Social Model v3.0

[alms66]

Leaving it as-is for demo 8 is fine by me. I don’t see any harm in discussing the issue now though, so…

I believe from a purely logical stand-point that these units must decay over time if the original problem is corrected. Some of the men will die, others will quit the rebellion, and with the original cause of rebellion no longer in place, very few new recruits would be coming in. Just to be perfectly clear here, by ‘corrected,’ I mean ‘set to exactly what they wanted in the first place.’ So, if a group wants slavery abolished and the government abolishes slavery, what reason is there for rebel units to still exist? (Unless the rebels view capitulation as a sign of weakness and decide to seize control of the government anyhow, though I don’t think we’ve gotten that sophisticated yet) Now, if an existing rebel unit exists and the policy is not set to exactly what they want, recruiting more people makes sense, though it should be at a reasonable rate based on how close the policy is actually set to what they want. Over time, the recruitment rate may not be enough to sustain the rebel unit, and the rebellion effectively ‘dies’ at that point, and the unit should be removed.

As for there being too many units created, there may be, though I’m more inclined to say that the number of units is ok, but the strength of the units is too high. In every rebellion there will be those that choose sides, and there will be many more simply looking not to get hurt or involved. I’ve seen instances where large percentages of an entire tile’s population ended up as rebels, and such large numbers are unreasonable (granted, I intentionally set policies to ridiculous levels to see what would happen). Even the most evil, despotic, and corrupt regimes that have undergone rebellions have had a vast majority of the population be bystanders and not rebels.

From the Government Model:

3) Province's Autonomy Level (PAL): It's a variable describing how much autonomy the local govt has. A low PAL means the local govt mostly just follows instructions from the central govt, while a high PAL means the local govt is left to resolve province problems the way it considers it best. Here it's important to emphasize again that the local govt, no matter how much autonomy has, remains "operative", so central govt policies remain always intact. The more autonomous a province is, a bigger part of the tax collection is kept in it (not collected by the central govt and therefore out of reach of ruler's hands) and also, if autonomous enough, some military autonomy is gained, meaning the ruler can recruit less troops in the province. Depending on the regime, the military autonomy can also imply the formation of local military units called "Feudal Units".
………….
4) Feudal Units: As said in 2), local govt's political structure resembles the central govt's one. If in the latter the combined political power of the military and the ruler is high, then we assume in local authorities there're important numbers of warlords and such. If a province has sufficient autonomy, these warlords "take over", forming their own military forces (Feudal Units) to protect their interests. Feudal Units obey the warlords, so they're not controlled by the ruler (player), but by the AI. They have only a defensive role, fighting foreign invasions coming into the province, so most of the time they remain inactive. The taking over by warlords doesn't imply a secession, so the province remains under ruler's control like any other.

Feudal units are meant to represent a local (provincial) lord’s forces, not any type of rebel unit. So in the event that the player either grants a province autonomy or uses a feudal government system, only then are feudal units created and used. Thus, I don’t understand your comments about feudal units. Unless, perhaps, you mean that rebels currently behave like feudal units (they don’t attempt to capture the capital – which is what rebels are supposed to do).

Also, in re-reading the riot model, I don’t see any indication that riots should cause rebels at all, but that rebel formation is an entirely different event from rioting.

[Mark_Everson]

I'm in favor of reducing the units generated by a factor of 3 or so, and leaving it at that for Demo 8 b/c of the feature freeze. The disbanding of rebel units issue will go on the feature request list when I next update it, and we can discuss after D8 is out.

It would help reduce MM a bit if some of the rebels disappeared when policies changed for the better. The question is how much coding effort would be required, and whether it's worth it on balance?

[LDiCesare]

Yes, rebel formation is different. But it's a matter of PAF. If PAF is low, you get only riots, if it's high, you get some rebels. The riots model is very good and very complex too, so I started by coding a very simplified version of it.
The big point is the current model doesn't provide info for when units are supposed to disappear.
I agree that disappearing units would help reduce MM, but I think it's just too cheap, and the player should do soemthing more in order to get rid of them. On the other hand, having them disappear/decay after time looks ok. It's just immediately vanishing that I don't like. If htey decayed over 5 to 10 turns, it would be ok for me.
As for feudal lords, looking at history in France, there's often little difference between a feudal lord and a rebel.

If everyone agrees we can leave that part as is for Demo 8, let'w do that. Afterwards, yes we should keep in the rebel units of various kinds the information coresponding to why they rebelled. It's not that hard to keep, since I already created a dedicated sub civilization which can store all information for them.

[LDiCesare]

Originally posted by alms66
quote:
Originally posted by LDiCesare
Well, all policies are more or less important, but having 100 figures for each is probably a bit too much, 10 would already be quite big. Anyway, slavery of 57 means you allow some slavery but with some limits. It could be something like not being allowed to kill slaves, allowing them to buy their liberty back. Very low slavery numbers would mean only indenture is allowed for instance. That probably needs explanations and refining, as just figures won't cut it, and 101 different descriptions is a bit too much.

I'll have to second that. Ten-step or less sliders would be much easier to work with. Perhaps a structure like this can be used to define the steps (one possible example of slavery defined)?

code:
<slavery>
<step>
<label&gtNone</label>
<description&gtNo slavery is allowed.</description>
</step>
<step>
<label&gtIndentured</label>
<description&gtOnly indentured slavery is allowed.</description>
</step>
<step>
<label&gtFull</lable>
<description&gtAll forms of slavery allowed.</decription>
</step>
</slavery>


What you see above is just the interface portion, there would have to be some sort of "effect" portion that defined the behavior of the sliders.

Yes, either a <value>50</value> or a continuous progression based on the number of options (I prefer the former for flexibility but the latter is easier to modify).

[Mark_Everson]

In a code commit Laurent said:

Not all PAFs can cause the same effects. In particular, ethnic discontent can lead to rebels but bad policies events shouldn't. So fixed the compound social event to create rebels only when allowed.

Why shouldn't bad policies cause rebels? The American Civil War originated in policy differences rather than ethnic discrimination.

[LDiCesare]

Because they don't in the model. The various PAFs are:
Nationalistic Rebellion Feeling (NRF)
Self Determination Feeling (SDF)
Anti-Discrimination Feeling (ADF)
Bad Policies Feeling (BPF)
Poor Welfare Feeling (PWF)
Revolutionary Feeling (RF)
Replace Ruler Feeling (RRF)
I only coded the Bad Policies and Anti Discrimination ones.
Civil war would probably fall in SDF/NRF.

Then we have:
Events that can be triggered (invoked) are:
Verbally Exposing Discontent
Riots
Revolutionary Forces Formation
Guerrilla Forces Formation
Attempt to Murder the Ruler
Military Coup
Army Betrayal
Declaring Independence
Rebel Duke

I considered what I coded to be Riots and Guerilla forces formation, the latter being caused by ADF and not BPF. But then BPF can cause revolutionary forces formation.
I must eventually check that again and implement the whole series of events.

[demipomme]

Moved from game152

Originally posted by LDiCesare
Well, all policies are more or less important, but having 100 figures for each is probably a bit too much, 10 would already be quite big. Anyway, slavery of 57 means you allow some slavery but with some limits. It could be something like not being allowed to kill slaves, allowing them to buy their liberty back. Very low slavery numbers would mean only indenture is allowed for instance. That probably needs explanations and refining, as just figures won't cut it, and 101 different descriptions is a bit too much.

Why not allow the specific policies. For example under slavery you could have some tick boxes:

Slavery:
Legal []
Legal to kill []
Allowed to buy freedom []
Must be given a wage []

I know this is probably harder to model than a number scale, but personally I prefer numbers with units.

I also think that rather than have "ethic groups" as currently modelled, it would be better to have "social groups". Each social is an indivisible group of the population. Each social group is separated from other social groups by one of the following:

-Race
-Nationality
-Religion
-Class
-Language

The name for the social group is made up of the adjectives for these categories e.g.

Caucasian, Welsh, Christian, working class, Welsh speaking people

The benefit I see in doing it this way is it allows change in the attributes of a group. A group can be killed, or religiously converted, they could adopt a new nationality or language and move from say working class to middle class.

To clarify two of those attributes:

Race is the only attribute that an individual in the population cannot change. It is physical i.e facial features, hair colour, skin colour.

Nationality is what nationality a group feels affiliated with, even if that nation no longer exists. For example many of the migrants to America still refer to themselves as Irish-American or something similar. Their home is America, but their nationality is Irish.

Governments could choose four different attitudes to each Race/Nationality/Religion/Class/Language. These would be: favour, allow, discriminate, social cleansing.

It would be beneficial but amoral for a government to try to homogenise its population.

[LDiCesare]

Why not allow the specific policies. For example under slavery you could have some tick boxes:

Slavery:
Legal []
Legal to kill []
Allowed to buy freedom []
Must be given a wage []

I know this is probably harder to model than a number scale, but personally I prefer numbers with units.

As long as there's a hidden figure beneath for the computations to do, it's ok. But your proposal would make for a lot of check boxes and be very hard to model in terms of negotiation system if everything is binary.

There are social groups too inside each ethnic group. Favoured policies are computed by social group. When you get riots, you can see it as you weill get details saying for example that Workers of EG Persians wanted a policy of whatever.

Race change through mingling with others. I already noted places like La Reunion, but the Indians are not very much like the original Aryas (blonde hair is not very common there). That changes very slowly over time, but it does.

[demipomme]

Originally posted by LDiCesare
There are social groups too inside each ethnic group.

I know.

What I mean by a social group is not a social class. A social group is an indivisible group of people that differ from other social groups by class, race, religion, nationality and language.

As well as each ethnicity being divided into social classes, it is also possible to look at it as each social class being divided into ethnicities.

I accept race does change, but this is extraordinarily hard to model because it's linear rather than discreet. You're either christian or you're not. You either speak english as your first language or you don't. You're either Irish or you're not. These are discreet. Class is discreet with blurred borders. Race is almost completely linear.

[LDiCesare]

What I mean by a social group is not a social class. A social group is an indivisible group of people that differ from other social groups by class, race, religion, nationality and language.

I'm saying that's already the case: Population is divided by ethnicity and each ethnicity is then divided in social classes, leading to social groups. You already have persian workers and macedonian aristocrats in the same square. Since ethnic group is (nationality + religion), you already have that except for race and language.
Race can be added but I'd rather add phenotypes than race (i.e. hair, body colour and other visible traits).So people who are quite different genetically (say a Basque and a Celt) may notlook very different, so if they speak the same, have the same religion, they will not be discriminated against and will mingle more freely than people of different physical traits.
Language has been discussed elsewhere and I agree that it's something we want.

[yellowdaddy]

...Looks like the Language bus is filling up with passengers!
(the "racial" image elements I proposed correspond to the phenotypes idea, in that: all face elements can have a code; can be grouped together in any arrangement to correspond to any ethnicity; and, you can model for racial mixing that has never occured in history - such as an Aranta [Australian Aborigine]colonisation of China or an Iroquois colonisation of Western Europe.)

[demipomme]

Originally posted by LDiCesare
I'm saying that's already the case: Population is divided by ethnicity and each ethnicity is then divided in social classes, leading to social groups.

What I'm really saying is not 'add something new', but 'rearrange what you've got'. The reason I think this is because 'ethnicity' is such a broad word - so if we could get the word ethnicity out of there that would be good.

I am thinking it from the point of view of the screen the player sees. One view could show the nationality demographic, one view could show the language demographic, one view could show the religion demographic, one view could show the class demographic and one view could show the phenotypes demophapics. Rather than see ethnic groups as the first thing, they see social groups, which are distinguished from each other by the five factors already mentioned.

By the way I think the racial phenotype concept is a great idea.

And I'd like to buy a ticket for the language bus!

[LDiCesare]

I've been trying to implement fully the riots model these days.

All PAFs are computed (and stored) at the social class level. In other words, each social class within each Ethnic Group in each province has this set of PAF values:
1 Nationalistic Rebellion Feeling (NRF)
2 Self Determination Feeling (SDF)
3 Anti-Discrimination Feeling (ADF)
4 Bad Policies Feeling (BPF)
5 Poor Welfare Feeling (PWF)
6 Revolutionary Feeling (RF)
7 Replace Ruler Feeling (RRF)

Right now, ADF and BPF are coded.
I checked the others and remembered why I hadn't done them in my first implementation. I'd like to discuss some:

1 NRF: Depends solely on nationalism of the ethnicities. IT can't be fought except by lowering the nationality of said ethnic group (which is something the player can hardly control, but I think it can change with time) or by promoting a new nationality, which is not always a solution. I wonder whether it should depend on a nationality tech too.
2 SDF: Depends almost exclusively on Province Autonomy Level (PAL), which is not coded. I don't know what PAL is supposed to do outside the social model, but clearly it would be up to the ruler to change this value. A high autonomy would increase the risks of riots and the effectiveness of the administration (which is: lower its cost). Problem is we don't have a cost for the administration in yet if I'm correct (since that's bordering econ, I don't know too much). What should the cost of admin be, does it exist yet in the econ code?
3 ADF: Coded.
4 BPF: Coded.
5 PWF: Depends on Per capita income and historical pci. Mark, how do I retrieve a per capita income in the econ code? It also depends on administration effectiveness which depends on PAL. This also depends on social policies value, which increases the need of infrastructure. The point is: How is state (here welfare) infrastructure modelled from an econ point of view? The optimum government infrastructure increases with social policies and decreases with private property, but the way the equations are done is: Administration effectiveness decreases if you increase social policies without increasing infrastructure, which will lead to more chance of riotting. The problem is the cheapest way to stop people from riotting for bad welfare is to reduce social policies if I am not mistaken. There should imo be a direct effect of social policies in reducing the welfare riots factor. Also, from an econ point of view, is there a difference in the kinds of inftastructure held by the government? Or is it a big lump (which would actually be easier to code with and for the player to manage/understand)?
6 RF: Revolutionary feeling is quite straightforward.
7 RRF: A bit like the former.

To sum it up:
On 1) I have some dounts whether it should be present in the early game, unless nationalistic values of the EGs start low and grow through the game.
On 2, I need to know if province autonomy should have effects on the economy (probably reducing corruption?) and how.
On 5 I need 2 pieces of econ information: per capita income and government infrastructure. I'll forage in the code, but if they aren't in, it will be a loss of time so I'd rather hear from Mark. I also think social policy should have a direct effect in the PWF.
On 6 and 7, the RF and RRF mostly need new special bad effects to be coded (e.g. murdering the ruler).

[Mark_Everson]

Hi Laurent:

I am pretty booked today, but here are some quick answers etc. keyed to your numbering. I generally agree with what you say and the connections that you suggest, so if I don't say anything assume that I support your view.

1. I think nationalistic feeling can be just a hard-coded low number for now. Say 15-30% of the full-blown nationalism that is seen today. As we get other things implemented we can make it more versatile.

2. I will add an Administration infrastructure to the econ model and put my guesses at good values into it. We can calculate admin effectiveness from it, and possibly other parameters.

5. PCI isn't implemented yet, I'll try to do it soon, it's not hard.

I should be able to get you my to-dos above by next weekend, maybe earlier.

[Lord God Jinnai]

[SIZE=1]
2 SDF: Depends almost exclusively on Province Autonomy Level (PAL), which is not coded. I don't know what PAL is supposed to do outside the social model, but clearly it would be up to the ruler to change this value. A high autonomy would increase the risks of riots and the effectiveness of the administration (which is: lower its cost). Problem is we don't have a cost for the administration in yet if I'm correct (since that's bordering econ, I don't know too much). What should the cost of admin be, does it exist yet in the econ code?

That seems completely backwards. Higher autonomy means they'd have more control over themselves and less likely to feel the urge to break away. Of course you might never be able to integrate them into your society if its too autonommous.
Also increasing autonomy would lower adminstration costs as there is less of a beurocratic or other form of control. Those appointed are generally more local as well and because of the more flexible nature generally are able to be done cheaper and more in-tune with what the local populace would like.