Population, demography, and recruits

[LDiCesare]

I looked for some thread or model on poplation and found little or nothing. The most complete piece fo information I have is in axi's poll about how much detail we want to go in for population/ethnic group modelling. That is, nothing in terms of models.
I want to code population uprising when enemy troops try to invade. I don't want to have all the population to fight, and cannot use only a percentage of the population to fight, as from one turn to the next, the number of fighting men would artificially grow.
So I want to code population along these lines:
In an ethnic group, keep track of fighting/nonfighting. This I translate into men/women and adult/children. One could also add elders, but I will not because they didn't always stop fighting when they got old, and we are still in ancient age modelling.

What I really want is to say I have f.e. 10000 inhabitants in a square. How many can fight? The military/social/riot models will decide how many do fight when there is an uprising, an army is drafted, etc.
What are the effects on population on the turn following a loss of male population?
How is growth affected?
How many children become mature?
How many mature die?
Does the presence of military units (males) in the square affect the population growth (there will be bastards, marriages with local women...)

State of the code (as far as I understand - Mark and Gary can correct me here-)

1)Population in square is divided by ethnicities. Ethnic groups have a population figure. How both are related is totally opaque to me.

2)Growth depends solely on food. The growth formula is unclear to me. If the method names are to be trusted, excess food simply turns into people.

3)Military units take population from a square. They just take what they need, with no distinction of sex or ethnicity.

4)When disbanded, half of military personnel is added to the square.

Suggestions and needs

1)Population in square to be divided by ethnicities, each ethnicity by sex and age category, age category being either child/adult or child/adult/old. The rationale for having sex population per ethnicity is that when BigBadWolves invade SheepLand, the Sheep male population will raise to fight them. Those who die will probably be replaced by BigBadWolves. When Sheep come to liberate their country, only BigBadWolves will try to fight them. Surviving Sheep will raise to help the liberators. Also, historically, the Saxon killing of Celts in Britain and the Breton slaying of male Armoricans require a male population ethnicity to be followed. This raises the question of: what decides the ethnicity of a child. Father? Mother? Mostly because of conquests, it was father, but there are populations where it is the mother who matters (semites, some polynesians, amerindians).
Codewise, how do Population in squares and EthnicGroup population data member relate? Is some code obsolete or is there a third party class somewhere?

2) Population growth model?
We need mainly food and women to have population growth. Men are needed too, but not as many as women. What is the effect of a lack of men on population growth? What effects marriage policies have on this? (Does strongly enforced monogamy lower fertility?)
Also, how many young become mature, how many mature die (or turn old, and olds die) in one turn?

3) Military would take from mature male population, no ethnicity distinction yet. Long term, ethnicity could be taken into account.

4) All population would be put back to the square, ethnicity to be determined.

Not adressed points.
There are many. I want to point some of these here but NOT discuss them because they won't be coded soon. I just log them, please speak about them only if you think they have to be coded right now for some reason.
There is the effect on social classes. Should they matter as far as army drafting goes (fighting aristocracy, serfs...)
Child death. All children don't make it to adulthood. How many die? The disease factor comes here too.

[Mark_Everson]

Hi Laurent:

Thanks for starting this discussion up! You have a lot of good points. We clearly do need some estimate of a militarily-eligible population, probably broken down by ethnic group.

There is in fact an old fairly-developed model, though it was quite far down. Here is a link to the old Population and Demographics thread.

IMO the old model got a bit too detailed. I think some of your suggestions also are verging on far too detailed. I believe keeping track of sex, and so implicitly which different ethnic-group combintations children come from, could be a huge swamp that doesn't much add to the game. We need to fight Clash becoming a world-simulator at the expense of things the average player will actually get enjoyment from.

I haven't the time to re-read the old thread right now, so I'm not sure how much it has to say about the military-recruitment issues you're focusing on.

In response to your comment that conquerors are all male... I don't disputs that the warriors who did the conquering might have been all male. But througout history there were almost always supply and support personnel associated with an army, and usually many of them were female. I want to resist again this notion that you and Gary have that an army is a purely male population. This leads to the notion that only males settle from an army, and worse the assertion that an army can't settle virgin territory. I belive that is both broadly false historically, and will needlessly complicate the game in a way that the player gets essentially no benefit from.

[LDiCesare]

Proposal

I just wrote an extensive post that was lost before I could post it.
I try again:

Proposal:

Don't separate population by sex.
Separate population by EG.
Separate population by young/mature.

Growth:
Currently, pop(t+1) = food(t). I find this a bit crude, but OK as an upper limit on population.
There is an unused equation in the code which uses Food per Head and a 4% growth factor.

I want to do a growth computation every 5 years.
I separate young/mature as 0-14/15-60 years.
This means over 5 years, 1/3 young become mature and 1/9 mature die.
Reproduction rate should allow for MAX increase of population over 5 years. For instance, in order to double population in 50 years, MAX = 0.07. I propose that figure. The unused code uses 0.04. Proposals?
Considering a flat ages pyramid, 3/4 of population breeds so:
Pop(T+5) = Pop * (1+MAX) = Pop * (11/12 (deaths) + births)
births = (1/12 + MAX) = 0.16.
This gives:
Young(T+5) = Young(T)*2/3 + Mature(T)*0.16
Mature(T+5) = Mature(T)*8/9.
Making 2/3, 0.16 and 8/9 parameters that can be tweaked.
Then we must consider food: The above computed population is maximum population. Compute food per head (FPH) for it and if FPH<1, multiply figures by FPH.

This is a quite simple model. At least simple to code, and with only 3 variables to tweak (MAX growth in 5 years, age for adulthood and for average death of "old age").
I would make all variables global to start with. We can fiddle with adulthood/death age later based on whatever EG/medicine tech we want.

[LDiCesare]

Recruits

Now if we have an adult population, all calls to population from the economy and military (and other models except disease or such) should draw upon it instead of total population. This can disturb existing scenarios a bit as it would lower production by 1/4th. If my proposal is accepted, I will check Dawn and Delenda to make sure they still work mostly the same. That probably means I will have to do things in the economics, or I can provide a getWorkForce() method to be used instead of population which just puts the additional 1/4th in so the scenarios see nothing.
Raising military units will thus have more impact on the economy as it will remove from the active population (1/4th more impact than before).

Then I will be able to decide a proportion of the attacked population in a square will decide whether it is worth revolting when an invader comes by. I would do it only when an army enters the square. What should the proportion of population of a given EG ready to take the arms in order to defend itself be ? Maybe 1/3rd? They would take arms only if by doing so they provide their side odds of 2 vs 1 or more, and would be shabby warriors (revolting roman peasants would be warriors and not legions). Is that 1/3rd figure OK or does anyone has a better proposal? This figure being totally arbitrary.

I will start coding as soon as I can (which can start anytime between monday evening and friday evening), so please give feedback.

In particular I need feedback on how the code EthnicGroup population figure relates to the PopulationData figures. I didn't see any obvious link.

[Simon Loverix]

The demography of the ancient world is very similar to that of the present-day third world: a high rate of child deaths and few people reaching old age. I suggest linking demographic types mainly to infrastructure, to discern developed and undeveloped societies. Technology doesn't matter so much as having access to it.

[LDiCesare]

I don't bother about infrastructure and tech in this implementation. Developped societies are far in the coding so I ignore them for now.
Note that the weeding of all ages equally affects the rough age pyramid unrealistically. The model I propose is not very realistic, and cannot be unless we decide who starves to death, and such. I don't want to go into more realism however, as I don't think we need it, and we certainly don't need it now. I am only concerned by not having a mature population popping every turn out of aehter available for warfare.

[Mark_Everson]

Hi Laurent:

I have pondered this a bit and think that doing any serious demographics coding before demo 7.1 is released is just asking for trouble. A model with young who aren't economically productive will upset the existing balance in the economics model in non-trivial ways. I really don't want to deal with that now, since I've already got too much on my plate. And I don't think anyone else can practically track down all the angles to it in economy right now.

But I have a proposal that I think will work okay with the existing population model. My suggestion draws on some of the good suggestions you have made, and I don't think will give the game any serious instabilities. This would allow us to put off more-detailed demographics till at least D9, which I think is more like it given that there are a lot of things more important than detailed demographics to get in first.

Local defensive army formation proposed model and issues

1. Any auto-generated defensive armies will need a special status so that they will be merged back into the local population when the issue is decided in their square.

2. I think they should be handled as a separate class of unit, called militia or something, that is even worse than warriors in terms of combat effectiveness. If these units are special, it will help in facilitating item 1. Something like 1/5 of the population should be eligible to form these units.

3. I like the idea that the local self-defense forces will only appear if they can do some good. However I think requiring them to attain 2:1 odds to appear is too restrictive a criterion. I would also like to see chance play a role in this, since the locals before mustering would never know exactly the troop strength they potentially would face anyway. My first take on this is to determine the potential local troop strength by adding the strength of militia that could be mobilized to any friendly troops in the square. Then the potential local: attacker strength ratio is evaluated.
* If the strength ratio is below 0.75: 1 no militia will ever be created
* if the strength ratio is above 1.75: 1 the militia will always be created
* for intermediate cases the chance of militia creation grows linearly from 0% to 100%

4. We must ensure that the people don't fight someone they would Prefer to control the square. I'm not sure exactly how to do this at the moment. If all the current scenarios had mono-ethnic civs it would be simple to find out if the local people are disposed to like or not like an invader. However, in delenda I know at least the Romans have several different EGs in their starting civ. If this is the only problem with this proposal I suspect we can come up with something that will work at least temporarily. There might already be something in the social model that will help in this, have to check with Gary, since he's the only one that's aware of what works now and what doesn't in that model.

5. An approach like this could seriously change the balance in a scenario like delenda that has large cities. I don't think that's entirely a bad thing, but we need a keep in mind that some weak tuning of the balance of forces might be necessary if this change is made.

I do think that something like this is worthwhile to do for D7.1! However, if you accept my notion of putting aside the demographic modeling for a later time, we should probably move this discussion to the military thread.

[edit: include notion of proper timing for more-detailed demographics]

[LDiCesare]

I agree that unproductive young would upset the economy. Tweaking the military to get Delenda even fairly working would require lots of work too. I'd rather make the whole thing a D8 development. It will help get D7.1 out faster. You'll see when I increase the upkeep costs of the armies back in Delenda, it changes a lot of things. This would be even worse. So I say: Let's get D7.1 out fast so I can start this for D8. (It is no use calling D7.1 7.1 if it is to come 6 months after D7 anyway.)
As for militia being somewhat random, this sounds good. (Wasn't it someone named Mark Everson who asked armies not to attack if they were less than 2:1 or something? ).
The kind of units spawned will depend on the tech level (available units). I will choose the best unit available that will be somehow marked as militia. Thus I can use warriors or something more pitiful.
I agree with preferred EGs. I doubt Gauls or iberians had any clear-cut preferrence for either romans or carthaginians, so even a simple EG matching with ruler EG could do if there is nothing better available.

[Mark_Everson]

Bonjour Laurent, I agree with your points.

Originally posted by LDiCesare
As for militia being somewhat random, this sounds good. (Wasn't it someone named Mark Everson who asked armies not to attack if they were less than 2:1 or something? ).

That name does ring a bell... How about setting the lower threshold at 1.25:1, and the upper at 1.75:1. I had been just trying to get in the notion that the militia battles should be rather risky.

[Gary Thomas]

For the sake of getting rid of D7.1 I really think we should leave militia out completely.

Then, for D8, use Hans Delbruck's figure of 1/8 of the population.

Incidentally, the current code holds a total population for a square, and a distribution by ethnicity. I am not really convinced that extending this model to include sex and age ratios is going to add anything to the game except unnecesary complexity. My suggestion would be to add another parameter (akin to recruitment and dispersion) called martiality or some such, being the proportion of the population who would join the militia. If extra realism seems desirable, make martiality a characteristic of ethnicity. So then those ethnic groups who favoured the civilization would provide the appropriate proportion to the militia, or if they were conquered to the invaders or partisans. The martiality would be fairly high for, say, Sikhs, and low for, say, Amish. It would be 100% for Dorsai.

I doubt that any more detail than this will contribute anything to the game.

Cheers.

[LDiCesare]

I think not managing population by ethnicity but by proportion will be a nightmare. It will lead to lots of computation just to deal damage.
F.e. A square of 1000 EG A and 1000 EG B. Proportion is 0.5A, 0.5B, population 2000. Suppose 100 A die in a revolt. Instead of having 900A,1000B, we have to compute 1900 total, 0.45A, 0.55B (making sums and divisions and keeping one additional variable).
Also, does 100% martiality mean 2-year olds attack their opponents with clubs?
Having a young/mature distinction (I don't mind about sex) allows to slow growth a bit and to force the player to wait for 2 or 3 turns before they can grow an army out of a square. Mature has the advantage (or disadvantage) to have an impact on the economics. Having an additional militaristic factor in the EG's is OK for me, but longer-term IMO.

[Mark_Everson]

Hi Gents:

Gary's stated approach is pretty much what I was hoping to do on demographics. That is, not much! I think more detailed demographics would complicate many parts of the game, especially economics, in a way that doesn't give players much in return. Why have all those headaches for nothing?

On militaricity I would go perhaps a tiny bit further than Gary's stated position, which I get to in item 3 below. To state my position, which is mostly what Gary had, but responding to some of Laurent's points:

1. Have a fraction representing the background level of militarily capable individuals. This is the previously-stated Militaricity or whatever. To crudely account for demographics it could have an absolute maximum of 2/3 or some such. That would effectively exclude young and very old.

2. Militaricity certainly could depend on EG, and I think that adds some good features.

3. For each EG in each Square (that is, each Population Element IIRC) the fraction currently available should be kept track of. Call this Available Militaricity for the moment. When militia are formed, or armies conscripted, the maximum population avaiable would be the Available Militaricity (AM). That way if a militia were formed in a square, and was subsequently slaughtered, immediately thereafter there would be no individuals left suitable for armies. Each time the population grows a value TBD would be added to the AM until it reaches the EG's Militaricity. In this way militarily useful individuals could be depleted but would replenish with time.

4. To a non-trivial extent men suited for military service are also those most likely to cause trouble in other ways, like riots. We may also be able to use AM in the social model as the number of potential hot-heads that might riot. I have read some speculation that one reason post-revolution France was so relatively internally peaceful, even with the wrenching social changes, was the the levee en masse put most of the potential rioters in military service! What do you guys think?

It seems to me that 3 doesn't hugely complicate things, and yet gives a natural limit to local rebellions if they are ruthlessly stamped out.

[Gary Thomas]

I think not managing population by ethnicity but by proportion will be a nightmare. It will lead to lots of computation just to deal damage.

Can't see the logic of this. Population is dealt with as a proportion now. The Distribution class has lots of utilities for dealing with changes. This code is there and works now. It is completely general and is used for other things that require a distribution, as happens a lot in the social model. Changing this will hugely complicate the management of populations.

Also, does 100% martiality mean 2-year olds attack their opponents with clubs?

No, they slip poison into the drinks of the conquerers. You have to read the books to appreciate it. Anyway, we can't have Dorsai, they are too far in the future.

In general I agree with everything that Mark has said, except that he did not specifically agree about my preference for no militia in D7.1. If we keep on sneaking in extras, D7.1 will never happen. It was supposed to be a minor improvement to add ship movement. Also I am currently keeping two versions synchronized (D7.1 and D8), and would prefer not to have to.

On the other hand, Mark's point 4 looks like a good candidate for D13.

Cheers

Gary

[Mark_Everson]

Originally posted by Gary Thomas
In general I agree with everything that Mark has said, except that he did not specifically agree about my preference for no militia in D7.1.

I agree D7.1 shouldn't include the militia stuff. We should get out D7.1 as soon as practical. Sorry my time limitations have been part of the slowdown...

[LDiCesare]

First of all, I also agree to do nothing for D7.1.
I mean, nothing about militia. I still have to send some stuff about units costs, and I will be done.

Now Mark said:

3. For each EG in each Square (that is, each Population Element IIRC) the fraction currently available should be kept track of. Call this Available Militaricity for the moment. When militia are formed, or armies conscripted, the maximum population avaiable would be the Available Militaricity (AM). That way if a militia were formed in a square, and was subsequently slaughtered, immediately thereafter there would be no individuals left suitable for armies. Each time the population grows a value TBD would be added to the AM until it reaches the EG's Militaricity. In this way militarily useful individuals could be depleted but would replenish with time.

That is what I want. But I want it by EG because a mixed-EG square with A's and B's will not revolt an masse against A conquerors. This means that when B's Available military suffer losses, I must keep track separately of A and B's AM, which I must then confront to previous total population and ethnic proportions in order to compute the new proportions and total population.

That is something I consider more complicated than keeping per EG figures of population and AM. Now if the social model requires these distributions, we can keep them, but I'd rather have them as output computed on demand than something I compute everytime I deal damage to the population. I would like to add the fact that if you have militaricity depend on EG, there is also something I would like to depend on EG, which is birth rate. In modern times, Europe (and probably US) demographics vary vastly per EG. This with immigration leads to right wing governments being elected in some countries, so I think it has an impact.