Why granaries don't work

[DuncanK]

Maybe this only aplies to industrial civs. I haven't tried it with other civs. With the French I don't have to trade early settlers for and early granary, although the settler may be delayed slightly. I build the granary right after the last settler (first one built). I wouldn't be able to build another settler right away anyway because my population wouldn't be sufficient. So I will have to trade some military probably. Probably some warriors to explore with. So maybe in some cases this agruement is about granaries vs. exploration or granaries vs. war not granaries vs. settler production.

[punkbass2000]

I'm thinking the Americans would probably be an excellent civ to go the granary route with... perhaps that shall be my next game.

[DuncanK]

Originally posted by punkbass2000
I'm thinking the Americans would probably be an excellent civ to go the granary route with... perhaps that shall be my next game.

I think they might be the perfect granary civ especially if you have low food production. I always get my as kicked as the americans for some reason though. I think I don't build up my military up enough with them.

[Bamspeedy]

Originally posted by DuncanK
Maybe this only aplies to industrial civs. I haven't tried it with other civs. With the French I don't have to trade early settlers for and early granary, although the settler may be delayed slightly. I build the granary right after the last settler (first one built). I wouldn't be able to build another settler right away anyway because my population wouldn't be sufficient. So I will have to trade some military probably. Probably some warriors to explore with. So maybe in some cases this agruement is about granaries vs. exploration or granaries vs. war not granaries vs. settler production.

Yes, in some situations, you may not have to trade any settlers (the settler might just get delayed a few turns). This is where you have just +2 food and working 2-shield tiles (game on forest, mined bonus grassland). In a city like this, you have the production, but just don't have the food to grow fast enough. I've built the granary before any warriors or settlers before, but the lack of exploration usually hurts you. So building another scout (or 2 warriors) before the granary is where the settler vs. granary debate comes in. Also, when you have a cattle or wheat, you now have the food, but not the shields, making the settler vs. granary debate stronger because you would be hitting size 3 sooner.

[DuncanK]

Originally posted by Bamspeedy


I've built the granary before any warriors or settlers before, but the lack of exploration usually hurts you.

Yeah, I agree if you are playing an exploration civ like America. With an industrial civ that is not an exploration civ, like the French, I think you still build the granary.

[Bamspeedy]

Well, the French should still build some exploring warriors before the granary. You don't want to throw your settlers into the darkness, not knowing where to go. By exploration, I mean finding new city sites and establishing contact with the AI (which will help your tech pace), not necessarily just grabbing goody huts.

[DuncanK]

Originally posted by Bamspeedy
Well, the French should still build some exploring warriors before the granary. You don't want to throw your settlers into the darkness, not knowing where to go. By exploration, I mean finding new city sites and establishing contact with the AI (which will help your tech pace), not necessarily just grabbing goody huts.

Yeah, with the French warriors and walls are all you can start with. So yes you build warriors first. Then when you have enough population for a settler you build a settler. If you don't have a 3-food square, then you build a granary.

[Dominae]

Americans are the only civ that can efficiently (industriously!) chop Forest to build a Granary right from the start (I discuss this in the Americans thread). Although you'll probably want to build a couple of Scouts before you start on your Granary, you'll still be way ahead of the competition (so to speak; the AI gets free units). Nothing beats the Americans for early-game expansion.


Dominae

[kmill25]

The only way I build grainaries early is through the Pryamids. Other than that, I'll wait until I have nothing better to do with the shields.

[HaVoC]

Building Granaries without consideration can often be harmful, since uncontrolled growth can lead to your cities revolting every few turns.
I think there are four reasons for when to build a granary.

1. You have access to lots of luxury goods so that growth will be no problem.

2. A city has a very limited food supply.

3. You have a city in an uncolonized area and want to use it to produce lots of settlers / workers.

4. Your city is already large and grows slowly.

Otherwise granaries aren't useful. It's a case-by-case desition.

[realpolitic]

I've found a more compeling reason not to build granaries now that I'm playing at Diety level, I'd get crowded out. Now I just build 2 cities, so I have 8 free units. The quick gain in shields with newly conquered cities is faster than the long term gain in food, I assume it takes about 60 shields to conquer 30 more than to settle.

It's good to see that the discussion here has some mathematical component here. If I seem a bit fanatical I apologize, I had been helping some government institutions deal with Califoria's budget cuts with some simple arithmetic, and was greeted very favorably, until it conflicted with a manager's cherished ideas, I got called every name in the book, and got thrown our on my ear. I'm glad this discussion is mathematical, it's weird to see grey haired men insulting each other like they were in third grade. In the 21st Century there are lots of people with Masters Degrees who don't understand the value of arithmetic. Thanks to those of you who showed what happens in the long term, but I think at that point, conquest is best.

[Mazarin]

well, maybe a mathematical approach would be best -assuming you want to consider map size, number of opponents, terrain and many other factors...but as this is just a game we should rely on our expieriences and conclude that granaries do pay off in many cases and are not worth the investment in many others.

[DuncanK]

I tested this out. I started a new game with the Americans. I made the world large, made 9 AIs, roving tribes, and a normal climate etc..., Monarchy level.

I saved the game on the first turn. Then I played the game twice until 1790 B.C..

On the first game I built 2 granaries in my first two cities. I used the clear forest strategy for both granaries. I built 1 scout before the first granary and 1 warrior before the second granary.

On the second game I did not build any granaries. I built 2 more scouts and I would say at least 5 warriors in place of the granaries.

Here's the run down on both games at 1790 B.C.:

Game 1 (w/2 granaries):

10 cities
4 settlers
6 workers
Population 386,000

Game 2 (without granaries):

8 cities
5 settlers
5 workers
Population 260,000

Notes:

1) 2 of the settlers in game 2 were going to build cities on the very next turn.

2) I had tech parity on both games (all the early military tech and writing).

3) I had a bigger military in game 2 (mostly warriors, but some spearmen, archers and horsemen)

4) I had a couple hundered more bucks in game 1 (assumably because of the smaller military throughout the game)

Analysis:

I think I did better in game one and therefore I'm now convinced that granaies are the way to go with these settings.

Bigger is Better. I stopped the game at the end of my expansion stage in game 1. This is where I would have built up my military and conquered an AI in order to get more land area. Even though my military was bigger in game 2 in 1790 B.C., my production was greater in game 1. Also, I was more towards my miltary production stage and at the end of my expansion stage in game 1 than in game 2. In short, I was farther ahead in my stategy in game 1.

The large early miltary (trade of for the granaries) tended to be a drag, more than a benefit. I still was not able to get a lot of barbarian villages with my warriors, because the AIs got a lot of them.

The extra scouts in game 2 were not able to get too much more huts if any.

Realpolitic,

As someone trained in Economics I know that mathmatical analysis has limited potential in many areas. I usually ignore mathematical analysis when it comes to these games, because it usually doesn't hold up in actual game play. We should all rely on our game experience instead of crunching numbers. I believe the numbers will lie to us more often than not.

[Bamspeedy]

Double post-
See below

[Bamspeedy]

Havoc-
If you use the governors or check your city every time it grows in size you won't have any problems with revolts.

1. Like has been said before- The luxury slider is a wonderful thing. On deity, you will be running 0% science anyways, so cash is not a problem. A size 6 city running as much as 50% luxury still has more money to devote to science/tax than a size 3 city with 20% luxury. So using the luxury rate early in the game is not a problem. After you get the granary built and other cities can build workers to hook up luxuries and build military police to send to the capital, then you can lower luxury rate alot.

2. Well, if the city has high-food it can still benefit from a granary, but yes it may not be quite as much worth it (it terms of # of turns it takes to build a settler).

If you are producing +5 food the fastest you can build settlers is 8 turns without a granary and 4 turns with a granary.
At +4 food it can build a settler every 6 turns with a granary and 10 turns without a granary.
At +3 food it can build a settler every 8 turns with a granary and 14 turns without a granary.
At +2 food it can build a settler every 10 turns with a granary and 20 turns without a granary.

If the capital is producing +4 or +5 food/turn and have a granary, that city can usually have a build order of settler, settler, settler, settler.... All your other cities can build nothing but military. Without a granary your capital will usually have to build something else in between some of those settlers because your population drops so much and you need to let your capital grow again.

3. Any city that will produce alot of settlers and workers should build a granary.

4. You're right if there are still tiles for it to use. If the city is already maxed out at size 12, then don't bother to build a granary until just before you build a hospital.

In DuncanK's test, that was on a large map, which is ideal for a granary. And on his point #4, he also had more income because of the faster growth and his granary cities being a larger size.

Realpolitic plays tiny maps, so a granary is not the best move there.

I often hear the date of 2000 B.C. being thrown around as the point where granaries start pulling ahead of when you don't build a granary. On tiny maps the expansion phase would be over well before 2000 B.C. On larger maps you'll still be expanding at 1000 B.C., sometimes later than this.

[vulture]

Some fun facts and figures with granaries. I had a look at some simplified models, just to see what kinds of factors are important.

First time around, I assumed that all cities get to work tiles producing 2 food, 1 shield, 1 trade. One scenario has the first city building a granary, the other has the city doing the size 1-3 settler oscillation.

The city without the granary is simple. It has a 20 turn cycle, producing 20 shields worth of units (2 warriors, 1 spearman or 1 archer, depending on your techs), 1 settler and 50 trade. So after 60 turns you have accumulated 60 shield of units, 3 settlers, and 150 trade.

With the granary, I decided to produce 3 warriors to start with (that 3rd one only delays granary production by 3 turns in practice). I then oscillate between sizes 3-5 for settler production (you could get a second settler out more quickly, but this is just illustrative). Settler no. 3 appears in turn 61, only 1 turn later than in the no-granary situation, at which point you have also prouced 100 shields of units (in batches of 30, 20, 20, 10, 20). In the first 60 turns, you produce 210 trade (and that's after allowing for trade lost to luxury spending, assuming you have no luxury resources to hand).

Second scenario: assume we have 1 shielded grassland available, producing 2/2/1, and all other sqaures are 2/1/1 as per the original scenario.

Without the granary, the situation is the same, but you get to produce more shields for units. Ironically, some of your settlers may turn up one turn later, depending on how you want to waste shields (either going to wealth for a turn or two when you can't fit anything in before the settler build begins, or fitting a warrior in at the expense of delaying the setller by one turn). Settler 3 still comes along in turn 60 (or 61, possibly 62, depending on how you play it), but you have 100 shields worth of units (in blocks of 40, 30, 30, so a religious civ could fit a temple build in there as well).

With the granary, settler number 3 appears in turn 52, but with only 70 shields of units (30 at the start for warriors to explore, then 20, 20). In 60 turns you would generate 195 trade and 90 shields.

Results part 1: Using a single city as a settler pump, with 2 excess food at all times, you are equal or ahead with settler generation by the 3rd settler, and have considerably more trade (gold or science), even if you have to use the luxury slider to control happiness. The amount of shields that can be spent on units is comparable. So you 'break even' in settler production in 50-60 turns.

On the other hand, you lose out in terms of production from the first new city, by 20 turns in the first scenario, or nearly 10 turns in the second. That's quite a lot of shield production you are giving up. In particular, if you use all cities to churn out settlers as fast as possible, (assuming no granaries in the subsequent cities), then you don't catch up in terms of number of cities until you get to 15-20 (depending on how long you assume it takes a settler to move and found a new city). The REX phase is usually over before you get to that many cities (in my experience, you are doing very well to get 10 cities down without warfare on a large map, Emperor level). Of course that kind of rapid expansion leaves you over exposed to AI invasion, since you don't have resources to spare for much else aside from a few warriors, 1 spearman per city and lots of settlers.

So, my quick look suggests that the effectiveness of granaries depends to some extent on the excess food to shield ratio. More shields and less food makes granaries more effective. Fewer shields and plentiful food make them comparatively useless in the REX part of the game (since you are limited by the number of shields, not by the amount of food). If you are going for an all out (badly defended) REX then granaries don't help. If you are using one city as a settler pump, then they may, but the most natural settler pumps (high food) don't need them. If you have to use a +2 food city as a settler pump, then it is probably best to build a granary. If you can afford to rush buy a granary, all the better; you get the best of both worlds. It is rather situation dependent though, so get out and explore the world before committing yourself.

Granaries, of course, are useful after the REX phase too... (and don't forget, the *only* time you can grow too fast is when your city growth is outstripping your ability to develop squares (I've overlooked worker production in this analysis); as long as all your citizens can work developed tiles, the worst you do by having to use entertainers/luxury slider to control happiness is break even, and most of the time you gain some benefit (when compared to keeping your cities below the happiness limit)).

[Ribannah]

Whether to start with a Granary or not indeed also depends on whether your worker(s) can keep up with improving the terrain. If the extra citizens are working unimproved tiles, you are probably better off by starting with a settler (or, like I often do, with a second Worker) and take another look after that. If you have a game or a whales tile, to start with Granary may be an excellent idea.

[DuncanK]

Does everyone participating in this thread agree that only industrial civs should build a granary in the very early game? Some people think that only the Americans are worthy, and able, of starting with a granary. I think that only the Americans should build a granary before their first settler, considering that the map is large enough and there is room for expansion.

[ducki]

I think anyone with the tech might be presented with a situation that calls for a granary before a settler.

[DuncanK]

Originally posted by ducki
I think anyone with the tech might be presented with a situation that calls for a granary before a settler.

Only expansionist civs start with pottery. I've done it with expansionist civs that are not industrial. You need two workers to clear the forest for the granary. I think that's costly. I don't think its worth it, because its taking too long to get that first settler out.

[Mazarin]

DuncanK,

You should also consider the advatage of settlers from goody hut: by delaying you first settler you increase the chances of getting one...since I read Aesons scouting thread, I always build the granary first (after 3 or four scouts) and I've never been disappointed

[Bamspeedy]

Depends on the tech rate. If you can get Pottery quickly (20 turns or less) by running 100% science, then the granary first will still work (use a barracks or wonder as a pre-build for the granary). If playing on a huge map where it will take over 30 turns to get pottery, then build a settler first obviously. The babylon deity game that I posted earlier I did not start with pottery and I did not build an extra worker right away.

You don't always need to clear a forest for the granary (although it is easier if you are industrious). It depends on how many shield-producing squares you have nearby.

[ducki]

Originally posted by DuncanK


Only expansionist civs start with pottery. I've done it with expansionist civs that are not industrial. You need two workers to clear the forest for the granary. I think that's costly. I don't think its worth it, because its taking too long to get that first settler out.

Also consider a non-Exp civ getting Pottery from a goody hut, which is the only reason I said it the way I did...

... I think there are plenty of circumstances that warrant a granary before a settler; however, there are so many factors to consider that listing every possible instance would be ludicrous. To name just a couple, a very low food start, a very high luxury start, a start where you get a settler from a hut, a start where you get Pottery from a hut, a huge map with fewer than standard civs.

Granted, oftentimes, a granary first is a waste, but there are also many times when it makes sense. You can't make a hard and fast ruling due to the randomness of the game and the vastness of the variables.

[Ribannah]

That is why I put a lot of time into the first turns of the game, trying to project all kinds of opening play sequences.
Sometimes it takes an hour before I make my first move!

[DuncanK]

Thanks for the reply to my post everyone. I think I will try a granary first with some other civs. Still, I'm reluctant to delay the production of my first settler. When it comes down to it 2 cities without granaries is better than 1 city with a granary. In the case of the Americans I can see how you could have 2 cities, one with a granary, where maybe another civ would just have either 2 cities (without granaries) or 1 city with a granary.

Mazarin,

I will go back and read that thread. Thanx

[Panag]

hi ,

the best is to get a second city near a good food tile asap and get a granary there , ....

and start to build settlers at a high rate

have a nice day

[vulture]

Originally posted by panag
hi ,

the best is to get a second city near a good food tile asap and get a granary there , ....

and start to build settlers at a high rate

have a nice day

I disagree. Cities with good food really don't need a granary. In my first PtW game (ongoing, mentioned elsewhere) I had the good fortune to have a great start position with lots of cows on grassland. Each city with 1 cow and 1 shielded grassland could grow 2 sizes in 14 turns, and would take about 12 turns to build a settler. In that kind of city, a granary is all but useless, since it would take many turns to produce and just knock 2 turns off the settler production time. IMHO the time to build a granary is if your best food city has only 2 excess food, but plenty of shields. You get the most benefity from a granary in a city that can produce sufficient shields for 2 settlers (or more) in the time it takes to grow 2 sizes, since you can halve your settler production time in that case. If you produce between 30 and 60 shields in the time taken to grow 2 sizes then a granary is a marginal improvement for settler production. If you can produce 30 shields or less, a granary is worse than useless.

High shields, low food, means granary is useful.

[Panag]

Originally posted by vulture


I disagree. Cities with good food really don't need a granary. In my first PtW game (ongoing, mentioned elsewhere) I had the good fortune to have a great start position with lots of cows on grassland. Each city with 1 cow and 1 shielded grassland could grow 2 sizes in 14 turns, and would take about 12 turns to build a settler. In that kind of city, a granary is all but useless, since it would take many turns to produce and just knock 2 turns off the settler production time. IMHO the time to build a granary is if your best food city has only 2 excess food, but plenty of shields. You get the most benefity from a granary in a city that can produce sufficient shields for 2 settlers (or more) in the time it takes to grow 2 sizes, since you can halve your settler production time in that case. If you produce between 30 and 60 shields in the time taken to grow 2 sizes then a granary is a marginal improvement for settler production. If you can produce 30 shields or less, a granary is worse than useless.

High shields, low food, means granary is useful.

hi ,

well a granary can help to build your settlers faster , ....

it just depends where your first couple of cities are , a scout is great to scout around and look for suitable places

have a nice day

[Bamspeedy]

Each city with 1 cow and 1 shielded grassland could grow 2 sizes in 14 turns, and would take about 12 turns to build a settler.

How do you figure that? 1 settler every 14 turns is the fastest you could build a settler (on average) if it takes 14 turns to get 2 population points.

Cow on grassland:
Irrigate it and no granary=1 settler/10 turns
Irrigate it and granary=settler/6 turns
Mine it and no granary=settler/14 turns
Mine it and granary=settler/8 turns

But in your case, where there are multiple cows for several different cities, then yes, it would be smarter to get all of those other cities set up faster, so each cow can be used right away.

If a city is your only high-food city, then a granary will still help, because the other cities that only have +2 food won't help your settler creation much, since they can only build 1 settler/20 turns. As the city gets to size 3-4 it would be picking up more shields also, speeding up the time it takes to build a granary. So it depends if your worker can keep up with the population growth or you have any bonus grassland/forests around.

I've tried the granaries a little more, and I think building 1 settler before the granary is better for the high-food city. The second city can build more scouts/warriors to help in your exploration and it lets your worker get caught up in terrain improvements for your capital.

In my current game, I have 1 cow on grassland, game on forest, and some bonus grassland tiles in my capital (non-industrious civ). Irrigating the cow, and building a granary the first settler gets delayed by 9 turns. But produces the 3rd settler sooner than if I hadn't built a granary.

Building a settler before the granary I was able to use the other city to pump out more scouts. Although, having more improved tiles for my capital to use didn't really shorten the time it took to build a granary except for maybe 1 turn. So by building settler before granary I was behind the granary first build by 1 settler, but I had so much more exploration done.

If you can produce 30 shields or less, a granary is worse than useless.

I agree on that. I imagine that floodplain cities would have a harder time coming up with the shields. But with cow on grassland starts I don't have a problem at all with having enough shields, provided there are enough bonus grassland around. To build a settler every 6 turns, you just need an average of 5 shields/turn. With an irrigated cow (+4 food), you can spend 2 turns working on a forest and 1 turn not working the forest and still get your 10 food/3 turns if you have a granary. Without a granary, you have to not work the forest, because that slows your growth (5 turns at +4 food to get your 20 food).

[vulture]

Originally posted by vultute

Each city with 1 cow and 1 shielded grassland could grow 2 sizes in 14 turns, and would take about 12 turns to build a settler

Originally posted by Bamspeedy


How do you figure that? 1 settler every 14 turns is the fastest you could build a settler (on average) if it takes 14 turns to get 2 population points.

Cow on grassland:
Irrigate it and no granary=1 settler/10 turns
Irrigate it and granary=settler/6 turns
Mine it and no granary=settler/14 turns
Mine it and granary=settler/8 turns

Possibly I explained it badly (or I'm missing your point entirely here...), but I meant that If it takes 14 turns to get enough food for a settler, and 12 turns to generate enough shields, then it takes 14 turns to build a settler. If we add a granary to the mix, then it takes 8 turns to get the food, but still 12 turns to generate enough shields, so it takes 12 turns to generate a settler. Obviously I'm ignoring the effects of mining the grasslands to generate more shields here (IIRC in the game I decided against building a granary, since I had so many good settler pumps to hand already, and so didn't need to hurry the tile improvements for the settler cities once I'd connected them up with roads).