The true cost of worker stacking

[Offa]

What an energetic little thread.

Chess programming for instance is fascinating, and of course the web is riddled with info about this. The programmes are very clever, and to say they use brute force is unfair, as much work is done in optmizing the analysis. However, they do tend to either follow library positions (opening and end game) or search through a great number of possible sequences, albeit in a very clever way to avoid unnecessary analyses. Coupled with a way of analysing a position numerically, (eg piece count, give marks for passed pawns, take points away for isolated pawns etc etc) quite slow computers can murder people like me at chess.

Many other games are much less amenable to computers, even though they seem straightforward superficially, eg bridge programmes are pretty poor. Civ is probably pretty tough, however, the opening sequence for a worker is not so different from a chess position, and so you would think the computer could optimize much better.

Anyway, on topic, I think the 2 worker turn ~= 1 town turn rule is very impressive as a concept, and certainly emphasizes the importance of making workers do the right things.

I do think all of this discussion assumes there are no roads. Workers do not use up bits of their energy travelling. If they arrive via roads with any movement left they can do a full turns work improving the tile. So if there are roads, stacks are good, as the tile is improved faster.

WOA, would you consider using worker stacks to connect a vital luxury or resource early in the game? For example getting a lux connected a couple of turns early might let you increase your tech rate and get to republic a couple of turns earlier...

 

[WackenOpenAir]

Yes, i consider it for sure. As you might have noticed, i like to analyse and calculate stuff, and i try to consider every option i can think of

I have in the past stacked 2 workers to road an iron mountain for example. I also do consider stacking 2 workers for some other jobs sometimes. It just happens to be worth it so rarely that for the newbies to keep it simple, not stacking is a good general rule.
Much like the rule to irrigate plains and mine grassland under despotism, there are exceptions, but most usually it is true.

 

[LulThyme]

Ok about chess programs.
Let me make myself clear.
All the best chess computer programs DO NOT use brute force.
Although it has been demonstrated that a supercomputer using brute force DOES play at around grandmaster level.
Yes it does not see a gambit coming if the sacrifice is before the event horizon and the reward after, but the thing is it can consider more moves than a normal program (were talking over 30 plies easy) which somehow compensates.
It would get beaten by a good program on the same computer, but would still kick ass all around...

Still, while there are tons of easy changes that could have been made to the AI to make it a LOT MORe competitive (how about actually create armies for example), making a GREAT civ AI would be very hard...

 

[ainwood]

Re stacked workers - a quick analysis I did make me think that it really depends on the situation.

EG: Below is an analysis for two options: Two workers, mining two adjacent BGs. A city is working both of them. Stacked turns gains an extra 3 shields.

	Description		Tile 1			Tile 1			Total Shields
Turn	Worker 1	Worker 2	Worked?	Mined?	Shields	Worked?	Mined?	Shields	
0	Move	Move	TRUE	FALSE	1	TRUE	FALSE	1	2
1	Mine	Mine	TRUE	FALSE	1	TRUE	FALSE	1	4
2	Mine	Mine	TRUE	FALSE	1	TRUE	FALSE	1	6
3	Mine	Mine	TRUE	FALSE	1	TRUE	FALSE	1	8
4	Mine	Mine	TRUE	FALSE	1	TRUE	FALSE	1	10
5	Move	Move	TRUE	TRUE	2	TRUE	FALSE	1	13
6	Mine	Mine	TRUE	TRUE	2	TRUE	FALSE	1	16
7	Mine	Mine	TRUE	TRUE	2	TRUE	FALSE	1	19
8	Mine	Mine	TRUE	TRUE	2	TRUE	FALSE	1	22
9	Mine	Mine	TRUE	TRUE	2	TRUE	FALSE	1	25
10	Move	Move	TRUE	TRUE	2	TRUE	TRUE	2	29
11	Move	Move	TRUE	TRUE	2	TRUE	TRUE	2	33
12	Move	Move	TRUE	TRUE	2	TRUE	TRUE	2	37
13	Move	Move	TRUE	TRUE	2	TRUE	TRUE	2	41
14	Move	Move	TRUE	TRUE	2	TRUE	TRUE	2	45
15	Move	Move	TRUE	TRUE	2	TRUE	TRUE	2	49


	Description		Tile 1			Tile 1			Total Shields
Turn	Worker 1	Worker 2	Worked?	Mined?	Shields	Worked?	Mined?	Shields	
0	Move	Move	TRUE	FALSE	1	TRUE	FALSE	1	2
1	Mine	Mine	TRUE	FALSE	1	TRUE	FALSE	1	4
2	Mine	Mine	TRUE	FALSE	1	TRUE	FALSE	1	6
3	Mine	Mine	TRUE	FALSE	1	TRUE	FALSE	1	8
4	Mine	Mine	TRUE	FALSE	1	TRUE	FALSE	1	10
5	Mine	Mine	TRUE	FALSE	1	TRUE	FALSE	1	12
6	Mine	Mine	TRUE	FALSE	1	TRUE	FALSE	1	14
7	Mine	Mine	TRUE	FALSE	1	TRUE	FALSE	1	16
8	Mine	Mine	TRUE	FALSE	1	TRUE	FALSE	1	18
9	Move	Move	TRUE	TRUE	2	TRUE	TRUE	2	22
10	Move	Move	TRUE	TRUE	2	TRUE	TRUE	2	26
11	Move	Move	TRUE	TRUE	2	TRUE	TRUE	2	30
12	Move	Move	TRUE	TRUE	2	TRUE	TRUE	2	34
13	Move	Move	TRUE	TRUE	2	TRUE	TRUE	2	38
14	Move	Move	TRUE	TRUE	2	TRUE	TRUE	2	42
15	Move	Move	TRUE	TRUE	2	TRUE	TRUE	2	46

Weakness in this analysis is that it ignores things like shield-overrun on projects, roads, what the workers do next etc etc.

 

[Huzzamaster]

I have massive stacked workers usually, but I seem to think it suits my improvements better. When I'm improve to improve a certain city then it's good, but usually I work in the method of brancing out. My improvements basically get better from my capital and main cities, but with road systems throughout my entire empire. So really I don't always do it. But I can look into it more on the smaller picture, the turn to turn basis.

 

[Admiral Kutzov]

You're comparing apples and oranges. WOA seems to be trying to give newbies a general principle to follow in an early game. DocT is bringing in situational necessities. You've left out the effects of corruption & happiness and their bearing on the game. You've both got valid points, but set some parameters and see where it takes you. Take as given a 2 city empire with a non-industrious civ for the human and 3 workers. No resources or lux to be hooked up, no war pressure. Each city has one BG and the rest grasslands. Both cities have fresh water. No roads anywhere. Please explain to the idiots like me what you do with the workers and why. Yes, I realize this is unrealistic because of the starting worker.

 

[punkbass2000]

OK, here's a specific example where I would consider nothing but stacking. This is early in a PBEM, though I don't have the start position. I moved NW to found where my city is, then built a Worker immediately. My first Worker went to the Cow and Irrigated/Roaded, the second went S to the BG to begin Mining, and my first joined him to finish up the job and then they both go to the NE BG as a team. I really doubt not stacking would be advantageous here. I know not all situations are like this, but thie is the start of the game when Worker moves are most crucial. My capital needs improved tiles ASAP. Losing a couple Worker Turns in order to have each tile ready sooner was well worth it, IMO.

 

[WackenOpenAir]

OK, here's a specific example where I would consider nothing but stacking. This is early in a PBEM, though I don't have the start position. I moved NW to found where my city is, then built a Worker immediately. My first Worker went to the Cow and Irrigated/Roaded, the second went S to the BG to begin Mining, and my first joined him to finish up the job and then they both go to the NE BG as a team. I really doubt not stacking would be advantageous here. I know not all situations are like this, but thie is the start of the game when Worker moves are most crucial. My capital needs improved tiles ASAP. Losing a couple Worker Turns in order to have each tile ready sooner was well worth it, IMO.

This early in the game, your first city that needs to become a settler factory, is a different thing. I think i have mentioned that.
For this, i use excel sheets to calculate exactly what is optimal.
I have 2 ways here to speed up things for the capital at the cost of worker turns, 1 is stacking, the other is not building roads.
I prefer the not building roads since that helps you much more in the beginning. I also have to note, the cases where it is worth it to build a worker before granary when building a settler factory are extremely rare. (declaring it UNworthy if it delays the factory by more than 2 turns, declaring it doubtfull when it delays it by 2 turns and declaring it worty when it delays the factory by less than 2 turns. Since building a worker after the factory is ready takes 2 turns)
With the excel sheets, i search the fastest way to get the 4 turn factory operational, then i will see how i can do this with the minimum efficiency sacrifice needed.
I will however indeed do anything if it can get my settler factory operational faster. At this point you are not talking about just a city turn, but a complete game turn. 1 turn here = at least 1 turn in your victory date.
The second and third city, if they are building settlers or becomming a settler factory, also can be be worth a sacrifice. You really need to calculate this ahead however, so you only sacrife efficency when it does really benefit you.

 

[Offa]

I also have to note, the cases where it is worth it to build a worker before granary when building a settler factory are extremely rare. (declaring it UNworthy if it delays the factory by more than 2 turns, declaring it doubtfull when it delays it by 2 turns and declaring it worty when it delays the factory by less than 2 turns. Since building a worker after the factory is ready takes 2 turns)

Although I suspect you are right, doesn't this calculation neglect the extra work that the extra worker has done? Making a settler factory, and then making a worker two turns later, doesn't seem quite as good as making a worker and then a granary if this delays the factory by only 2 turns.

 

[Whomp]

Maybe I'm unclear here and I would never underestimate you WOA but is the wasted movement of a settler popping without roads taken into account?

 

[bradleyfeanor]

There seems to be much talk about this thread being directed to newbies. Well, I'm no newbie but it has certainly helped me out. I have been religious about worker stacking, and I never noticed that I was wasting all those turns on movement.

Thanks Wacken!

 

[WackenOpenAir]

@ offa:

The worker after granary takes 2 turns. If your factory is delayed by less than 2 turns, there obviously is an advantage.
If your factory is delayed more than 2 turns, there is 1 turn loss in trade for some worked tiles. This 1 turn is the most important type of turn possible. It is a turn that will cost you 1 turn on your victory At least. I do not think the few tile improvements provided can ever make up for that.

Now if the delay is 2 turns, then i said doubtfull. This is a tradeoff Basically it does give food equality and thus it should be advantagous because yuo have the extra worked tiles. However, i intended to build a settler factory, not a worker factory. I was not planning to build a worker first when the fact is ready, but a settler. I am not yet sure about this tradeoff, therefore i said doubtfull. The one time that i did calculate a 2 turn delay, i did build the worker.

@whomp.
A road for a settler to move is nice and certainly valuable. However, if it(building a road) delays the factory by one turn, it delays ALL settlers by a turn. While no roads do not delay all settlers because by later settlers, there will be roads already. Also of course, between getting the requesteds for your factory and your first settler, you might also have build a road. (not always, it also often happens that the first settler is earlier than the factory is really complete, and the first one takes 5 turns)

I must say, i cannot say much about this aspect in the choise between stackign for the capital and skipping roads because only once have i build a worker before granary yet. (Only once did my excel sheets show a delay of only 2 turns by doing so)

@bradleyfeanor:
Thanx

 

[Whomp]

Good stuff WOA. Ultimately it may be only one settler who has to hump it. Thanks.

 

[mad-bax]

I think ainwoods post has been rather overlooked. It deserves attention. The purpose of worker turns is to gain something (food/shields/gold/movement).

For a simple example that even I can understand. If you move three workers onto unimproved grass and then road it, it costs 6 worker turns. If you move 1 worker onto an unimproved grass and road it then it costs 4 turns. By stacking the workers I gain 2g and extra movement for two turns. Whether one method is better than the other depends on the situation IMO. But then look at clearing jungle on the other end of the scale. You can have 24 workers each taking 1 turn to move to their own patch of jungle and clearing it in 24 turns. After 25 turns you have 25 extra food and about 8 extra shields (1:3 BG ratio). If you use all the workers stacked it takes 48 turns to clear the same number of tiles, but every other turn gives you something. 1 shield early in the game is worth 10% of your best fighting unit, and later in the game is worth less than 1%. So not only do you get food and shields earlier by stacking, but they are worth more too. It's like buying a tech from the AI for 10gpt. It might hurt on the first turn of the deal, but by the end it is not even a consideration.

Ideally, for me, flat land in your empire should be roaded by individual workers, and irrigation, mining and forrestry should be done by stacking. Railing should always be done by stacking.

Finally, just a note on Risk. If a slave is 37 turns into clearing a jungle tile and a barb steps up next to him, what then? If you improve a tile in a single turn this can't happen as the improvement is completed before an unfriendly unit can get to the worker.

It's roading that is key. Once the tile is roaded then stacking is not wasteful. A nice newbie rule is to never allow a worker to leave a tile before it is roaded.

 

[vmxa]

I did not read all the replies and maybe things got clearer, but here is how I would look at your thesis.

I will skip the movement to the tiles as it is the same for all.
T1 (plains) 1+1 food as two tiles are involved =2F
T2 1+1 = 2F total = 4
T3 1+1 = 2f tot =6
T4 2+2 = 4f tot =10

Next with stack (btw not going to be common in first 100 or so turns)
T1 1+1 =2
T2 1+1 =2 total = 4
T3 2+1 =3 total = 7
T4 2+1 =3 total = 10
T5 2+2 =4

Now the thing is we got the extra food at T3, is it going to matter or not? If it pops that worker or settler sooner, we are ahead, if not we lost out.
This does not consider that we lost worker turns as the first method has both workers on their way to some other task.

Other issues could be a desert tile. Here we get 1 food at turn 3 and split we get none till after T4. So will it matter in this case? Maybe, if the town was starving long enough before we got irrigation going. Will 1 food be enough to matter at this time? Are we trying to get water farther out?

In the main I would agree that ganging workers in the AA is not the way to go. In the late MA, that changes as worker turns are not the most important thing verus getting some task done now.