Should Excess shield/food overflow to next production?

[TheNiceOne]

Originally posted by JonathanValjean
I think food should be stored, but not shields. In real life, food can be stored, but manpower can't be bottled up to be used later. So, something else.

But in real life, manpower work is spread evenly throughout the year. The real-life equivalent of a 60-shield city is a city that produces 5 shields each month.
If they are to build something that costs 70 shields, they will be finished at the end of february the second year, and will have 10 months of that year left. It's meaningless for the city to then waste the remaining 10 months. Instead, they should use the remaining of the year's production for their next project - i.e. let the shields carry over.

 

[Bretwalda]

I voted yes, mainly because micormanaging is not much fun for me, and I like playing, but also dont want to fall behind because of this. So that would make the game much nicer for me.

There is a solution, check diz out: http://forums.civfanatics.com/showthread.php?s=&threadid=57158

 

[anarres]

I voted yes, it makes sense, and speeding my turns up is something I really need to do....

Bretwalda, that link looks cool, I am checking it out now!

 

[JonathanValjean]

Originally posted by TheNiceOne
But in real life, manpower work is spread evenly throughout the year. The real-life equivalent of a 60-shield city is a city that produces 5 shields each month. If they are to build something that costs 70 shields, they will be finished at the end of february the second year, and will have 10 months of that year left. It's meaningless for the city to then waste the remaining 10 months. Instead, they should use the remaining of the year's production for their next project - i.e. let the shields carry over.

You make an excellent case! Now if I could just change my vote....

 

[Coorae]

I voted yes definantly

It doesnt really matter if it is unrealistic, you are already doing things like switching the great library to the pyramids when its almost done.

to me micromanagement is not fun unless I am nerdy.

 

[Arnold]

Idea of overflow make play faster for competitive player, and make casual player (and AI) more competitive.

I not understand thinking of why some vote "no". Perhaps they not play game enough yet. I vote yes, and hope Firaxis can make change in time for C3C.

 

[a space oddity]

This change will certainly not be in time for C3C, it's being beta tested already.

BTW I voted no, I like the game the way it is.

 

[Arnold]

Space oddity, how you know change not in C3C? You are beta tester? Qitai already propose this change in suggestion forum many month ago.

 

[dojoboy]

I think it should, but it needs to be tied into an advancement tech. At "x" point of the game it becomes possible.

 

[Ville]

Maybe for same kinds of units you could be able to continue without wasting shields when unit is finished.

 

[nyboy10128]

I was torn with this sugestion but voted no. If it was coupled with the sugestion someone made stating that you lose any carry over if you changed production, I would have voted yes. It has always bothered me that you could change what you are producing after you started without any penalty. It does't make sense that halfway through building a tank you could switch to university without wasting all those tank parts that that couldn't be used for the university.

 

[Galcador]

Voted yes. I think this is the single biggest improvement they could make to the mechanics of the game. Less micromanagement for the human, and it would surely improve the performance of the AI civs.

 

[Roland Johansen]

People who are voting here should try it first before they vote, and that is possible because Mumchembled already made a utility that does this (food, production and science cascade can be toggled on and off seperately). Bretwalda already posted the link but I will repeat it here for the ones who don't want to scroll up

By the way, I voted yes. I agree with TheNiceOne arguments on this completely. The way it is now forces you into unrealistic, tedious, unintelligent micromanagement. And it would be the most easy improvement of the AI, that I can think of.

 

[eg577]

"It helps improve the weak AI" is not a valid reason at all. I already went through this, but I shall reiterate my point as clearly as I can:

Instead of recoding the game so shields carry over, why not just recode the game so the AI knows how to micromange? Think about it. Because a computer can micromange twice as much as the average human can bear, this would "improve" the AI more than making shields carry over.

I realize that the poll was just "should shields carry over: yes or no". If you want shields to carry over for realism or something else then that's your reason. But, "it would help the AI" is not a valid reason. It would be like saying we should remove artillery because it would help the AI. Does that make any sense? No. It isn't too much to ask that the AI use artillery. It isn't too much to ask that the AI micromange.

 

[alexman]

Yes, actually, it is alot to ask. For sure too big of a change to add at the last moment.

But leaving that aside, it is your point that is somewhat irrelevant. The carryover would indeed help the AI, as people claim. Just because you found a way to help the AI that you like better, doesn't mean that this change would not help the AI!

This is a game mechanics change, primarily to reduce micromanagement for the human. A side effect is that it would help the AI. I don't see anything invalid about that statement.

 

[Roland Johansen]

I agree that it would always be better to improve the AI than to change some great game-element so that the AI has less problems with it. But I don't consider the wasting of shields and food a great game element (this is personal I admit) so if the change to shields and food cascade improves the AI, then that is an added beneficial side-effect for me.

As for improving the AI: I'm not an expert in AI-programming, but as far as I know this is far more difficult than people believe it is. There is a big difference between being able to explain to a friend how to micromanage and telling a computerprogram how to micromanage. You'll have to write very general routines that always do the right kind of micromanagement. I think this is very difficult. Determining what is the right level of production is pretty difficult for a computer.

An example: We have a city in the late middle ages (musketmen and knights are around). Good cities in this era have a productionlevel of around 20 and so does our city. The city has no hills and the only way to micromanage it is by choosing between building mines or irrigation. To produce knights a production level of 18 would be ok (4* 18 =72, minimal loss), to produce musketmen a production level of 20 would be perfect (3*20= 60, no loss). So the AI has to decide somehow what unit is needed more. This is a very vague decision and thus difficult for the AI to make. Say it chooses for knights because it wants to start a war (AI's never plan things but this is a theoretical super-AI). It builds the mines and irrigation in such a way that a production level of 18 is reached. Now it is attacked by another AI and some of its musketmen are killed. Some routine tells it, it needs musketman, so the just changed terrainimprovements are changed back so that a production level of 20 is reached. By the time the terrainimprovements are changed the weak attacking AI forces are already beaten and the AI wants to build knights again to take some cities of this AI opponent.
You see where I'm going... A lot of wasted workeractions. This is just an example and in reality the AI has much more difficulty deciding if musketmen or knights are needed and other buildings and units have to be considered also. And this has to be done for each time-period and cross-linked with all kinds of strategies. And the workers have to be present at the right spots on the map to do this micromanagement.
One of the biggest problems with the AI and micromanagement is that the AI doesn't think about the big picture or about what will be happening in 10 turns and that is important for some kinds of micromanagement.
The level of micromanagement that I'm using to defeat the AI is unattainable for an AI, I think. That's another reason for me to vote yes for this change, although not the most important one. That would be that I don't really like this kind of unintelligent micromanagement.

If it were easy to learn micromangement to the AI then Soren would have done it. Maybe some level of micromanagement is attainable for the AI (change between mined hill and irrigated grasland when beneficial, although the term beneficial is not that easy to define). But as I said before I don't know that much about AI-programming so even this might be too difficult.

 

[Qitai]

nyboy10128> You could have voted something else instead of no

 

[eg577]

Alexman,

Like I said before, if someone's reason for wanting shields to carry over is that it is more realistic or it reduces micromanagement (a subjective benefit) then fine, that's your reason. The only point that I made, hopefully more clearly the third time, is that if your only motivation in the decision is that the AI benefits from the change then its an invalid reason since it prexcludes a bigger benefit for the AI. Now, who says the other bigger benefit for the AI is on the table? No one, but if one change (i.e. shields being able to carry over) is on the table, why not another? Now, if helping the AI is only one of many reasons to want shields to carry over then you have to weigh whats better: less micro and a little help to the AI or a bigger help to the AI. I think this was worth pointing out since no one probably considered it. By the way, I did not take this poll in context of Conquests at all. Speaking in general, giving micromangement routines to the AI is not too much to ask.



Roland Johansen,

Here is the thing: it is possible to micromange to a certain degree making purely mechanical decisions. Such decisions may not be as good as making possibly better, but vague decisions (which a human would have to do say), but such mechanical decisions have a saving grace - they are mechanical! The AI could easily apply that mechanical routine to every city, every single turn for two seconds of effort. Human players can only bear total micromangement in the begining stages, and then only in a handfull of cities later on. If the AI's routine is just 50% as effecient as the best human's say, it still wins by a longshot.

Your example wasn't approaching the problem correctly if we are to have a mechanical system. Rarely when I myself micromange do I go "Hey, look at this...if I changed to something else I would waste fewer shields". 99% of the time I go "Ok, this is what I want to build and that is final. NOW lets reorganize the labor."

So the AI wants to build a knight in Chicago, period, end of story. Now he looks at the labor. Increase the shield count until the number of turns decreases by one (or two, or whatever), but only if you can get by without reducing the number of turns to next food growth. The change of commerce gained is completely irrelevant, since at worst you are moving a laborer from a roaded square to an unroaded square (which will be roaded soon no matter how bad your worker AI is). Presto!

Now there is also the other direction. Say you can't find a way to decrease the number of turns to build even by one turn. So, start decreasing shields in favor of more food and commerce!

What if the city is going to grow midway during the project? Could that make a difference? Yes! New shields gained midway from an addition laborer make a big difference in the reorganization of the labor, something which is often especially tedious for human players to keep track of.

What about if you are going to grow midway through the project, but that citizen needs to be a tax collector/entertainer to maintain happiness? Why then, you would let yourself increase the number of turns to grow if it would let you decrease the number of turns to build or increase commerce!

Obviously this is just scratching the surface of how far you can go with purely mechanical routines. What we have on paper right here, if applied to every city every turn, would easily surpass human effeciency.

 

[SirPleb]

Originally posted by eg577
Alexman ... if your only motivation in the decision is that the AI benefits from the change then its an invalid reason since it prexcludes a bigger benefit for the AI.

I think that Alexman's post clearly precludes this question, he clearly stated (as did Roland Johansen and others) that he sees this as a nice side-effect.

Originally posted by eg577
Obviously this is just scratching the surface of how far you can go with purely mechanical routines. What we have on paper right here, if applied to every city every turn, would easily surpass human effeciency.

I disagree with this. The logic you describe is already a fair bit of programming but it wouldn't come near to competing with what I do. In the later game it would actually do more than I do but by then it doesn't matter - by then my early gains have multiplied and can't be offset by gains made by the AI later on.

I think that the logic you describe has a few weaknesses. Examples: more logic would have to be added to "plan" for combined production over a number of turns to acheive a goal with some turns on food and some on shields; your statement that "commerce gained is irrelevant" is wrong in the assumption that only roads matter (rivers also matter, as do resources and coasts); and corruption must be factored in. (The AI currently does not factor it in as far as I can tell. That is illustrated by the governor's appalling decisions to emphasize shields over food in totally corrupt towns. That and some other specific cases would be much easier to fix than the logic you propose but Firaxis has not done them yet.)

Of course one can say well sure, that only scratched the surface, you just need to add code which does ...........

I think the important point there is that even the logic you already described, with its loopholes, is a lot more work than the simple change Qitai described at the start of this thread. It requires more code, more testing, and there's more chance of unforseen side-effects. And those things all get worse once you start adding layers of logic to patch the loopholes. Qitai's suggestion has considerable advantages over what you describe in that it:
1) Is intended to improve the fun factor for the human player. The improvement to the AI is secondary.
2) Is a lot easier to program, to test, and to anticipate the consquences.

 

[Pembroke]

While it is certainly true that the AI could be improved either by adding shield carry over or by adding AI micro management code I think it's a safe bet to say that adding the carry over is less work. The existence of a fan-made utility done without having access to the source code implementing the carry over is a strong indication that it's probably very much less work.

Thus Atari has the following options:

A) Improve the AI by adding carry over => minor work => less money needed.

B) Improve the AI by adding micro management code => major work => more money needed.

Let's now look at these options more closely:

A) Less money.
B) More money.

Let me repeat that: A - Less money. B - More money. Less. More. Again: Less. More.

Guess which one has the better chance of being implemented in a free patch?