When should I stop building Civs? A: Don't build them.

[personwithhat]

Well no . You have produced more 1936 equipment which are irrelevant by 1941 and that's why your "strategy" don't work most of the time . You don't only need to keep track of the number of gun produced but also their quality . Don't get me wrong , it can be worth it to go full Mils on some country but it's more the exception than the rule .

I've already listed an extensive list of equipment that is available in 1936 and maintains relevancy up to 1942 at the very least.
Proper planning can avert the production of any 'useless' equipment.

 

[donald dawkins]

Comparing factory count means nothing. More civs would give you more factories, but you will have them working for a much shorter time period.
1 factory working for 2 years has output comparable to 2 factories working for a year.

but what about after 2 factories working for 2 years? Not even that, but 2 factories working for 1 year, then 7 factories working a year after (2 producing year one, plus the 5 new military factories made in year 1)?

by building mils only your effectively giving yourself 1 year significant advantage in material production against an opponent and then a final year of comparative strength in terms of equipment production. After that you might as well start A new game as you’ll never be able to match a civ building nations production after that.

Equally, whether or not you even gain equipment superiority for a mil only strat is entirely dependant on when your opponent switches to mil construction and when you declare war on them, as well as your opponents starting equipment reserves. so really to have a very strong advantage in my presented cases you would need to declare war a year prior to your opponents switch to military production, which is very difficult to do as the nation about to be attacked is informed of an i,pending attack via notification and can begin militarising.

In other words, in most cases to truly profit off of a mil only Germany form example, you would need to declare war on a Soviet Union in 1938 to Maximise your material superiority Before the average soviet player begins the mil switch in 39. This simply would not work as a Germany or most other foreign powers due to tech limitations, general equipment/manpower restraints and the geographical barriers that favour soviet defensive strategies in early wars (see Stalin line). This does not take into account the very large reserves of early war equipment the Soviet Union starts with.

So with a mil only strat, what happens in year three when you haven’t been able to defeat your opponent and now he’s got almost double the mils? Not only this, but can your equipment advantage out way the potential strategic cost of having to potentially fight both the allies and the Russian at the same time as early as 1938, when you have not even taken Poland or czechoslovakia yet?

mil only can only work for all or nothing early conquests, it may work against very green soviet players and the limited French and British democracies, but once the the early conquest are made, unless you have completely won the game (taken out the soviets, taken out the UK, taken africa), you are guaranteed to lose.

 

[bitmode]

Case #1 hence depends on more factories (and resources) -> Adding more CG and resource requirements than Case #2.

Sure, case #1 eventually needs more factories and hence CG to catch up. I was referring to the early game. The MIC-only strat not only has to make do with the CIC available at start. And not only does it cannibalize that CIC through the CG it adds. It also adds those CG more quickly initially. Building only MIC in the first year will yield more total CG at the end of that year than when building only CIC.
Now this might well not move the needle of total equipment produced by much and you might be right in your overall conclusion. But it is an effect the formulas do not represent.

I'm simply assuming that you are using the full 15 civs per factory at that given point.

Japan already starts with only 14 CIC available and that number persistently goes down as more MIC are added. This effect will be much more pronounced in countries with higher CG % like the US. By making M and C constants within a production line, you merely shift the problem to the number of concurrent production lines. When looking at the aggregate of all construction lines, M and C are functions of the prior choices between case #1 and #2.

The problem is that this doesn't seem to be enough of a factor to affect any simulated runs I've done (to a noticeable extent)

Yes, that may way well be. I guess it matches my underlying point which was that the math itself is not any more convincing than detailed simulations or play testing unless it makes very weak assumptions.

I'm assuming you know how much equipment you need for the usual basis of your divisions. E.g. 200 garrisons? 100? and what template? etc.
That should help with motorized/support equipment.

I usually don't know and tend to use support companies sparingly (yes, I'm a heretic).

 

[personwithhat]

Yes, that may way well be. I guess it matches my underlying point which was that the math itself is not any more convincing than detailed simulations or play testing unless it makes very weak assumptions.

True, but I'd like to point out that no model is precise without measurements and simulations. Like any theory it needs to be supported on a mountain of evidence.
My main goal here is just to re-open up discussion, I can probably iron out a better model that is closer to what the game is.

Just felt that the results I have are already actionable, at least I've had more luck building pure-mils in my games.
At least, the whole "Build Civs until 2 years before war" statement should be discounted.

Japan already starts with only 14 CIC available and that number persistently goes down as more MIC are added. This effect will be much more pronounced in countries with higher CG % like the US. By making M and C constants within a production line, you merely shift the problem to the number of concurrent production lines. When looking at the aggregate of all construction lines, M and C are functions of the prior choices between case #1 and #2.

True, different model would be required to handle aggregated choices between case #1 and #2. Right now I only model the decision from a point in time and decide between which case to do. It's not as precise but seems to hold up in tests close enough to be usable.

I think the main issue(s) lie in the fact that Civ costs were increased, and Civilian economy penalizes Civ construction too.
There are also quite a lot more Military factory bonuses from Focuses/etc. than there are for Civs......

 

[personwithhat]

but what about after 2 factories working for 2 years? Not even that, but 2 factories working for 1 year, then 7 factories working a year after (2 producing year one, plus the 5 new military factories made in year 1)?

I don't even follow what this statement means. Even disregarding the missed military prod, a Civ will only pay for itself (prod wise) within 2-3 years assuming good construction bonuses etc.

by building mils only your effectively giving yourself 1 year significant advantage in material production against an opponent and then a final year of comparative strength in terms of equipment production. After that you might as well start A new game as you’ll never be able to match a civ building nations production after that.

That's quite incorrect. By building mils only you gain yourself an consistent advantage over the Civ-building player relative to how many civs they've built.
The more civ's they've built the larger your advantage, until such a time as their Civs build enough mil's to outproduce you.
Which, according to the tests and the basic model here, is 5+ years after their last Civ.

 

[Louella]

with a particular adviser, one of the focuses, and the right economy law, you can achieve instant conversion of civs to mils as the UK.

 

[foamingjetty]

This .pdf reminds me of the kinds of things I make for members of my gaming community for fun when they ask for optimizations in this or that random category. +1 looks like a fun time was had here.

But any time I do something like this for my friends, I find myself updating my answer for days afterwards because the assumptions we have to make to allow for a thorough mathematization of a question tend to limit the usefulness of the conclusions to only cases where those assumptions are irrelevant. I strongly doubt that can be safely said here.

I'm not going to say "I'm a math professor" or "I have x thousand hours in the game" as though these (true) statements are enough to give me credibility on the internet. It's just that the amount of experience-based dismissal of this conclusion should encourage you to reconsider some of the simplifications made in service of fun equations gymnastics. Saying your conclusion is wrong in game terms is not the same as saying you are mathematically wrong.

Just as one example, you say frequently that "it's the amount of equipment over time that matters, not the number of factories two years later" and "having fewer factories over longer time is a net benefit because you use less resources." You're claiming that a stockpile purely of 1936 guns, built without interruption or upgrades throughout the game, will win wars as effectively as a diverse armory of combined arms. Having a larger overall capacity to build 1939+ era arms is important for reasons you're excluding from your numerical analysis.

That is, saying "every gun you build all game long is worth 1 gun, and your fighting ability = sum of all 1's built during the game" is by itself a critical flaw.

That's before considering the limitations in building infra, ports, airfields, repairs, or trade, which you already acknowledge. Curious if you try to consider any of these complications in your analysis in the future.

 

[blahmaster6k]

It's entirely possible that he subscribes to the Feedback Gaming School of HoI4, which dictates that you win the game by putting one research slot on artillery day 1, hard research it, and keep doing so until you've finished the entire tree. Then you just spam out 7-2s with all your factories on guns and artillery. This would explain his lack of consideration for resource needs for rubber and oil, since if you just build infantry and don't use planes or tanks you're likely not going to be trading for much of anything as long as you have plenty of steel and at least some tungsten. It would even slightly validate his assumptions about factory output being equally as important at every stage of the game. The problem is that this can work in single player against the bad AI, but 7-2 spam won't do anything to a competent player.

 

[personwithhat]

You're claiming that a stockpile purely of 1936 guns, built without interruption or upgrades throughout the game, will win wars as effectively as a diverse armory of combined arms.

Never claimed that. If you're as experienced as you say, you should be able to clearly read the written word.

I keep restating this, but there's a plethora of 1936 equipment that you can build while you wait for your primary gear (Gun_II etc.)
E.g. Support, motorized, recon planes, light tanks, fighters for interception, etc. etc. etc.
In some situational cases Gun_I is useful. Doesn't hurt for garrisons to have them, as their entire job is to delay so the real units can march in.
I don't recommend sticking to 1936 guns all game unless you have some strat specifically for this.
You should also consider that most nations can get at least some 'useful' equipment by early-mid 1937, and dump all their mils on it.

The fact of the matter is that by building Civs you lose military output. You may end up with more output of a particular type (lets say more Fighter 3's) later on in the game. You do, however, lose a significant amount of nearly everything else.

That is, saying "every gun you build all game long is worth 1 gun, and your fighting ability = sum of all 1's built during the game" is by itself a critical flaw.

Ineptitude or inexperience with logistical requirements for your divisions is not my problem.
If you manage to somehow find yourself in a situation where you have no use for anything you are building.
Then I suppose you can build some civs for the first year. But no more than that, not if you plan on having a relevant military output by 1941 or so.

If you completely discount the lost military production, then a civ in 8/10 infra with an 80% military construction bonus, will pay for itself in 1.8 years after it finishes building. If it builds only mils until it pays off! If it's building other Civs, will take much longer since Civilian construction bonuses tend to cap out at 40% or so.

 

[personwithhat]

Then you just spam out 7-2s with all your factories on guns and artillery.

7/2's won't push jack into fortified infantry without air/tanks for break through. Obviously.

It would even slightly validate his assumptions about factory output being equally as important at every stage of the game

It's of equal import only if you've planned out a use for each, or aimed for a target stockpile (since I'm sure your experience can assist you there).

The problem is that this can work in single player against the bad AI, but 7-2 spam won't do anything to a competent player.

I play competitive MP games just fine ty, there's no need write garbage then pretend I said it.
In fact, I don't even play vs AI because it's atrocious.

 

[blahmaster6k]

there's no need write garbage then pretend I said it.

I never said you said that, I was speculating about potential use cases that would make your model valid and trying to figure out why you arrived at the conclusion you did. Especially considering that this appears to be a case of "start by stating your conclusion, then find a model that supports this conclusion." I could be wrong, but it gives off that appearance and I would wager a guess that others in this thread would agree.

I'm honestly surprised you say you play competitive multiplayer, because your entire premise is more or less "the meta that has been converged upon over four years of hoi4 and was shaped by thousands of players' collective experience is completely wrong, and here's why." Surely there must be a reason people have near universally arrived at a conclusion opposite to yours over the past four years. I would encourage you to think about what that reason is. Not everything can be solved mathematically, and you don't gain credibility by using circular reasoning to dismiss others' criticisms of your poor assumptions.

 

[foamingjetty]

Never claimed that. If you're as experienced as you say, you should be able to clearly read the written word.

You're clearly feeling stressed by negative responses. I'm not trying to dog on you, I enjoyed your math. What I'm saying is I *did* clearly read the written words on your pdf. You *are* saying that this calculation of only 1936 guns and nothing else is still relevant to the game, while the community is responding that it's more complicated than that.

Don't take offense when you assume away a lot of complexity, and people respond by saying it's not that simple.

You say with words something and say with your calculations something very different, that's all.

 

[personwithhat]

You *are* saying that this calculation of only 1936 guns and nothing else is still relevant to the game, while the community is responding that it's more complicated than that.

The calculation in the PDF doesn't mention 1936 guns.
I dumped a quickie sample run on the forums and used Gun_I to simplify the total IC comparison, much like if I were doing docks I would spam convoys.
I could have built motorized or support equipment, doesn't matter. The point was to prove that Civs affect total military output in a negative way.

when you assume away a lot of complexity,

I've explained the reasoning before for every assumption.
If you have issues with my model, then feel free to post recommendations or a better model of your own.
Repeatedly saying "resources are an issue" or "Gun 1 is bad" doesn't contribute anything.
None of you have proof or are modeling on how this would be an active effect. You are merely stating opinions.

 

[personwithhat]

I never said you said that, I was speculating about potential use cases that would make your model valid and trying to figure out why you arrived at the conclusion you did. Especially considering that this appears to be a case of "start by stating your conclusion, then find a model that supports this conclusion." I could be wrong, but it gives off that appearance and I would wager a guess that others in this thread would agree.

Actually, I was running random economic tests to optimize a Japan build that I'm (still) working on.
It ended up that spamming mils/docks all game and ignoring Civs gave me more output than I would have had normally.

Hence the investigation, and then a write up in PDF form to share the information and maybe get some useful feedback.

 

[blahmaster6k]

Actually, I was running random economic tests to optimize a Japan build that I'm (still) working on.
It ended up that spamming mils/docks all game and ignoring Civs gave me more output than I would have had normally.

Hence the investigation, and then a write up in PDF form to share the information and maybe get some useful feedback.

Thanks for clarifying my misunderstanding. I feel better about your intentions now, but that doesn't excuse the poor reasoning involved in the model.

None of you have proof or are modeling on how this would be an active effect. You are merely stating opinions.

You have just as much proof as they do for their positions, that's exactly the point of what most people in this thread are saying. You may have reasoning behind your assumptions, but it doesn't matter because the assumptions necessarily make the model unrepresentative of the actual game. When you make ANY assumptions at all, your model becomes just another opinion. Because you just can't model the game accurately while making assumptions such as these, as @bitmode was saying.

You're trying to put the onus on everyone else to take the time to do the math as you did to disprove your (fallacious) model with one of our own, and you refuse to acknowledge anyone else's position unless they back it up with a mathematical simulation of equal complexity to your own. In reality, any model that would be accurate would necessarily be orders of magnitude more complex than the one in the OP. Quite franky, I had more than enough calculus for my entire life when I was at university and have no interest in doing any more. But even in the absence of dissenting calculus, you can't cite your own model as The Truth and use it to shout down disagreement when it's as flawed as it is.

 

[foamingjetty]

I've explained the reasoning before for every assumption.
If you have issues with my model, then feel free to post recommendations or a better model of your own.
Repeatedly saying "resources are an issue" or "Gun 1 is bad" doesn't contribute anything.
None of you have proof or are modeling on how this would be an active effect. You are merely stating opinions.

Okay, fair enough. You invested time in this and now feel protective of it. You have not by a very wide margin reasoned through every assumption your calculations make. My "guns 1" example is not the only example. But you're dodging it very hard, which means you either don't understand why your calculations are saying that, or you don't understand why that would be a problem.

It is a fallacy to propose that the only responses that are valid are ones that take the same form as yours. For my part I know my limitations enough to know that I wouldn't be satisfied with my own math attempt any more than I am satisfied with yours. There are a helluva lot of variables that make this hard to quantify in two pages, even if all players have scripted military moves.

Since this probably sounds hostile to you and we're not really communicating, I'm just going to let this go. Props to you for applying some math in any case. Have a good one.

 

[personwithhat]

Because you just can't model the game accurately while making assumptions such as these

If the tests correspond to the model then at it's at least possible that the model is a good indicator.
You're dismissing my assumptions on literally no basis other than opinion, while I have done quite a few runs that support it.

But even in the absence of dissenting calculus, you can't cite your own model as The Truth and use it to shout down disagreement when it's as flawed as it is.

Your disagreement is merely opinion. I have done sample runs that follow the model. If you do not believe it accounts for all variables properly (which it probably doesn't), then find and present a repeatable case that this model does not match.

This would prove that my model is invalid, and would help me triage and identify possible influences and corrections that should be done.
As of now, in the cases I've tested, my model matches close enough to function as a measuring stick.

It is a fallacy to propose that the only responses that are valid are ones that take the same form as yours.

Stating "Your assumptions are invalid" with absolutely no proof to back this up is not a valid point.
The onus of proof is on me, which I've presented already.
If you find my test cases too limiting or invalid, make some of your own.

For my part I know my limitations enough to know that I wouldn't be satisfied with my own math attempt any more than I am satisfied with yours. There are a helluva lot of variables that make this hard to quantify in two pages, even if all players have scripted military moves.

Feel free to list out all the variables that would impact the model and why. Even that would be a nice helper in structuring a theoretical model.
Maybe do a test run or two before dismissing the model.

 

[personwithhat]

A brief refresher on the scientific method:
1 - Run test cases
2 - Make hypothesis that best fits available evidence
3 - More test cases. Don't match? Back to step 2.

After quite a few test cases and evidence:
4 - Seems enough evidence. You now have a theory. Repeat step 3 with wider audience.

Ignoring any tests I do, as well as all my reasoning, on the basis of:
- "I dont think it will work"
- "Why nobody else done this before"
- "We done it already and you're wrong"
are not valid arguments.

Some of the points brought up in this thread are quite valid.
Some of the counter-points I bring up, IMO, are quite valid as well.
I've provided a repeatable test case (and I've done others myself, before).
Now it's your turn, if you feel like defending your points.

 

[Simon_9732495]

Now it's your turn, if you feel like defending your points.

No, it's your turn, and that's why:

The title of this topic is: "When should I stop building Civs? A: Don't build them."
Your paper says:

Unfortunately, in all but the most extreme cases,building a Civ does not pay for itself. The truly optimal decision should be to exclusively build Mils.

You have noticed by now that you are wrong with the title and this statement and you start backpaddling.

For example you say now:

If you need more resources and you build Infra or Civs for resources, that is fine.

You admit that there are cases where building CIVs makes sense. For example to be able to import resources in the future.

So, you should update your claim.
Are you saying, that with static bonuses and producing always the same equipment and discounting import, any building other than MILs, repair and whatnot...
THEN building Mils only is better. Is that your claim?
If that is all what you are saying. Ok, well, obviously.

If you are claiming more, could you please update precisely what it is.




Edit, some more thoughts:

Maybe a way to think to get a grasp on how to produce MILs and CIVs in a optimal manner in a real game:

"I'm Germany. How can I have as many Medium3 Tanks (1941 tech) in 22. June 1941 (start of Barbarossa)"

Therefore you need as many MILs as possible at the point where Medium3s are available.
So you need to optimize your MIL count to a date far into the game. To reach that goal you build CIVs in the beginning.
That strategy may not be optimal in overall industrial capacity (IC) worth of equipment produced. But that is not a goal in a real game.

 

[Ksyr]

Now it's your turn, if you feel like defending your points.

I appreciate that you are trying to further the understanding of the game, but you are the one voicing an opinion and you are the one that have to defend it. Your math is simplified to the point where it is useless for real games.