How does the military work in HOD?

[Tikinaattori]

Yeah that's what I've been trying to do. Problem is, a lot of those factories seem to be pretty unprofitable during peacetime, meaning I have to subsidize them if possible, or let them close and re-open them during war time, which is a little pricey.

In my latest games, I have started to supply my army at least 65% ( 100% if I can afford it ), so that I have at least some artillery factories left when war starts. Since you always buy first from your own factories, your own demand keeps those factories alive. Also if you can sphere countries that have decent armies, that helps too, because they will buy from your domestic market first. When war starts, your factories can hire new people and increase production if possible. If you set your own peace time supply very low, your own military factories struggle really badly in the world market , because Great Powers usually have enough military factories to supply their own, and their sphere's armies, at that normal 30% peace time level, so there's really low demand.

 

[ifnt]

If I've just rolled a 9 i should be less likely to roll high next roll, if I've just rolled a 0 I should be more likely to roll high, unfortunately with an RNG, five very high rolls in a row is pretty common.

Law of small numbers - a behavioral fallacy.

 

[Miaow]

Law of small numbers - a behavioral fallacy.

Actually, it's not true—as in, his statement is not true.

Your past rolls have no incidence on your next rolls.

In the very long term, all roll values should come up roughly as often as each other. The order in which they come up is completely random, though, and any given order is as likely as any other.

For a two-roll set, you're twice as likely to get a 5 and a 6 than two 9s, because 5 and 6 is two possible combinations: 5->6 and 6->5. However, you're as likely to roll 0 and 9 as you are to roll 0 and 0 or 9 and 9.

 

[Nimic]

I don't know what the guy who loses all even battles is doing, because that's not my experience at all. I often do end up outnumbering the enemy, simply because I plan ahead and they do not, but even when not I usually come out on top due to army composition and tech.

 

[ifnt]

Miaow, your statement is entirely correct.

Actually, it's not true—as in, his statement is not true.

However, I'd argue you misinterpreted my point on the law of small numbers. It's a term used to describe the fallacy that people tend to think in mistake what happened last should not come again. It's a behavioral "law", not a scientific law (like its remote cousin law of large numbers, lol).

p.s. Even Wikipedia did not give a good description of this - it happens that as a graduate student I am working on research regarding probability perceptions - once I got time in the summer I'll give my attempt in describing LSN...

edit: oh - I misinterpreted your "... as in ..." part. Nevermind

 

[Chamboozer]

And you missed the point of what TypoNinja was saying. He wasn't talking about reality, but rather about how he thinks the system should work. In other words, the rolls are random but ideally, in order to be fair to both sides in a battle, they should artificially be made less likely to land multiple high or low numbers in a row.

 

[ifnt]

define fairness?

 

[Chamboozer]

define fairness?

According to him, not having an element of randomness great enough to undermine good preparation.

Planning, preparation, good tactics, strategy. These are the hallmarks of this type of game. Every aspect and factor should be under the players control. Some things we have limited abilties to affect (like what pops do) but they act in understood ways, and influencing their behavior is possible.

Taking control of the game away from the player is the cardinal sin of game design,

His post is right there on the other page if you'd like to read it.

 

[unmerged(624310)]

And you missed the point of what TypoNinja was saying. He wasn't talking about reality, but rather about how he thinks the system should work. In other words, the rolls are random but ideally, in order to be fair to both sides in a battle, they should artificially be made less likely to land multiple high or low numbers in a row.

Precisely, I know the system is random, my problem is that random is not fair. Many assume it is, because if two people each get a random result X number of times, both are getting the same treatment, but in practice it doesn't work.

Its easy to see why, Random is only fair on a large enough scale, depending on how rigidly you want to define fair, that scale can be quite large. For any given subset of random outcomes results tend to get 'streaky' (as in good or bad luck streaks). Simple home experiment to demonstrate, flip a coin, however many times your patience holds out for. Go with X's and O's for heads and tails while recording results. You have 50/50 for any result but your records will probably not (well they might, it is random after all) look like XOXOXOXOXOXOXOXO they'll look more like OXXXOOXXXXOOOOOOOXXXOXXXOO. Now that second string of X's and O's has the same of each, but if we are only use any group of four because combat was 4 rounds, problems start.

My point was that RNG is a bad way to simulate an element of chance, its biased on the sample size that games tend to use, you want weighted number generation to force short term normalization. Even something as simple as 2d6 instead of 1d10 for the dice roll would give a bell-curve for results instead of a line, and help average results, but the base problem will still persist, because following die rolls are not affected by previous ones, and need to be in my opinion.

Random is falsely assumed to be interchangeable with Fair or Even much of the time, especially in my experience, in computer games, and I don't think this should be the case.

 

[unmerged(734083)]

Oh nice. I'll try this in a new game- thanks for the advice!

 

[eleinvisible]

I had twice as many troops as them, Good army composition, decent generals. Outnumbered them in the fighting. Doesn't matter, dice roller screws you every time.

Fight defensive. Fallback toward your fortresses, bait them into attacking a smaller force then reinforce. It's the tried and true Victoria II method. Do no fight offensively until their morale is low and your numbers greater.

 

[ifnt]

Precisely, I know the system is random, my problem is that random is not fair. Many assume it is, because if two people each get a random result X number of times, both are getting the same treatment, but in practice it doesn't work.

Its easy to see why, Random is only fair on a large enough scale, depending on how rigidly you want to define fair, that scale can be quite large. For any given subset of random outcomes results tend to get 'streaky' (as in good or bad luck streaks). Simple home experiment to demonstrate, flip a coin, however many times your patience holds out for. Go with X's and O's for heads and tails while recording results. You have 50/50 for any result but your records will probably not (well they might, it is random after all) look like XOXOXOXOXOXOXOXO they'll look more like OXXXOOXXXXOOOOOOOXXXOXXXOO. Now that second string of X's and O's has the same of each, but if we are only use any group of four because combat was 4 rounds, problems start.

My point was that RNG is a bad way to simulate an element of chance, its biased on the sample size that games tend to use, you want weighted number generation to force short term normalization. Even something as simple as 2d6 instead of 1d10 for the dice roll would give a bell-curve for results instead of a line, and help average results, but the base problem will still persist, because following die rolls are not affected by previous ones, and need to be in my opinion.

Random is falsely assumed to be interchangeable with Fair or Even much of the time, especially in my experience, in computer games, and I don't think this should be the case.

Interesting argument. I agree from the beginning up to the first sentence in paragraph 2. I also can agree with your bottom line somewhat - although what you mean by fairness is still not very clear to me. If fairness of randomness is defined as not overwhelming good preparation, then I'd ask on the definition of "good preparation" as well.

I see your point in wanting to make the 0-9 throws balance out in shorter terms. Negative correlation perfectly does that - however, the conditional mean would depend on the last throw and in this sense, it is not conditionally fair. Have you heard about techiniques economists and statisticians use when studying time series, say AR(1) or MA(1)? [Sorry about my approach, I've got too used to thinking as such. Blame grad school!]

 

[Beagá]

Fight defensive. Fallback toward your fortresses, bait them into attacking a smaller force then reinforce. It's the tried and true Victoria II method. Do no fight offensively until their morale is low and your numbers greater.

Pretty much this, it´s rule number one. Only attack when you REALLY need to, and those events are few and far between, like when an enemy is sieging an allied province that is on a plain and you have an excellent offensive general - and if it´s before machine guns.

 

[unmerged(624310)]

Interesting argument. I agree from the beginning up to the first sentence in paragraph 2. I also can agree with your bottom line somewhat - although what you mean by fairness is still not very clear to me. If fairness of randomness is defined as not overwhelming good preparation, then I'd ask on the definition of "good preparation" as well.

I see your point in wanting to make the 0-9 throws balance out in shorter terms. Negative correlation perfectly does that - however, the conditional mean would depend on the last throw and in this sense, it is not conditionally fair. Have you heard about techiniques economists and statisticians use when studying time series, say AR(1) or MA(1)? [Sorry about my approach, I've got too used to thinking as such. Blame grad school!]

If I have, its not ringing a bell. I read a shockingly large amount, but lack formal instruction in most subjects. For instance that bit about coin flips came from a book that detailed why certain 'randomized' medical study results were actually fairly useless due to random not being the same as unbiased depending on how you use it.

By 'fair' I mean that number generation is weighted enough to preserve relatively even/balanced outcomes even in smaller samples sizes. Lots of people use Random when I think they wanted something a little closer to averaged numbers or even weighted distribution. Its very possible for Random results to completely favor one side or another, and while we want some of that, your troops may just fight harder or something, but the current spread 0-9 is wide enough that two even forces can end up with one utterly crushed for minimal losses. This is, in my opinion too wide a spread for a system where we can't influence battle outcomes after they start. We get initial conditions only, army composition, generals, where the fight happens, and possible the ability to send reinforcements.

The outcome of something as binary as a battle (and possible a war) shouldn't hinge on something the player has so little control over. Now if the battle system was more involved, I could make choices, like telling everybody to grab some cover and keep their heads down to mitigate a poor roll that's getting lots of troops killed we'd have another discussion, but right now battles are largely spectator sports.