The true cost of worker stacking

[WackenOpenAir]

I've had this discussion before on this board, and still many people do not seem to realise the real cost of stacking workers before there are roads on a tile.
I thought of a new way to explain it, just did so in reply to another post, but decided this deserves its own post because of the vast number of people that do stack workers in early game.

suppose you have 2 workers in a city and you have 2 tiles to irrigate:
NOT STACK:
1 move both to different tiles.
2 both irrigate turn1
3 irri 2
4 irri 3
5 irri 4

STACK:
1 move both to tile one
2 irrigate 1
3 irrigate 2
4 move both to tile two
5 irrigate 1
6 irrigate2

Now the cost of it:
A worker is about half the cost of a settler. Little less in shields, but food is the real cost of these units so i calculate half.
If you think you can't compare it like this for the matter i am going to: If it would be worth less than half a settler, you wouldn't build any and build only settlers instead since cost IS half.

A settler = a city.

so 1 worker turn = 1/2 city turn.

With the example i gave, 2 workers were working 1 turn longer. so that is 2 worker turns = 1 city turn.

So you payed 1 cities full production (food + shields + gold) by stacking teh workers on that 1 job.

If you want, i can also state the cost in different ways.
2 worker turns = half the time needed to irrigate a tile. = half the reward of irrigating a tile = half a food per turn FOR THE REST OF THE GAME.

Now you might say: i'll build some extra workers so that i can still work all tiles i use and not lose that cost.
In this case, the cost is the extra workers you build. This cost looks probably less impressive. per 2 workers you stack, you need only 1/3th of a worker. = 7 food+3 shields. And then, you can stack them for the rest of the game. That cost looks much less impressive.
This indicates the real value of food and workers which is underestimated by everyone. You could say that all these costs for the same thing here are equal in value.

 

[whitecow]

All I can say is WOW! I always stack workers now i know though gotta go rethink my game.

 

[BasketCase]

There is a flip side to it in the case of clearing jungles. If you send a bunch of workers to different jungle tiles for the above efficiency, all the jungle tiles will be useless for 24 turns. However, if you have a nearby city that needs the tile, putting 6 workers on a single jungle tile can have it ready for use in 6 turns.

I run into this fairly often, because my usual method of establishing territory in jungles is to build a city on a jungle square to get an instal-clear.

 

[shadow2k]

As with everything, it really depends on the situation. You can't do straight math with Civ, as that's what the AI does, and why it sucks. You're the smart one, and you have to take situations into account.

If your citizens are working the two unimproved tiles, then stacking the workers could be to your benefit, despite the wasted worker turn. Check your math again, because food is power, and the situation is exponential. It may allow you to grow quicker, pop a settler quicker, or even net you another worker quicker...or just match shields to production due to growth, getting you a swordsman quicker. Whatever the case may be...

Two citizens working unimproved plains, for example...

NOT STACK:
1 move both to different tiles. 2fpt
2 both irrigate turn1 2fpt
3 irri 2 2fpt
4 irri 3 2fpt
5 irri 4 4fpt

They net a total of 12 food during those turns.



STACK:
1 move both to tile one 2fpt
2 irrigate 1 2fpt
3 irrigate 2 3fpt
4 move both to tile two 3fpt
5 irrigate 1 3fpt
6 irrigate2 4fpt

That's 17 total food for those two citizens.

You may be irrigating in a line to bring water to the plains from a river. You *can't* irrigate both at the same time, as the farther one won't have access to fresh water yet.

Your city may have one citizen needing a 2fpt tile right away, but not the second for a few turns. It may actually be of benefit to waste that turn to get the first tile irrigated asap, while the second one isn't being worked...yet.

You may be a masochist, and have raging barbs on, having to cover your workers with protection. Stacked workers take one unit, seperated units take two.

And so on, and so forth.

There are any number of reasons why it may be better to stack your workers. The biggest reason in my eyes is to get that tile productive sooner.

 

[WackenOpenAir]

yes to cut jungle i often make exeptions. 1 turn on 24 is also much less loss of efficiency than 1 turn on 4.
I do try however not to stack more than 2 on a jungle tile if not needed.
Avoid or delay building those cities that have no acces to 2 food tiles if possible. If not possible, yes then you have to because the cost of not doing so is even bigger.

shadow, you first count food in 5 turns, then count food in 6 turns. Not a very good comparison i would say so maybe you check your math...
count them for 6 turns, and you see 16 vs 17 food.

Edit:
PS with those jungle cities, immeadiately build workers from these cities. At least then you have a reward for cutting that jungle in that your city will stay small and can use its 2 surplus. Otherwise, you will keep stacking and cutting jungles and have half your worker force occupied keeping up with the growth of one city. while your city is staying small, you can then continue cutting at moderate speed with smaller (2 workers on a tile, instead of 4/6 like you might do for cutting that very first jungle tile when your city has no better tiles to use) or no stacking at all. then after a while, with much less investment, you will have multiple tiles cut and the city can start growing.

stacking to have an extra shield just a little bit earlier i will only do when setting up a settler factory and it will help the granary finish a turn earlier.
I use excel sheets to calculate such options and yes in those situations it may sometimes happen that i have to stack.
Also, when setting up settler factories, i often build mines and irrigation and move on without building a road. This also wastes a worker turn.
There is however one thing in the game that is more important that worker turns and that is worker/settler factories. Excel sheets i use here to optimize those first 20 turns in the game. Randomly stacking won't normally help you in this phase either. Only do so at the select moments it does really help the settler factory.

when irrigating a line, i often build a road first with one worker, who thus spends one movement getting to each tile. Then irrigate with another. This way you don't stack them, but you do get the road AND irrigation in 4 turns per tile without wasting any turns since the irrigating worker doesn't lose a turn on moving. If you start building the road early enough, you can also stack and irrigate with multiple workers since there now is a road and thus they don't lose the turns.
Of course, yes there can be situations where you simply need to get watter somewhere fast. Try to think about these things though when choosing where to settle and in what order.

 

[Evincar]

And what happnes if those two tiles are close enough so that you can move and start irrigation the same turn ?

 

[Offa]

And what happnes if those two tiles are close enough so that you can move and start irrigation the same turn ?

This means that both tiles already have roads. I believe that in that case worker stacks are a good idea as they get the job finished faster. The problem with stacks is the wasted turn moving onto an unroaded tile.

I think worker stacks can be useful for vital roads, but WOA's analysis is always sound. And he does like workers.

 

[punkbass2000]

Along similar lines, when you're roading a tile that is "unimportant" (obviously it's important if you're roading it and counting Worker turns is still pertinent, but I more mean "urgent"), use a Slave if possible and save your Workers for already roaded tiles. A turn moved onto an unroaded tile is a whole turn regardless, but you only lose a half-turn of work with the Slave. A third of a turn if you're Industrious.

 

[Khan_Asparuh]

Hi all,
my 2ç about worker stacking;
it's useful if the tile is allready roaded. Wasting more then one worker turn to go there is a no-no, so sometimes I prefer roading everything and then improve.
In fact, i usualy send slaves to road, clear jungles(later in the game) and so on as their wasted moves are cheaper.
But when the tile is roaded and if you can stack more units on it, it's worth.
Imagine a city with 3 roaded but not mined tiles.
Send then 3 workers (the number needed to mine in 1 turn) on each. You'll get +3 shields in 3 turns.
Now try to stack them.
1 turn - 1 tile mined - +1 shield
2 turn - 2 tiles mined - +2 shields
3 turn - 3 tiles mined - +3 shields
So you won 3 clear shields with the same work time. True for every improvement type.
Greetings

 

[microbe]

suppose you have 2 workers in a city and you have 2 tiles to irrigate:
NOT STACK:
1 move both to different tiles.
2 both irrigate turn1
3 irri 2
4 irri 3
5 irri 4

STACK:
1 move both to tile one
2 irrigate 1
3 irrigate 2
4 move both to tile two
5 irrigate 1
6 irrigate2

Case 1, neither irrigation tile are available until turn 6.
Case 2, the first irrigation tiles are available at turn 4.

A settler = a city.

so 1 worker turn = 1/2 city turn.

I fail to see how could that be. 1 worker != 1 worker turn.

 

[WackenOpenAir]

>Case 1, neither irrigation tile are available until turn 6.
>Case 2, the first irrigation tiles are available at turn 4.

This has been mentioned by the one who couldnt calculate. without stacking, they are both available at turn5 (not 6)
With stacking they are it turn 3 and 6. That is -2 and +1. for a total of 1 turn worth of irrigated tile. Now not talking about the part that you fail to see:
if you think 1 turn of irrigated tile is worth 1 turn of worker jobs, that means it would be worth it to have a worker occupied to keep 1 tile irrigated during the whole game as then also the number of lost worker turns equals the number of won irrigation turns. I think that obviously is not worth it.

>I fail to see how could that be. 1 worker != 1 worker turn
I fail to see how you can fail to see that.

 

[Alpha Draconis1]

I believe we can sum it all up in this; stacking workers is a gamble on a case by case basis. Like real life (and chess), linear mathematics do not work - there will be times when one is better than the other. When to choose one way or the other isn't as difficult rocket science, but it's certainly closer to a polynomial equation in calculus than to an algebraic line.

. . . Actually, I learned that in Eco 102. Never thought I'd see the day when I'd be able to apply fundamental calculus to a PC game.

 

[LulThyme]

What you said isnt very precise.
What you may be trying to say is that its hard to have an easy to present rule that works all the time.
In a specific case, analysis can often tell us what IS the best case though.
AS a couple person said including the OP, EXCEPT if the tile is somewhat very urgent or important, stacking worker before roading is losing worker turns, which lots of people did not realize.
this is a fact and has nothing to do with linear algebra.
And your point about chess?
A supercomputer playing a brute force algorithm (counting only the pieces, with no strategy at all and going as far as possible in the tree) plays at super grand master level...

Btw 'an algebraic line' as you call it is defined by a polynomial equation...

 

[punkbass2000]

"And your point about chess?
A supercomputer playing a brute force algorithm (counting only the pieces, with no strategy at all and going as far as possible in the tree) plays at super grand master level... "

What?! Where? I'd be very interested to see this. My understanding is that brute force chess programs aren't incredibly good. They'll beat most, of course, but will never stand a chance against a high-level player. A brute force machine would never see a gambit or exchange sacrifice coming. Pure material is far from everything in chess.

 

[Alpha Draconis1]

What you said isnt very precise.
What you may be trying to say is that its hard to have an easy to present rule that works all the time.

Er, I'm not sure I get you. What I said and what you said mean the same thing to me. I don't see how one is more precise than the other.

AS a couple person said including the OP, EXCEPT if the tile is somewhat very urgent or important, stacking worker before roading is losing worker turns, which lots of people did not realize.
this is a fact and has nothing to do with linear algebra.

You misunderstood. I was not talking about the worker turns themselves at this point, but referring to the "answer" or "rule" by which to follow when deciding whether to stack or not to stack, or in other words, following a recipe on when to stack even if you lose turns. That they lose turns was a given.

And your point about chess?
A supercomputer playing a brute force algorithm (counting only the pieces, with no strategy at all and going as far as possible in the tree) plays at super grand master level...

The strategy is programmed into the computer. Furthermore, that strategy is simpler because it's more akin to the "always war" playing style; there are a lesser number of variables to consider.I just chose chess as an example in the specific context of when to lose turns or not, NOT as as an overall comparison to the civ game.

Btw 'an algebraic line' as you call it is defined by a polynomial equation...

Not always. Algebra has at it's core the monomial equations, and the phrase I used was "a polynomial equation in calculus" to compare the complexity of analyzing a polynomial in terms of derivatives and integrals that is used so commonly in economics (production, price, demand, and supply) compared to the relatively simpler math involved in algebra. . . and why am I discussing this again? I think we got lost somewhere up there.

So to load a previous save; I merely meant that deciding on a rule to worker stacking is as complex as calculus is to algebra. Or maybe I should have used arithmetics as an example.

 

[Doc Tsiolkovski]

In most cases, stacking Workers is more effective. Period.
First, the situation where you should not stack them:

You have 3 Workers on tile A. Tile B, C, and D need roads, and are all adjacent to tile A. Plus, no important city is working any of them.
In this case, you of course should move one Worker to each tile. You'll have 3 roaded tiles at IBT turn 4, or in turn 4 if you use 'Road to...'.

Obviously, if a core city is working one of these tiles, or you need to connect a resource, you will move all three Workers to that one tile, and get the road the next turn. Or, if you need to protect the Workers (Barbs, AW).
Then, situation is different when C and D are not adjacent to A. Stacking is better.
Third, why do you want to road a tile at all?
Either to connect something, in which case the tiles are naturally not adjacent.
Or, because you want to mine/irrigate/clear/RR something. In which case it is nonsense to have the crew for the second job waiting until the can move on a tile...

7 Workers will road and irrigate 7 tiles in:

Stacked:
Move to tile A (1); 3 Workers to tile B, 3 road A, 1 irrigates A (2); 3 Workers irrigate A, done, 3 road B, 1 moves to B and irrigates (3); 3 Workers to C, 3 complete B, 1 moves to C and irrigates (4); and so on, 8 turns to irrigate and road 7 tiles.

Not stacked:
Of course, you need either Electricity or a lot of luck to have access to fresh Water on all tiles...
But even in that lucky case, 7 Workers will need (7 + any additional movement) turns. In the extreme case of all tiles being adjacent, that's 8 as well; in any other case, it's more.
And that means the city gets its food 8+ turns later...

It's very situational if single Workers are better. There are instances. But the biggest drawback is that you improve a tile for a purpose, and thus, you time it with city growth, and it doesn't matter if you have 3 or 4 tiles improved at once in a couple of turns. What matters is that you get the one tile the city uses done ASAP. Then move to the next tile, and have it improved once the city needs it.

 

[WackenOpenAir]

Yes it is different per situation, but i am very sure it is best not to stack in far most of the situations.

If you stack, you lose worker efficiency. On long term that is a big loss. The less efficient you use your workers, the more often you will have urgent cases in the future that might require stacking. A snowball effect.

If you don't stack, you work more efficient after a little while, you will be working ahead of your needs and never have any urgent cases.

i'll just not react on your example in detail as it is i am not planning to puzzle out what you are trying to write, possibly it was your intention that noone will puzzle it out judging by the way you wrote it.
But i do know that if you move 7 workers to each a tile, you used 7 movement turns. If you move them in groups of 3 to road in 1 turn (and you can do irrigation however you want when the roads are there) then you spend 21 movement turns. So in total, working your tiles will take 2 turns longer than not stacked.
Of course you can vary whatever you want by and do irrigation first, then road with 3 and move 1 away from the stack or whatever you want. But that won't make a difference.
If you move stacks of 3 on an unroaded tile, you lose 2 worker turns. You cannot chance that, and you don't need to make unreadable examples of combinations of stacking and not stacking. It is simple enough: move more than 1 worker on an unroaded tile costs you worker turns.


About chess:
I am studing the arts of poker at the moment with the intention of someday creating an AI that plays poker.
Poker playing AI's seem very hard to create as noone yet succeeded in creating one that beats the best players.
I also there came across the comment that chess AI is so much easier to design because you can simply calculate it out completely.
This is true in theory, there will be a day when computers will reduce chess to a game that noone can win or that will always be won by white.
Similar to the game where you have a 3 by 3 board, 2 players and alternatingly place your mark on a field trying to get 3 in a row. (i don't know the english name for it) Noone can win it if the other doesn't do something stupid.
The CPU power needed for this will be available in a very distant future though (in computer terms, not in galactic terms...)
For now, an AI needs strategy to choose which out of the billions of possibilities (over multiple turns ahead) it should calculate.

 

[stachnie]

Both possibilities have their advantages and disadvantages. If I need to improve terrain ASAP and I have lots of workers nearby, I use them in big stacks but if I have not as many of them and some time, I manage them to minimize the number of idle turns.

An example: I am non-Industrious, in Democracy (or Industrious and in Republic/Monarchy) and already discovered Replaceable Parts. I have a small town surrounded by Plains and I devoted two slave workers to improve terrain. If I need to make a road and irrigate some tile ASAP, I will move them to the tile (1 turn), make a road (1 turn) and irrigate (2 turns) - 1 tile improved in 4 turns. If I do not need it immediately, I will move workers to 2 separate tiles (1 turn), make roads (2 turns) and irrigate (3 turns) - 2 tiles improved in 6 turns.


Best regards,

Slawomir Stachniewicz.

 

[Doc Tsiolkovski]

Wacken, I think nobody argues with you how to must efficiently improve a huge chunk of unimportant territory. In this case, you shouldn't waste any Worker moves, and not stack them.
An example would be a corrupt, former AI territory. Or milking runs.
But you completely seem to refuse to accept that the Worker itself doesn't matter - it's the city that needs Food/Mining. As fast as possible, but tile after tile.
Or, you need to get a road somewhere: Why? Unneeded surplus resource? Single Worker.
Towards an enemy: Stack. Because any tile your roadnet expands more towards the frontline helps.

And, your example of the waste of worker moves is simple a rare situation, not the rule.
It just nearly never happens that you don't have to waste 2 moves for your second single Worker, 3 for the third., just to get to that tile.

What I don't like about your proposal is that word "effective". It may confuse newbies.
It is effective for milking. It is fatal for the early game.
Say, you have 4 Workers at 2000BC. Please don't tell me you would move each of them on an extra tile to irrigate, instead to move all 4 to that Cow?

 

[WackenOpenAir]

what do you mean with those 2 moves?

It is rare that i have to move multiple turns to reach the tile i want to improve. Also in early game.

Yes, also early in the game, or rather, espescially early in the game, i do not stack workers. Your cow example, i will rather send 1 worker 4 turns before founding the city so that it is irrigated when i build the city. And if not, no i will not stack workers on it, certainly because my workers are spread around over my empire and thus would some would need to move multiple turns to reach that tile.

Sure, if you stack them continuously, they will all reach it in the same time, but then they are not spread around your empire and thus when you want to work a tile somewhere else you might even spend multiple turns to move their with multiple workers.

In early game, i keep my workers well spread around and by using them efficiently, not stacking and not moving more than 1 move between 2 tiles that need to be worked, i can normally work ahead of my city growth.

And when i compare my qsc results in the gotm's i played, i don't think the way i play my early game is very "fatal"

I'll be damned if i stack 4 workers on 1 tile in 2000BC, also if that is a cow tile.
The earlier and the more important the terrain, the more important it is not to stack my workers and to maximize efficiency.
(not 1 tile specific important like the cow, but empirewise important as in the core of my empire pre 1000BC)

I will indeed irrigate 4 different tiles. Not 4 tiles for one city that won't be used for ages. I will irrgate for tiles for 4 cities. Preferably they will all be irrigated exactly by the time those 4 cities grow in size to need them.

And yes, i certainly and completely refuse to understand that the worker doesn't matter !
Spending extra worker turns on working 1 tile a bit faster now means all tiles worked by that worker in the future will be delayed. It is a decision that fits the ultraleft wing politicians.
Spend everything today and be poor tomorrow!