitic]

I stand by my anlysis of granaries in the early game. Expansion and especially conquest(when there's a sucker near by) are much better. However, when there's no decent space left, and you're at peace (eigther because you're waiting for a treaty to end or your opponent is ready for you or you need culture to prevent flips), then granaries become a possibility, especially as worker factories to build and move population later.

[Mazarin]

it's all about how much space you have to expand If the next enemy is just 10 tiles away, build barracks, military units (maybe one settler)...and take over the cities he has built with archers...if you have plenty space: build granaries in your first cities...they will pay off!

[Datajack Franit]

I build granaries only by getting the Pyramids-otherwise I just skip them and wait middle ages

[Franses]

I hardly ever build a granary either. Their are some special circumstances that makes me decide differently but that's about once every 10 games or so.

[realpolitic]

Zachariel - here's the comparison for your city --

Capitol with granary -

60 shields cost 1 gold/turn
expansion in 4 turns = +.25 pop/turn

Capitol with daughter city (and spearman)
50 shields

capitol expands in 7 turns =.14 pop/turn
daughter expands in 10 turns=.1 pop/turn
-----------------
.24 pop/turn

Not much difference! Except you get less territory

It's also possible to get such fertile land on one city it grows out of control!

The Zulu have a lousy UU, and expansionism isn't so hot -- the price of all those techs crash when they are traded for maps, (the more civs have a tech the lower the price)so it's best to buy less important ancient techs then instead of being expantionist


Datajack - to get the Pyramids you have to start with just a handful of cities and you don't get anything until you've built the equivalent of 7 granaries - it may sound great but it doesn't work. Better to have 1 AI build it and the others waste their shields.

[DaveMcW]

Since I know how much you like math, realpolitic...

A settler requires 30 shields and 40 food. It only requires 20 food in a city with a granary. So let's see how fast we can get them to expand our empire.

Size 3 city with granary working 3 irrigated plains:
+2 food/turn = settler in 10 turns = 0.1 settlers/turn
+4 shields/turn = settler in 7 turns = 0.14 settlers/turn
0.1 settlers/turn (limited by food)

2 size 1 cities with no granary working irrigated plains:
+2 food/turn = settler in 20 turns = 0.05 settlers/turn
+2 shields/turn = settler in 15 turns = 0.07 settlers/turn
0.05 settlers/turn (limited by food)
0.05 * 2 = 0.1 settlers/turn

Tie.



Size 3 city with granary working 3 irrigated plains, including 1 wheat:
+3 food/turn = settler in 8 turns
+4 shields/turn = settler in 7 turns
0.13 settlers/turn (limited by food)

2 size 1 cities again, this time one has the wheat:
+3 food/turn = settler in 14 turns = 0.07 settlers/turn
+2 shields/turn = settler in 15 turns = 0.07 settlers/turn
0.07 settlers/turn (balanced food and shields)
0.07 + 0.05 (from before) = 0.12 settlers/turn

Tiny advantage for granary.



Size 3 city with granary working 3 irrigated plains, including 2 wheats:
+4 food/turn = settler in 6 turns
+4 shields/turn = settler in 7 turns
0.14 settlers/turn (limited by shields)

2 size 1 cities, each with 1 irrigated plains wheat:
0.07 * 2 = 0.14 settlers/turn

Tie, but the granary could do better by increasing population and therefore shield output.



Size 5 city with granary working 5 irrigated plains, including 2 wheats:
+4 food/turn = settler in 6 turns = 0.17 settlers/turn
+6 shields/turn = settler in 5 turns = 0.2 settlers/turn
0.17 settlers/turn (limited by food)

2 size 2 cities, each with 1 irrigated plains wheat and 1 irrigated plains:
+3 food/turn = settler in 14 turns = 0.07 settlers/turn
+3 shields/turn = settler in 10 turns = 0.1 settlers/turn
0.07 settlers/turn (limited by food)
0.07 * 2 = 0.14 settlers/turn

Granary wins.



Size 5 city with granary working 5 irrigated plains, including 3 wheats:
+5 food/turn = settler in 4 turns = 0.25 settlers/turn
+6 shields/turn = settler in 5 turns = 0.2 settlers/turn
0.2 settlers/turn (limited by shields)

2 size 2 cities again, this time one has two wheats:
+4 food/turn = settler in 10 turns = 0.1 settlers/turn
+3 shields/turn = settler in 10 turns = 0.1 settlers/turn
0.1 settlers/turn (balanced food and shields)
0.1 + 0.07 (from before) = 0.17 settlers/turn

Granary wins.



Size 3 city with granary working 3 irrigated plains cattle:
+5 food/turn = settler in 4 turns = 0.25 settlers/turn
+7.5 shields/turn = settler in 4 turns = 0.25 settlers/turn
0.25 settlers/turn (balanced shields and food)

2 size 1 cities working irrigated plains cattle:
+3 food/turn = settler in 14 turns = 0.07 settlers/turn
+3 shields/turn = settler in 10 turns = 0.1 settlers/turn
0.07 settlers/turn (limited by food)
0.07 * 2 = 0.14 settlers/turn

You should be shot if you don't build a granary here.



In conclusion, the effectiveness of a granary is based on the amount of bonus food you have. I always try to start my first or second city around bonus food so I can build one.

[Mazarin]

Intresting post, DaveMcW...I'm just not sure whether it is enough to reduce your observations on how fast you build settlers: e.g. on narrow maps (especially on deity) I often don't have enough room to build more than 3 cities...(I also don't want to ) so while building the granary, my enemies will take away all the land and I'll be reduced to an OCC. In this case it is important to build my two first settlers ASAP -and then to concentrate on military...no time for granaries. It's completely different on maps with plenty of land: although your second and third cities will come later your territory will be bigger when you have finished your REXing phase. In this case, I prefer building two or three granaries. So the most important reason for me is what my map looks like -not how much bonus food I have

[DuncanK]

There are two more values for granaries; Economic and Production. A city with a granary that also produces settlers and workers makes more money and produces more. That is because its population is replenished faster. Still I only build them in the early game if I don't have any super-food cities. I tend to rely on my super-food cities to produce settlers and workers and building a granary in these cities would be overkill. I would say that I only build a granary in the early game maybe 10% of the time.

[Bamspeedy]

realpolitic- You are looking at this from a one dimensional point of view. Sure, if you have a choice of only 1 city with a granary or two cities without a granary for the whole game, then yes taking the 2 cities is a no-brainer.

But, why can't you have a granary AND build more cities?

I'm sure you read my response in your similar post over at CFC, but I'll post my study to benefit more people. If you compare two cities (one city with a granary, and one without), by the time the city without a granary reaches size 12, the other city (with the granary) had already been at size 12 for 75 turns! And has produced 400 more shields than the granary-less city.

This will amount to even more shields if you have the city building any settlers/workers, because the city with granary can bounce back up in population faster than the granaryless city (or be able to produce twice as many settlers/workers, depending on how you look at it).

If you don't build a granary in a city until it reaches size 7, then you already lost out on at least 150 shields.

And not to mention the economic aspect that DuncanK pointed out.

I will agree that in some situations, building a granary may not be feasible (high corrupt cities, being pressured by war from other civs, culture needs are more important, etc.), but granaries will pay off more in the long run and this makes the Pyramids so valuable (if you start on a decent sized continent).

If playing tiny-small maps or a very crowded map where you'll only have room for 4-5 cities, then settlers before the granary will be smarter. But on less crowded maps and larger maps, building the granary first will pay off much more. The granary first tactic usually doesn't start outexpanding the settlers only tactic until 2000 B.C., so it depends on whether or not everyone will still be in the expansion phase at 2000 B.C. And after your done having your capital spit out settlers, your population bounces back up in size so quickly to build wonders.

[Zachriel]

Originally posted by realpolitic
The Zulu have a lousy UU, and expansionism isn't so hot -- the price of all those techs crash when they are traded for maps, (the more civs have a tech the lower the price)so it's best to buy less important ancient techs then instead of being expantionist

Zimbabwe hits pop 4 in 2800 BC and then produces Settlers every 5 turns or so. Though I am admittedly not a "Civ3 numbers expert," DaveMcW's cogent analysis matches my own gaming experience.

By the way, the expansionist trait worked wonders in this position. With the extra cash and techs from exploration, with cash from the sale of maps, with the advantages of the Granary, and most importantly, with knowledge of the local terrain, Zululand's expansion was rapid and decisive.

(I do agree that there are times when delaying the Granary is a good idea, especially on smaller maps, or when there are good tactical reasons for producing an early Settler.)

[Zachriel]

Originally posted by DaveMcW
You should be shot if you don't build a granary here.

We prefer a tribal council (courtmartial) first, then an order for the convicted to walk off a cliff to an honorable death. But that's just us.

[joncnunn]

I usually wait to build Granaries until I have Hospitals. Most cities grow fast enoguh without them on the Emperor level.

With regard to that particlar start, if ever there was a calling postiion that called out for a Granery as the first city build as an expansiost civ, that is it.

[peterfharris]

I build a granary as soon as possible. I want that population to go back up as fast as possible because I want high production to build other stuff such as temple, barracks, veteran spearmen and archers. There is so much to do at the beginning and so little to do it with.

There is no way in hell I want my city population oscillating between 3 and 1. I want veteran spearmen accompanying my settlers and I want some archers in case of early war. I want that temple very early. I want to able to produce trips quickly if I get into a war. (Umh, yes I want everything). A granary makes all that easier for me by boosting population and thus productivity. I am usually the last to found my second city but the extra production makes up for that.

This works for me at Regent even on a tiny crowded map. (If the map is that crowded then the granary makes the capital grow faster so I can pump out more troops to get someone elses city).

[Charis]

I'll start by saying that I never appreciated granaries and never built them before playing in a succession game with Sirian, where their virtues became *quite* evident.

Since there seem to be some math lovers here, I wanted to toss out an idea that might make sense. Civ 3 is all about growth. Folks on both side of this argument don't argue about that! Those who never use granaries avoid them so they don't slow down their 'growth', while those that use them do so precisely to speed up their 'growth'



The difference is the time frame they're looking at.
Consider this graph:



This is a plot of generic power vs the number of turns.
(It's not meant to give specific numbers, but show a concept!)
Going with a granary is definitely slower, as you delay getting more cities, no units, as you build the granary. But after that you can crank out settlers or grow in pop at twice the rate. So the curve illustrates a higher exponential growth coefficient but with a time delay added in.

In the early stages ('B') the granary approach is far slower, while after a long time ('C') the extra growth from having the granary blows away not using them. Somewhere in the middle is the break-even point, 'A'.
Those centered on short-term, early war, first few cities think of their civs like the curves at 'B' and think 'granaries suck'.
Granary lovers think in terms of long term expansion, look at not only the difference in y-value but slope at point 'C' and think "Granaries? Only a fool would neglect them!"

The reason why there is *NOT* a universal answer to 'granary, yes or no' is because it depends on many factors:
- In games with very early war the short term unit count rules, and the advantage at 'B' is quite necessary
- In small games where you only have room for a few cities, the expansion is over before you ever get to 'C', perhaps before you get to 'A'. Getting settlers out quickly is paramount, forsaking the granary
- On a large map with no neighbor close it's hard to imagine not choosing a granary
- In some cases you're not food limited but shield limited, and a granary does not double the growth rate but does keep the delay, and is a poor choice
- If you can capture a builder neighbor who builds the pyramids, you get to follow the no-granary curve at the start then get the granary-curve's slope when you capture the pyramid, for the best of both worlds.

So the question you need to ask yourself is:
Am I going to run out of space to expand, or get in a war with a neighbor, before or after 'A'. Too bad you don't know where 'A' is in a real game. But experience will be a good guide -- *IF* you make the effort to try out and use granaries if that's something you never do now. Otherwise you're stuck on the solid curve forever - good players will have a feel for both curves and know which curve to choose for their game depending on objectives, neighbors, and map conditions.

This curve is the same one of interest when businesses must choose between heavy R&D spending or going for short-term profits. Which gets chosen too often? Let's say that many execs never build granaries!

I hope this helps, rather than induce folks to...
Charis

[Arrian]

To boil it down to bare essentials, it all comes down to how much room you have to expand. If you have lots of room, build granaries - they're worth it. If you're relatively cramped and expect to start fighting pretty soon, getting a couple of cities down and barracks built is the top priority.

-Arrian

[ducki]

If you are Expansionist and follow Aeson's Scout-Scout-Scout early build queue, I'd imagine an early granary to be a great idea.

If you delay your second and third cities, you should be below the AI average, making you eligible to gain settlers and towns from goody huts while at the same time, gaining the granary-growth factor.

Yes, it's a gamble, but so is the Expansionist trait in general. For non-Exp civs, I usually play it by ear. If I feel I am shield-heavy, I'll knock out a granary to equalize food and shields. If I'm food-heavy, that granary is probably going to take ages to finish and I might be better off just squirting out workers and settlers once I found a productive town.


I used to hate Granaries because I didn't like micromanaging citizen moods, but the more experience I get, the easier it is to know ahead of time what to do. Granaries are just another tool, like temples or libraries or roads or horsemen.
If you need it, build it, otherwise make more troops. (That one's for you, Theseus. )

[joncnunn]

Also Granies don't do cities without fresh water access much good in the ancient era.

I also play on tiny maps, and so unless I'm in an isolated start, expansion to the bottlenecks will be in the middle of "B" when the military troops are thrown in.

In addition, difficulty level plays a factor. On the Emperor level, you don't want cities growing faster than you can build the happiness improvements.

[DaveMcW]

Originally posted by joncnunn
In addition, difficulty level plays a factor. On the Emperor level, you don't want cities growing faster than you can build the happiness improvements.

The luxury slider is a wonderful thing...

In the worst case you can create some entertainers to slow down growth and create happiness. Extra food is never bad.

[joncnunn]

Extra entertainers consume food; this bumps workers off of more productive tiles and into less productive ones.

The luxary slider is usually unaffordable in the ancient era and often inefficent because it helps the wrong cities.

[Mazarin]

joncnunn...you should use the luxury slider more Especially in the Ancient Era putting 10-30% into luxuries will help your growth and production..and it does not help cities that don't need it: a size-2town usually won't invest any money into entertainment with luxury at 10%...bigger towns will.

[Bamspeedy]

joncnunn-what Dave is saying is that the WORST that can happen is they become an entertainer. How else are you going to keep that person happy? What is the difference between a size 3 city with no entertainers and a size 4 city with 1 entertainer? It will still produce the same.

Here's a summary of a deity game I played where I built a granary early on and then was able to build a settler every 4 turns, allowing me to keep pace with the AI on deity in the land grab phase of the game.

Babylon's Deity Settlers

The surrounding terrain (outside the capital) did not offer much bonus resources (except for the wheat on floodplains to the north), so other cities could only produce a settler once every 20 turns. The capital with a granary producing +5 food/turn could produce settlers just as fast as 5 other cities (COMBINED) that are only producing +2 food/turn.

[joncnunn]

Bamspeedy, Your map had much more open territory than the maps I play on. If I stop to build a Granery, I won't get more than 4 or 5 ciites before being blocked in on my maps. (I play on tiny maps)

In the case of despotism with no bonus food tiles where a size 3 city is working a mined hill tile, if it grows to size 4 with the new citizen being an entertainer, it will strave back to size 3. To maintaign it at 4 it's production is worse with an entertainer than it was at size 3 because the worker that was working the hills has to be relocated.

Under despotism, most of my cities use warriors as military police, the 10% luxaries would also tax away gold from cities than are fine in addition to the one with the problem.

[DaveMcW]

Sorry for jumping on you joncnunn. In your original post you state that you don't use granaries to:

1. build settlers
2. speed up growth in cities without an aqueduct

So the only reason left is:

3. make cities grow to size 12 faster

You are correct that a single size 12 city will hopelessly unbalance a military powerhouse. So avoiding granaries is best in your case (tiny map, warmonger).

Everyone should state what their map conditions are to avoid conflicting signals.

[joncnunn]

I also build Graneries right after the Hospital so they grow to 18+ faster. (By then I have Temple + Catherdrial + Market Place + access to sufficent luxaries.)

I'm only a warmongler when my rapid expansion didn't give me enough high quality territory. (Most important feature for me is location of the rivers.) I play on the Emperor level.

The 3rd and 4th Settler I built in my most current game were to establish borders with the river on my side of the Irq border. (On tiny maps, 2 cities in most locations close by will create a chock point by the time both cities expand culturally.) The next few settlers were founded in an outside-in pattern.

I still had to build a lot of Jag Warriors in my current game just to keep the peace with my Irq neighbor when they had Mounted Warriors. Some of the Jag Warriors also carefully blocked the AI from sneaking a settler thru my territory. It was a success, the Irq didn't attack me until I had Infentry when Tanks were still a long way off.

Should the Irq make the same mistake now (I just now entered the modern era), they will be exiled to a small island.

In my previous game as the Chinese also on Emperor level I never went to war with the neighboring Japense at all. (5 cities on the river during REX, 8 cities with Harbors. Total of 10 cities, which is sufficent for the FP on tiny maps.)

During my Emperor level game with France though, I fought 2 ancient era and 1 mop up early middle age wars with England to grab the whole landmass. (Closest river was in London and continued in the far side of their starting location from mine.)

[Datajack Franit]

My capital produces about 3 settlers, then around 1400 BC I start building the Pyramids- if nothing happens I have double production and can start building the GL- if someone builds it before me it's ok as long I researched literature so I can skip to the GL