Food-Banking for Score & Time Games

[Swordnboard]

This thread is dedicated to discussing a technique I first discovered in WastinTime's thread on achieving 15M score. As it turns out, the strategy is much more relevant for score games on Normal/Quick/Epic and really shines in time games, so to allow WastinTime's thread to move on all further discussion should probably go here.

The story so far...

I was doing some tests on quick/standard, and found that it might be possible to grow a city up to about 20% higher than its actual population cap because you can grow even while starving if you have enough food overflow. Furthermore, a city with sushi corporation can build theoretically infinite overflow by whipping itself down, meaning that the 20% increase can be exceeded quite a bit. This means that, with the proper timing, one's entire empire could be operating at something like double its natural population cap or more!

Basic gameplan:

• Generate infinite overflow by growing more than an entire population (half a population with granary) per turn
• Stack this overflow up by whipping down to (at minimum) the point where all citizens are working 2 food or better tiles. Smaller is better!
• Each city should strive to maintain similar amounts of overflow and population, so that they all reach their "peak" at the same time
• When the time is right, stop whipping and allow cities to grow way past their pop cap
• End the game at this artificial peak, yielding massive score (???)

Caveats:

• Quick Speed sucks -- Is it possible to have enough food for 1-turn growths + overflow on Marathon/huge?
• Timing of granary, lighthouse, sushi execs, etc. is difficult to get right so that all cities reach peak at the same time
  • If you mess up, the cities starve back down basically 1 pop / turn. After half your cities have peaked, you gain no score by waiting longer.
• What would we whip when building up overflow? Unfortunately workers aren't an option as they halt growth.
• This strategy takes up lots of time, pushing back the finish date quite far.

I'd love for someone with more experience on score mechanics to weigh in on whether this is feasible here.

Had a bit of time to do some more analysis:

On quick, the cost of growing a population is 14 for the first, then 16, 17, 18, 20, 21, 22...
This means that, with a granary, some populations aren't even more expensive to grow than the previous! So growing to use all of your tiles with at least 3 is optimal for food-banking speed.

On normal, the cost always increases by 2 (1 with a granary), so growing to use 4 tiles is good, and growing to use 3 tiles is neutral for overflow banking. We still want to grow into 3 tiles though, since our maximum population depends on both the banked food and the starting population point

On marathon, though, things start to get hairy. Even with a granary, the cost of growing a new population increases by 3 for every population you grow! This means that, even if your city could work 4 tiles all the way up until size 20, this would grind our overflow banking to a halt (3 more to grow + 2 eaten by the citizen is less than 4 more worked). Of course in reality our cities will look even worse! So this brings us to the terrible, horrible, no-good, very bad idea of banking food at 1 population. Heck, we don't even have to work a tile (which is good since we will be whipping this hypothetical city literally every turn ).

The setup: a city with a granary and sid's sushi co. It doesn't even need any workable tiles!
Each turn, the city grows from size 1 to size 2. It gains food from the city center (2) and sushi co (variable, note this was 42 at the end of the previous space game). The single population point eats 2 food, and growing to size 2 eats 66 food. The granary then refunds us 33. This means that the total change in food banked is (2 + 42 - 2 - 33) = 9. Whip down to size 1 and repeat as long as you are willing... voila! Lots of food in the bank!

EDIT: This provides an interesting argument for the Khmer. +1 food from the baray doesn't seem like much, but when coupled with a supermarket these two lowly buildings could give a 20% increase in overflow banking per turn. This also means every food from sushi counts.

But how far will this food get us? In order to grow in starvation mode, we must have enough food in the bank to grow in a single turn. No problem, right? We've been doing that every turn so far.
Our first population grown adds to our bank, same as before (bank gains 9)
Our second population grown requires 5 more food (bank gains 4)
And our third 5 more than that (bank loses 1)

Anyways, we can calculate out how far we can go with this before expending the bank, and it looks really bad. In fact, to grow n population consecutively we need on the magnitude of n^2 stored food in the bank! And the result is not even good until we can reach significantly higher than the old population cap (including worked tiles), which might be something like 40 or more! So to double our population, we might need about 80^2 = 6400 food banked, which at our rate of 9 a turn would take a whopping 700 turns to achieve .


Conclusions / tl;dr

• Banking sizeable amounts of food is orders of magnitude harder on marathon than it is on quick, normal, or epic
• Because the strategy requires whipping down to size 1 (on marathon, at least), we can't work tiles since there is too much unhappiness
• Due to how long this whole process takes, it likely won't improve score on marathon. Still need to analyze more, but I believe the strategy could be instrumental for a score game played on quick or normal speed.
• One side effect of the strategy (insane gold generation via Q-whipping) could be part of a ludicrous new late-game economy, which might remove the need for STRIKE entirely. The downside would be the loss of caste system, but with cristo redentor we could stay in caste half the time.

note: since I'm speculating about how to use Swordnboard's new strategy in a time game, much of this does not apply to the high score game WT is planning here.
It can indeed lead to a ludicrous late-game economy. I was thinking about how to make this work on a large scale, for hundreds of turns in a time game. One would want to keep teching, which is a problem if everything is whipped down to pop 1. (The loss of caste is not much of a problem when everything is whipped down to pop 1.) With mining there is still good production, which can be used to build science. But we also have to whip every turn. Solution is simple, missionaries! (execs would be better, but it seems they can't be cold 1 popped, at least not on quick.) Have cap-1 missionaries built, then every turn every single city whips a missionary and takes it off the cue to instead build science. Capital completes the missionary to turn all other missionaries into failgold. Missionary failgold in every city could be enough to cover your expenses if courthouses and stuff are in place. On the other hand, there's really no need to stay out of STRIKE, since not even workers are needed. Though something is needed to keep remaining AI in check... The generated failgold can also be used to buy other things you might need in US.

Some other implications of this strategy:

• A city must be pop 1 or pop 2 when it receives sushi, or it will never be able to get down to pop 1. Being at pop 1 is the most efficient for building food surplus, assuming happiness is not unlimited. Edit - not true. One can whip several items the same turn.
  
• If it would be possible to 1 turn Future techs (for bonus happiness) while doing this, then it would be better to start with larger cities.
• If whipping to pop 1 every turn, then it's better to settle on top of food resources for 3 food city tile.

However, there is one problem with this... Even on a tiny map, it's possible to have over 100 cities on B&S or archi. Playing a game like this on epic speed, banking food for 400+ turns means 400 turns of whipping missionaries and taking them of the cue in 100 cities, or performing the same mechanical task 40 000 times. Don't even think about a huge marathon map where you'd do it in 500 cities for 1000 turns...

Interesting is that food overflow is not applied to settler or worker builds. When I build a settler, the city feeds itself from my overflow and the rest of the overflow remains stored up.

Also developing something that could make time games bearable. Using saved production cues and a keyboard macro the operation now takes me about 1 second/city. If there only was a keyboard shortcut for whipping, then I could even do one hundred cities in one keypress... (Edit: didn't realize I can add mouse clicks to my macros. Now I can whip mishes and take them off the cue in multiple cities with only one press of a button. )

 

[elitetroops]

I've never played a time victory (except maybe sometime unintentionally 10 years ago when I couldn't beat noble...), but this got me interested. Mainly because it seems it could double the highest time victory HoF score to date. The more I think about it, the more complicated it gets.

• This technique relies on banking food for a veery long time, then let city grow. For optimal results, it needs to hit it's peak exactly on the turn you win. If you're too early, you lose 1 pop/turn.
• A time victory is won on a given turn, you can't influence this turn in any way. Every city needs a personalized plan for exactly how long to bank food and what turn to start growing to hit the peak on the correct turn.
• Calculations are fairly simple if future techs cannot keep up with whip anger. Then you don't need to account for citizens working tiles. If there are citizens capable of working tiles, then the calculations just got a lot harder for every city.
• Once they start growing, they will grow for 70-90 turns, depending on how much food you have. During these turns a lot can happen... Poisoned water supplies, for example. Health might become an issue in any case, if you can't keep up in future techs while banking food.
  
• Sushi culture. Meh. I was looking at some B&S maps and calculated one could get all cities close to pop 90 with the food available. However, using Sushi for so long with all that food would also make all cities legendary. Need to do some calculations here, it might still be better to use Sushi for a slightly shorter time than Cereal Mills. Or does this mean that one should first start banking food with Cereal Mills and switch to Sushi when it's late enough for cities to avoid legendary status?

What leader would be best for this? Sury and his +1 Baray was mentioned. Charlemagne and his Rathaus could be a good alternative that helps keeping up teching in the late game, which would allow larger cities. Shaka has almost same benefit, but better traits. I could consider Lincoln as well. Mainly because it's otherwise hard to find any kind of game suitable for the Americans. All cities want Supermarkets anyway, getting the extra happiness from them (and from CHA) can't hurt. +20% gold in corp HQ is also nice.

 

[Swordnboard]

Poisoned water supplies

And don't even think about vermin getting into the granary! How much would it cost to pay to have that much lost grain replaced? We might even have to turn off random events just for that one.

Anyways, I think managing happiness & health is definitely key here. But I don't think we necessarily have to get a future tech every turn. For cities with one or more food tiles, we can consider whipping 4-2 or 6-3 even (probably no good for marathon speed, though). This might let us whip execs instead of missionaries as well, or something else entirely and still build research 2/3 of the time. But more importantly, it slows down the unhappiness spam to where it might be more manageable with future tech. Being able to work tiles (even lame ones like ocean) not only helps our banking rate, but more importantly it dramatically reduces the amount of food eaten during the final growth push, raising our population cap that much higher.

We could also consider other methods of getting happiness. The only other unlimited one I can think of is vassal/colony spam, but every bit helps so maybe there really is a use for Lincoln. If we could somehow avoid STRIKE, culture slider might also be good (again runs into the cultural victory problem, but maybe there will be so little traditional commerce that it won't matter). There don't seem to be as many possibilities for health, but at least we would only need ~70 future techs instead of possibly hundreds.

 

[elitetroops]

Hmm.. That's a very good point! 4-2 whipping every second turn isn't necessarily any worse at all than 2-1 whipping every turn, depending on speed and what tiles there are to work. In many cases it might be even better for the food supply.

Let's say normal speed and food overflow is collected 200 turns. If 1 popping every turn, by the end of 200 turns there would be 180 whip . If 2 popping every 2nd turn, there should be only 80 . Now Monty also becomes very interesting. In this scenario he would only get to 60 .

Since happiness can be managed that well, I'm now even more considering the rainforest map. Cities could be pretty large, working only food tiles and green biology farms, while banking food. The map is smaller and not room for that many cities, so the result wouldn't be quite as spectacular as on a B&S map, but maybe the smaller map could make it more manageable for a first attempt of this new strategy. There's not enough resources for super strong Cereal either, but if every city is working ten 4+ tiles, this makes up for much of that.

Random events are dangerous. According to the random events list, vermin in granary can be avoided by paying a bit of gold. But there are other dangerous events as well.

 

[Bandobras Took]

Sushi culture. Meh. I was looking at some B&S maps and calculated one could get all cities close to pop 90 with the food available. However, using Sushi for so long with all that food would also make all cities legendary. Need to do some calculations here, it might still be better to use Sushi for a slightly shorter time than Cereal Mills. Or does this mean that one should first start banking food with Cereal Mills and switch to Sushi when it's late enough for cities to avoid legendary status?

I had a friend who accidentally won a culture victory because he had pushed up the culture slider so much to deal with War Weariness.

 

[Macksideshow]

What about carthaginians/hannibal for a leader. I'm not sure how teching is going to happen with all cities at low pop other than through trade route commerce. If tech is a bottleneck in this strategy is Carthage the best civ to handle that using a Great Lighthouse + coastal cities spam with cothons trade route economy approach?

Or perhaps we want a commerce core for teching that doesn't bank food?

 

[elitetroops]

If tech is a bottleneck in this strategy is Carthage the best civ to handle that using a Great Lighthouse + coastal cities spam with cothons trade route economy approach?

GLH obsoletes with Corporation tech. Trade routes are also unfortunately weaker with small cities.

How to tech is a big question. I looked at some time game in HoF tables and it actually had more score from tech than from pop, with 358 future techs (epic speed) and 2160 pop in 106 cities. I think that pop can be at least tripled, possibly even quadrupled, so beating the score is not a problem. But tech can still add a respectable amount of score on top of that. Doubling the population while not being able to tech might not be enough.

Much of the science would come from building research with Mining corp. Even on a tiny map +20 from there is not much of a problem. With factories that's 40. Maybe grow 6 cities large right away, they provide stronger trade routes for all other cities. (With UN resolution and Free Market there would be 6 trade routes/city.) Large cities could of course include a heavily cottaged Oxford capital and some cities that run a ton of scientists.

 

[Kaitzilla]

This idea is amazing for Time games.
Normal Speed and Quick Speed Score games too!

Great work Swordnboard.
I had no idea a city could grow higher than its food limit.