Riots Model v.1

[axi]

The Riots Model

I. Introduction

This model uses information from the social, government and economic models to create discontent measures in society. Discontent is then used to generate several different events from simple protests to revolutions and other major social events.

It's highly suggested to be familiar with the govt and social models before reading this.


II. Model Overview

The model is really simple. Discontent is measured in several ways taking into account different factors and unhappiness' driving force. Discontent moves people to action with a specific intention depending on the source of discontent. Actions take the form of "events" happening in the civ with a probability given mostly by the discontent level. Events give the player info about what's wrong in the civ, but can actually lead to dramatic changes, most of them in the govt (affecting the govt model), but even beyond, like creating new civs when actions pursue independence.


III. Clash Society and Discontent

As defined in the social and govt models, Clash society is divided in various parts. First, on an Ethnic Group level. Each EG represents a tribe (a nationality) living within the civ. In the roman empire, FE, you can find romans, greeks, celts, hebrews, etc. Second, on a Majorities/Minorities level. Minorities are all those ethnic groups discriminated because of their nationality or the religion they follow. Obviously, all not discriminated EGs are within Majorities. Majorities, in essence, are those people who see the civ as theirs and usually only one EG (with its religion or secularly) is within majorities. In the roman empire example, only romans are within Majorities. Third and finally, on a class level. Majorities population is said to be divided in classes: the aristocracy (Upper Class-UC), the Church (Religious Class-RC), the common people (Lower Class-LC), the high officers of the military (Military Class-MC) and the most important govt employees (Bureaucratic Elite-BE).


III.1. Pro-Action Feelings
If you pick a man in the civ, his discontent will depend on the position he has in society and the province he lives in. The latter appears because conditions aren't homogeneous through the civ. In order to create good modeling for all kinds of events (revolutions, riots, etc), discontent won't be treated simply as a crude unhappiness level. Instead, different sources for discontent will create different "Pro-Action Feelings". A Pro-Action Feeling (PAF) is the feeling that something needs to be done and actions are mandatory to achieve a specific goal. A low PAF means you feel some things are wrong, but not too bad to do anything about it, so you'll stay at home. High PAF means you believe you have to go outside and fight for what you want, somehow. There're several PAFs, each having its own particular intention depending on the part of society considering entering in action and the level at which actions are supposed to be needed.

Each Pro-Action Feeling is described now in the particular goal it's trying to accomplish, the type of actions ad hoc to what is wanted, the part of society willing to subscribe to it and the level (province or civ) at which it is needed:

III.1.1. At the Ethnic Group level

Nationalistic Rebellion Feeling (NRF): The tribe's feeling that they should have a civ and govt of their own instead of living under other civ's wings. Of course, NRF is null for the ethnic group with matching nationality with the civ's nationality. Actions this feeling encourages are verbal protests, street protests, riots and formation of guerrilla units among the ethnic group's population to achieve independence.
NRF is handled at the province level, having one NRF for each present Ethnic Group.


III.1.2. At the Classes level

Better Administration Feeling (BAF): The feeling that the govt needs to do better. That is, improve the quality of life of majorities, provide more and better solutions for common problems and use better govt policies. Actions to achieve this include verbal protests, street protests, riots and legally replacing the ruler. BAF is the closest concept in this model to the simple idea of unhappiness. As an example, BAF would be the only relevant Pro-Action Feeling in a nowadays 1st world country.
BAF is computed at a province level for the Lower Class and at the civ level for all classes.

Replace Ruler Feeling (RRF): The idea that the ruler needs to be replaced because he intends to make changes in the govt that are seen as a danger. Actions to achieve this include verbal protests to the ruler, street protests, riots, legally replacing the ruler, military coups and attempting to kill the ruler.
RRF is computed at the civ level for each class.

Revolutionary Feeling (RF): The feeling that a radical change is needed in the very govt structure and the only way to achieve it is overthrowing it. RF rises when enough people supports an ideology too different to what the govt regime is like. Actions to achieve this are formation of revolutionary forces among population, military coups and attempts to kill the ruler, depending on political classes.
RF is computed at the civ level on a class and ideology basis and at the province level for the Lower Class.


III.1.3. At the Majorities level

Independence Feeling (IF): The feeling that the province needs to go independent from the civ and become a new civ. Actions this feeling encourages are protests, riots, asking the central govt to give independence and, when high enough, a declaration of independence.
IF is computed at the province level.


III.1.4. At the Minorities level

Anti Discrimination Feeling (ADF): The feeling that they don't want to be discriminated anymore or at least receive a better treatment from the govt and majorities (like abolishing slavery). Actions include verbally asking no discrimination and better treatment and riots.
ADF will be handled at the province level, having one ADF for each minority ethnic group present.

There're two feelings above leading to independence (IF and NRF) and before continuing a short explanation on the difference between them should help: IF is specific to a province and can only be supported by majorities, while NRF is specific to a people (tribe), regardless of majorities/minorities status or province. The american declaration of independence would be a result of IF: english colonists (majorities in the english empire) wanting a particular province of the empire (american land) to become independent. Therefore, success in independence attempt only matters for America, not in Africa, not in India, and matters only for english in America, not for native americans, who keep being a minority. NRF is a matter of nationality. A tribe with its own identity living under other civ's rule in one or more provinces. India's independence would be a result of NRF. In this case a minority wanting their tribe free. Yugoslavia's civil war would also have been triggered by NRF. In this case one of the majority ethnic groups, serbs, in a multiethnic civ wanting to have its own civ, Serbia. An invaded country, like France during WW2, has french people with a high NRF, but there's no IF in France because french aren't majorities, germans are.

What I'm trying to say is there're sufficiently important differences in different struggles for freedom in history that the meanings of the concepts IF and NRF (and even Revolutionary Feeling, which covers struggles like french revolution) should be clear so we can have the right outcomes.

IV. How the Model Works

The model computes Pro-Action Feelings at the corresponding level (civ or province). Some of them must be computed each game turn and other need an update less frequently. Equations for each PAF are given in Appendix 1, but in the following section you can see what variables are included in those formulas (mostly coming from the social, govt and economic models) and the effect they have. Once PAFs are computed, they're used to calculate probabilities for Events. Events and their consequences are listed in section IV.2. Equations for computing Event probabilities are presented in Appendix 2. Every time an event occurs, the govt variable "Empire's Stability" is reduced proportionally to the importance of the event.

That's all what the model does.

IV.1. Variables and their effects in Pro-Action Feelings
The objective of this section is explaining how each PAF is going to be calculated and the events it can trigger. In general, each PAF has its own input variables, but some PAFs can be increased by other PAFs. Equations are in Appendix 1.

Nationalistic Rebellion Feeling (NRF)
We pick an ethnic group in the civ. If it has the same nationality the civ has, its NRF is zero, as per definition. If not, NRF is computed assuming the following: The tribe (ethnic group) is more willing to rebel the more nationalistic it is, so cultural attribute "Nationalism" from the social model is used here. NRF is also increased if people, in a given province, feel the province (land) belongs to them, so cultural attribute "Land Connection" is used. LC is specially helpful simulating the fact that a migrating tribe into your civ won't start up a fight on arrival, FE. Using these two we get the "basic" NRF. This basic measure can be...
1. Increased if the ethnic group is a minority. The worse the treatment minority receives from the govt and majorities, the higher NRF is. This means NRF increases the higher ADF is for the specific ethnic group.
2. Decreased if the tribe sees itself as a minority in terms of demography in the province. It's just unrealistic for a very small group of people to consider a rebellion. (demographic info)
3. Increased by high poverty. (PCI_minorities from the econ model)
4. Increased or decreased by the civ's govt stability. People can take advantage of an apparently unstable govt or, on the other side, they can consider valuable a stable empire (pax romana effect) and prefer not to rebel. Govt model's "Empire's Stability" variable will be used here.

Finally, Administration Effectiveness Level (AEL) in the province (from the govt model) is the last factor. AEL represents the govt capacity to actually run and control the province. AEL is a double-edged sword. If low, civ's control is really nominal over the province, so the tribe might find it easy to rebel. But the higher AEL is, the more evident it is for the tribe that they're under somebody else's rule. AEL will also make more effective any type of discrimination, so if the tribe is a minority, the higher this factor will count in NRF.

A comment: Using demographic info relative to total province population (point 2 above) gives the game another nice ingredient "for free". If the player can use tools to move population from one province to another, he'll be actually reducing chances of rebellions. This means forced migrations are introduced in the game with a clear and realistic objective, adding to Clash another world history common practice.

NRF Value is used to calculate probabilities for Events...
Verbally exposing discontent
Street Protests
Riots
Guerrilla forces formation


Better Administration Feeling (BAF)
The following effects are in BAF:

• People in majorities will look at the Govt Profile and depending on class and supported ideology, will compare govt policies and political structure to what they desire. The greater the difference, the higher BAF is. (info taken from the govt model)
• The greater the poverty in majorities, the higher BAF is. (info taken from economic model: Upper Class PCI and Lower Class PCI)
• Ineffective administration in a given province (low AEL)

BAF Values at the civ level for each class are used to calculate probabilities for Events...
Verbally exposing discontent
Legally replacing the ruler

BAF value at the province level (Lower Class BAF) is used to calculate probabilities for Events...
Street Protests
Riots


Revolutionary Feeling (RF)
A RF is computed for each present ideology and for each class, representing discontent of people supporting the given ideology caused by a Govt Profile too different to this ideology (info taken from the govt model). This calculation is made at the civ level, valid for all provinces. At the province level, RF can be reinforced for the Lower Class by a bad economic situation (Lower Class PCI in the province from the economic model).

RF Values at the civ level for each class are used to calculate probabilities for Events...
Military Coup
Murdering the ruler

RF value at the province level (Lower Class RF) is used to calculate probabilities for Events...
Riots
Revolutionary forces formation
Army Betrayal


Replace Ruler Feeling (RRF)
Each political class compares its political desires with the Ruler's Govt Profile. RRF increases the more different they are. The calculation is made at the civ level for each class. (info taken from the govt model)

RRF values at the civ level for each class are used to calculate probabilities for Events...
Verbally exposing discontent
Street Protests (in random province)
Riots (in random province)
Legally replacing the ruler
Military Coup
Murdering the ruler


Independence Feeling (IF)
IF is computed at each province. The core of IF is discontent produced by a "wrong" equilibrium between province's local govt and central govt. This means people want their local govt to have more power over province decisions than the central govt, and this desire increases with the province's level of isolation. For provinces near the capital province govt vs central govt is irrelevant for people, but the more isolated the province is, the more relevant this is because people feel central govt is less sensitive and effective to solve local problems. Two govt model's variables are used to produce this effect: Province Isolation, which is a measure of isolation considering distance and communications and transportation techs available; and Centralization, which is a province level variable saying how much power the central govt has over local decisions. This is the most complex decision the ruler would have to make for far away provinces, because Centralization will be a double-edged sword. The ruler might see as a good decision to give power away to local govts in order to keep IF low, but if Centralization is too low, it'll make evident the province is mostly autonomous, so people will realize there's no point in keep being part of the empire. IF can be increased by BAF and RRF, because IF is a feeling majorities can have and BAF and RRF are majorities discontent in other aspects.

IF value will be used to calculate probabilities for Events...
Verbally exposing discontent (asking for independence)
Street Protests
Riots
Declaring independence


Anti Discrimination Feeling (ADF)
ADF is computed at each province for every minority ethnic group present. It is a base value representing discontent for being a minority plus an increasing discontent caused by the kind of treatment given by the govt and minorities. To compute the latter, we use variables Ethnically Discriminated, Religiously Discriminated and Slavered for the given ethnic group taken from the social model and we use variables Ethnic Discrimination, Religious Discrimination and Slavery from the govt, taken from the govt model.

ADF values are used to calculate probabilities for Events...
Verbally exposing discontent
Street Protests
Riots

IV.2. Events

All probabilities for events are decreased if the ruler uses internal intelligence services as defined in the govt model. This services supposedly work trying to destroy people's capabilities of threatening the establishment.

All probabilities for events can be increased by international monetary aid, so you'll be able to give funds to help uprisings in other countries. Monetary aids should be handled by the Diplomatic Model.

Verbally Exposing Discontent
Just a message to the ruler where the text depends on the Pro-Action Feeling calling the event. An example from NRF would be "Mighty King, could you please give liberty to our people, the Scottish?". The govt model's "Civil Rights" variable will affect all probabilities for VED, so this people's expressions will be uncommon in tyrannies.

Street Protests
A peaceful demonstration by the Lower Class or Minorities in the streets demanding something given by the PAF calling the event. It is really the same as a verbal protest, since the player will only see a message about the protest in a given province with no other implication. The thing is, Street Protests have a proportionally lower probability of happening than Verbally Exposing Discontent, so the player will have a notion that the corresponding PAF has reached a higher level.
The govt model's "Civil Rights" variable will affect all probabilities for Street Protests, so this people's expressions will be uncommon in tyrannies.

Riots
A street protest becoming violent. In this case not only a message would appear, but a random amount of infrastructure is destroyed in the rioted province. "Civil Rights" WILL NOT affect probabilities for Riots, so in tyrannies you'd mainly see calm or sudden explosions of violence.

Legally replacing the ruler
When a class has 35% or more nominal political power (govt model), it'll be said the class has a legal right to replace the ruler. The equation to compute this event's probability considers that, so a class not having this much political power would never attempt this action. When the ruler is replaced, the Ruler's Govt Profile in the govt model is made equal to the class mentality. The player will keep playing from that new political position as the new ruler.

Revolutionary forces formation
This is the solution the Lower Class sees if too annoyed and if it's unable to legally replace the ruler. In the province calling the event, a number of Revolutionary units (military task forces) are produced depending on the majorities population in it. These units will fight for control over the civ, aiming to conquer the capital so they can change the regime. The presence of these units will reinforce the Revolutionary Feeling, so if not held, the situation can go worse and worse. See Appendix 3 for more on Revolutionary units behavior.

Army Betrayal
When a given ideology is supported greatly by the Lower Class and the ideology has a high Revolutionary Feeling, it's possible for some troops (that, per definition belong to the Lower Class) to betray the ruler and become revolutionaries. In that case, some task forces in the province calling the event change status to revolutionaries and revolutionary AI takes control of the unit.

Murdering the ruler
This is the solution the Upper Class, Religious Class or Bureaucratic Elite Class might come up with if too annoyed and if unable to legally replace the ruler. When the event is called, it's assumed as an attempt of murder with 50% chance. If failed, a message is displayed like "Attempt of murdering the ruler, busted! Conspirators captured!". This is only to give more flavor and to allow the player to realize how bad things are. If the murder attempt is successful, the ruler is replaced. The murdering class has about 70% chance of putting in charge someone of its line. If so, the Ruler's Govt Profile is converted to the class mentality and the player resumes the game from that new political position. If not, the player is allowed to choose freely a new Ruler's Govt Profile and resume.

Military Coup
If too annoyed and unable to legally replace the ruler, the Military Class might produce a military coup. When this happens, the military may or not choose to remain in power, which is decided looking at the MC mentality share not influenced by the overthrown ruler. In that mentality, the greater MC political power appears, the greater the chance for remaining in power. If MC decides to stay in power, the Ruler's Govt Profile is converted to the MC mentality and the player resumes from that new political position as the new leader of the military junta. If not, the player is allowed to freely choose a new Ruler's Govt Profile and resume from there as if the MC just gave power to other classes.

Guerrilla forces formation
In the province calling the event, a number of Guerrilla units (military task forces) are produced depending on the ethnic population in it calling the event. These units will fight for control over the province, aiming to liberate their people from the civ's control. The presence of these units will reinforce the Nationalistic Rebellion Feeling, so if not held, the situation can go worse and worse. See Appendix 3 for more on Guerrilla units behavior.

Declaring Independence
When the event is called, a new civ with a new nationality is created with the province as the capital province. Contiguous provinces belonging to the same "motherland" are checked in order to know their Independence Feeling. If high enough, they're considered as revolting too and annexed to the new civ. Majorities population in all these provinces change their nationality to the new one. An army is created from population to defend from a possible strike back from the motherland. Tech info is copied from the motherland. An AI character is created for ruler based on majorities culture and a Ruler's Govt Profile is created for him based on his characteristics. The Govt Profile is created taking motherland's political structure as a start and finding the political equilibrium.

Appendix 1: Equations for Pro-Action Feelings
NOT READY YET


Appendix 2: Equations for Events Probabilities
NOT READY YET


Appendix 3: Revolutionary and Guerrilla Behaviors

Revolutionary Forces
All units created of this type will pass to AI controlling. Each unit needs a revolutionary status (to make the distinction with guerrilla) and should store the ideology that created it. The latter is made in case of more than one ideology is producing revolutionaries.

Revolutionary forces want to take the capital. AI should be able to see that as the final goal, allowing forces to take other provinces or objectives as previous steps as needed. If AI evaluates the final goal is really difficult, it should change it to "create other civ", trying to conquer weakest provinces and create a new civ there.

If the goal is not changed and a revolutionary unit enters the capital, it's assumed as a revolutionary victory and the govt and ruler are replaced. Ideologically Negotiated Policies in the Govt Profile are made equal to the ideology the revolutionary unit has and Directly Negotiated Policies are computed using the normal govt model's Negotiation Procedure. The player resumes the game from that new political position. All revolutionary forces in the civ are deleted.

Any time a player has revolutionary forces in his civ, he can call "a loyal people's army", which in essence transforms the revolution into a civil war, although the change of name has no consequences and just adds flavor. When the ruler chooses to do so, military forces are created among majorities population in all provinces depending on province's demography and the support the govt and the ruler have, which is info the govt model has. As a cost, govt model's "Empire's Stability" is strongly reduced. The player will have to decide if the call will really help.


Guerrilla Forces
All units created of this type will pass to AI controlling. Each unit needs a guerrilla status (to make the distinction with revolutionaries) and the nationality that created it. Guerrilla's goal is to take over the province where they were created. If successful, the ethnic group will try to form its own govt. First, the game engine will check if there's already a civ somewhere with matching nationality. If so, the province will pass to that civ's control. If not, the ethnic group will try to form the govt (civ) there. To simulate the difficulty of this and to give the player some chances to recover control, we'll say the govt formation attempt has a 30% chance per year. If successful, a new civ is formed with the ethnic group nationality having the conquered province as capital province.

[Lord God Jinnai]

Axi: This is a good model and goes into descent level of description on Riots and what causes them.

Anyway, as with any new model, it's not perfect. Here's what I could see changed:

-----
Third and finally, on a class level. Majorities population is said to be divided in classes: the aristocracy (Upper Class-UC), the Church (Religious Class-RC), the common people (Lower Class-LC), the high officers of the military (Military Class-MC) and the most important govt employees (Bureaucratic Elite-BE).
-----
First off you're forgetting a class, the merchant/middle class. These weren't considered upper class by the upper class and were much better off than the lower class. The differance was their wealth was in money and trade, not land and power. Also some classes like religous and gov. employees will only exist within certain types of gov.

I like the PAF system you descibed, atleast in general...haveto see the details (i'm typing this as i read the model).

NRF could still be used for the civ of the ruling ethnic group. They may want their own country, or a minority of them do.

Why are BAF & RF computed differntly on differnt levels (ie province and civ-wide)?

-----
Legally replacing the ruler
When a class has 35% or more nominal political power (govt model), it'll be said the class has a legal right to replace the ruler. The equation to compute this event's probability considers that, so a class not having this much political power would never attempt this action. When the ruler is replaced, the Ruler's Govt Profile in the govt model is made equal to the class mentality. The player will keep playing from that new political position as the new ruler.
-----
OK. A ruler should still have some chance in not being replaced, even if he/she can legally be replaced. If you just "replace" the current ruler and nothing else, there's no point in it. Also, legally replacing is only possible in certain gov types.
-----
Murdering the ruler
This is the solution the Upper Class, Religious Class or Bureaucratic Elite Class might come up with if too annoyed and if unable to legally replace the ruler. When the event is called, it's assumed as an attempt of murder with 50% chance. If failed, a message is displayed like "Attempt of murdering the ruler, busted! Conspirators captured!". This is only to give more flavor and to allow the player to realize how bad things are. If the murder attempt is successful, the ruler is replaced. The murdering class has about 70% chance of putting in charge someone of its line. If so, the Ruler's Govt Profile is converted to the class mentality and the player resumes the game from that new political position. If not, the player is allowed to choose freely a new Ruler's Govt Profile and resume.
-----
This may be the case that the murdering party can have sway over the government policies, and maybe even who gets put on the throne, but it is usually a member of the current ruling family who is seen as easily controlled. Why the ruling family? Because you're less likely to start a revolution from the supporters of the ruling family, espically after the leader was just murdered and a new family took over, things would seem kinda suspicious about the new family.

Finally on another note. The character model will definatly interact with this model. Here's the thing though: In the character model, each person has their own personality and one personality may be to rule his own civ and cause a revolution that way, ie declaring his province independant. This has happened before and should also be incorperated.

[Lord God Jinnai]

I almost forgot...

You explained how events can decrease an empires stability, but not how to increase it.

[roquijad]

LGJ:

1) We did chose to not include the middle class. Main reason is simplicity and that we don't see much changes in game terms when adding them. Anyway, if any model feels the greatest impact if adding it, it'd be the govt model. The riots model would just adjust to it w/o much pain.

2) "NRF could still be used for the civ of the ruling ethnic group. They may want their own country, or a minority of them do."
I'm not sure what you mean here. I guess something like the ruling ethnic group wanting to get rid of other tribes in the civ. If so, that effect will be achieved by the govt and social models. In the govt model ruling people can define an Ethnic Discrimination level affecting other tribes in the civ. When ED is high enough, other tribes can be mistreated and will be encouraged to migrate, which is covered by the social model, although not developed yet.

3)"Why are BAF & RF computed differntly on differnt levels (ie province and civ-wide)?"
They're not computed differently, but at the province level the computed Lower Class civ-level BAF and RF are slightly increased depending on Lower Class PCI. Having LC BAF and RF at the province level help deciding where a street protest or riot will ocurr. Other classes only act at the civ-level because their actions (murdering ruler, military coup, verbal protests, replacing ruler) happen at that level.

3) "OK. A ruler should still have some chance in not being replaced, even if he/she can legally be replaced. If you just "replace" the current ruler and nothing else, there's no point in it. Also, legally replacing is only possible in certain gov types."
The chance of being replaced is incorporated, because "replacing the ruler" is an event happening with a given probability, not automatic. When a ruler is replaced, it's not just a "replacement". It has a strong impact on the govt model as explained refering to "the Govt Profile" and the "Ruler's Govt Profile". Check the govt model for more.

4) Character model and riots model interacting: For sure. Please read also the govt model and let's start to talk about possible interactions. Why don't you open a new thread for that when done reading these two models?

5) Empire's Stability increases by default each game turn slowly. So, if the civ suffers few events, the feeling of stability increases.

Rodrigo

[Lord God Jinnai]

When reading the social model i throught of something you didn't incorperate. Religious revoultions/discontent. Religion doesn't ness haveto be for 1 ethnic group. This would also give good reasons for having multiple GWRs in civs and such.

[axi]

Okay, now that you Americans are asleep, it's time for us Europeans to strike!

Disclaimer: I did not write the models, Rodrigo did. I only helped at the govt and riots (which at the time were one) and posted them here. There are many comments I would like to make too.

1) Classes: We have discussed introducing the middle class before. Although this might be more useful in the govt thread, I quote part of our discusions, which are self explanatory.

quote: axi- Upper Class (UC): The rich people, the aristocrats, the landlords, the bourgeois, the employers. Their function, differentiating them from the LC, is that they privately own all means of production not owned by the state (as defined by PP). Therefore, them and the state are the only producers of commodities and employers of labor. They produce labor too, but it is used only in the tertiary sector, for the production of services. rodrigo- Here's one of the major differences between you and me regarding the economic issue. I don't see the LC as having no ownership over production means. UC has of course a lot more, but doesn't control them all (excluding 1-PP). However, no matter the approach taken UC definition should keep being only "the very rich people" regardless of the richness source. What I'm trying to say is we must leave room for "bureaucracy richness", that is, people in the UC whose richness come from being in elite position in the bureaucratic apparatus. If the bureaucratic apparatus is wide enough, then an important part of the UC can be formed by these elite bureaucrats. That was the case in some period in the chinese history or in the USSR. So, we should keep away from the UC definition the source of its richness and only refer to it as the rich people. All what has to do with bureaucracy in this model is still TBD, but we have to leave room for it. axi- This is THE major issue of our disagreement. If you do not provide the UC and the LC with some differences of essence, it is very difficult afterwards to come up with real differences in their political outlook. That's why in the tables showing each class's ideological and cultural inclinations you have them all agree in almost everything, while I have tried to create some controversy. What you are suggesting is that the only thing that matters in class distinctions is the income and you are disregarding the economic role of each class in production. In my own proposal, the two classes have totally different economic functions. Of course this is not the whole picture either. No society is ever so strictly divided among capitalists and proletarians, those who provide the means of production and those who provide the labor. There is always another set of people who provide both and their income derives from both, in various proportions. This class is a form of alloy between the UC and the LC, following the rules of chemical mixtures (cantilever rule). I am refering of course to the Middle Class. We could always introduce it as a class into our system, with all the complexity it would cause, or divide it among the UC and the LC. Either way though, it is imperative that we should define the UC and the LC according to their roles and not according to their income. I am not suggesting that income has nothing to do with this either. For example we could use the following logic: For each society and for any given time we can define a 1st limit in the income, above which a person will not consume all of it but will start investing his little surplus and turning it into capital, thus starting to earn profits as well as wages. Of course consumption is rising as income does, but it is covered partly by the wages and partly by the profits. So the person in question may choose to reduce his working hours, thus reducing his wages. If we increase the proportion of profits over wages we will get to a 2nd limit, where the capital that the person in question posseses is able to employ the labor that he is producing, so he can choose to employ himself, if he finds it profitable enough. Above this, he will be able to employ both himself and others, but he can always choose to keep beeing employed by others. In order to cover his consumption though, his profits will still not be enough. If we proceed further, to a 3rd limit, where profits alone will be able to cover consumption, this person will have no reason to work at all, and will then become a true capitalist, working only if it pleases him. All three limits are irrelevant of what a person actually does with his career; they are relevant with what he is able to do with it. They only depend of the relation between consumption and income, which is characteristic of the society we are referring to. These 3 limits define 4 groups of people, of which the first is the true LC, the last is the true UC and the other two are actually the Middle Class. Using the 2nd limit though, a distinction can be made upon them, and the 2nd group can be considered adherent to the LC (since it cannot employ itself, so it is dependent), while the 3rd group can be considered adherent with the UC (since it is independent). The funny thing about this is that the exact value of this second limit varies greatly throughout the 3 sectors of the economy, since it depends from the Labor to Capital Ratio (LCR) used in each sector (So that a farmer can employ himself much easier than an industrial worker). For an economically independant nation, the PCI should be somewhere above the 2nd limit. I repeat that this issue is the most important of all and that it must be solved either among us, or, better still, in the forum! ......... axi- What I really don't understand is the functional difference between a bureaucratic elite, a nobility and a capitalistic elite that makes such a determination imperative. You will have to explain this to me. rodrigo- Nobility and capitalistic elites are basically the same, so there's no need to do anything there. We need the bureaucratic elite because "communist" regimes like in the USSR in which the govt didn't really want to build a welfare state, but to maintain an elite, will be impossible to simulate if we keep the definition for UC as the "owners of the production means". In the USSR the owner of production means was the State ruled by the elite, but not the elite itself. Therefore, this elite wanted to keep the USSR as a centrally planned economy (which was the source of the bureaucracy that made them an elite) and this is contradictory with the nobility/capitalistic elite ideology. The problem, in essence, is UC having two sources of richness and then it's necessary to simulate both behaviors and to not limit the definition to the "owners of the production means". However, your comments above plus some thinking of my own, make me see another way to solve the problem. All what you've said about UC and LC roles is great. I agree that my idea of LC having at least some control over production means will lead us to the introduction of the middle class. Since this would be a pain in the ass, I'm willing to accept "traditional" (nobility/capitalistic) UC having all production means not held by the State and LC providing labor only. In other words, you win: UC will be considered as the owners of the production means (aristocracy/nobility/capitalists).

2) Religious uprisings: There are three PAFs at the classes level: BAF, RRF, RF. There is a value for each one for the civ's RC which can lead to a religious uprising of the majorities, like what's presently happening in my country, where the church is organising street protests because they do not want religion ommited from id cards (high RC BAF, aided by high RC influence over the LC, caused street protests event!). If you want an uprising of minority religious groups, the ADF for the specific religious group does the work for you. Although I feel that it should be made possible that this triggers the formation of guerilla or revolutionary forces. (If I understand this correctly, the main difference between revolutionary and guerilla forces is that the first aim to take the capital while the last confine their action inside a certain province.)

3) BAF, RRF, RF: Everybody understands their difference, but I don't think it's clear why there has to be three of them, since they are expressions of discontent about the same situations and from the same agents. I quote myself again

quote: I must say that radical changes are somewhat different than all the others. In most times, all policies are reformist, which means that political agents will try to reform the current political system to suit their ideologies. But on special occasions, the policies of one or more agents will become subversive which means that these agents will consciously try to overthrow the government.

The difference between reformism and subversiveness translates into much more and severe events in the latter case, with the same discontent. So this explains the need for 3 different PAFs. Of course RRF and RF will be enhanced much more by characters and foreign influence and controlled less by propaganda and more by IIS, while the opposite will occur for BAF.

As for the per province computations, where would you place the revolutionary forces if such an uprising occured? Next to the capital or up in the mountains? In the industrial or in the rural areas? Consider some historical examples, i.e. the Spanish civil war, where the fascist insurgency begun from Maroc and then spread in these provinces of the mainland where conditions were ripe (the western part, neighbouring with Portugal and lacking industrial centers). This geographic division was essential for the military operations of the civil war.

4) Events missing: There can be more events than these. We had discussed that each class has it's own ways of acting. What I do not see here is economic actions, from the LC (strikes, general strike, undermining production, sabotaging the war effort, resistance to drafting), from the UC (taxes lockout, production lockout, refusal of cooperation from merchant agents) and from minorities (the same events as per the LC).

5) Characters: I should probably wait for the relevant thread, but anyway, here are some thoughts: It is understood that each political agent will have it's own character as leader (or at least it will be possible with the max characters setting). The character's traits will provide modifiers for the KL and Representation (if an ideologic faction) or for the de facto power (if a class) or for some cultural and demogaphic characteristics (if an ethnic or religious group) and also for the relevant PAFs (good leadership leads to more action). It will be this character that can be the target of foul play or the reciever of a bribe and his traits will determine the cost, the chances and the effects of such an action

Of course all these effects will be vastly larger if a Radical comes to play in any such position. And what's best, the effect of a radical will be post - mortem, they will persist after the character's death. Foul play against a radical is very risky: murder will probably backfire very badly (as in the extreme case of Rome, Hebrews and Jesus Christ) and a frame-up might do that too. An ideology ban (=radical's imprisonment) might lead to revolution (The Czar imprisoned Lenin, Lenin overthrew the Czar). A radical cannot be flushed down the toilet of history so easily...

------------------
"In a time of universal deceit, telling the truth is a revolutionary act."
George Orwell
[This message has been edited by axi (edited June 19, 2000).]

[roquijad]

LGJ:
First, a civ can have several GRW because a civ can have several ethnic groups.
It's true that no religious fervor exists in the riots model and in fact this was also suggested by Mark. My intention was to add that feature too, but I didn't because I didn't know exactly how to do it and the model had to be released this weekend, so it was just a matter of time to think about it. Religious fervor will exist in the next updated version.
----------
Where to put revolutionary forces: This is exactly why Lower Class' Revolutionary Feeling is handled at the province level. The event "Revolutionary forces formation" is a Lower Class' reaction. These forces will tend to appear (probabilities are involved, so nothing certain) in those provinces with higher LC' Revolutionary Feeling. LC' RF is computed at the civ level, but reinforced at the province by bad economic situation (which will be the only difference between provinces when computing the event probability)
----------
We can add more events. Let's go one by one carefully checking what variables and calculations are needed.
----------
Characters: I also think characters can add a LOT of nice effects if well interconnected with the govt and riots models. I admit I don't have a full knowledge about character's latest model. Can you e-mail it to me, Lord God Jinnai? After reading it we can start a thread on models interconnections.

Rodrigo

[axi]

***bump***

Just a reminder that there is OOA/OOD + development discussions to do here too...

You know Rodrigo, there are still some equations missing...

------------------
"In a time of universal deceit, telling the truth is a revolutionary act."
George Orwell

[roquijad]

I think Riots Model's variables will perfectly fit in F_Smith's coding structure. As for checking soon this model's behavior, we'll have to wait some more because the equations are not ready yet.

As one of the phrases in the Beast says, "good things happen to those who wait"

Anyway, thanks for reminding me that we need the equations here, Axi.

[F_Smith]

I have not looked this over, but it'll be next.

As soon as the cultures, ethnic groups, religions, tendency values, group attributes and governments are at a basic beta stage, this weekend, I'll jump on this next.

It'll fit in nicely as one of our first 'turn' logic attempts. If that goes smoothly, then we will build

Can you keep people happy over time, or will they riot on you? It'll be fun to test that out next week.

[roquijad]

RIOTS MODEL UPDATE

[the following is available in a MS-Word document]

The riots model needed its equations for PAF's (Pro-Action Feelings) and events probabilities. Here they are. I did my best to put things here in a way that could help implementing this in what we already have in the beast.

I understand things in the beast are organized like this:
A civ has provinces
A province has ethnic groups
Each ethnic group is divided in social classes

The riots model was slightly changed to fit this structure.
Also, PAF's and Events are now divided in Ethnic PAF's, Ethnic Events and Class PAF's, Class Events.


How to implement the Riots Model

Preliminary
1)The following attributes of ethnic groups are needed and are not yet implemented (that I know) in the beast:
Ethnically Discriminated (dummy variable: 0/1)
Religiously Discriminated (dummy variable: 0/1)
Enslaved (dummy variable:0/1)

Do not confuse these with govt policies Ethnic Discrimination, Religious Discrimination and Slavery. Of course they're related, but they're different. Govt variables define the level of discrimination, while ethnic variables define if the particular ethnic group suffers or not govt discrimination. To avoid confusion in what follows, ED, RD and Slavery will represent govt variables, while E_ED, E_RD and Enslaved will represent ethnic group variables.

The following ethnic variable is brand new:
Nationality Check (NCHK): Dummy variable (0/1). It is equal to 1 if the ethnic group nationality is the same the civ has and 0 otherwise.


2)The following variables are needed:
At the Province level:
Province Isolation: Ranging 0-100% (0-1). PI will be computed for each province using the distance (in km or miles) that separates it from the capital province, modified by the current tech level on communications and transportation. An equation for this is still TBD.
Centralization: 0-100% (0-1). How much of the local govt decisions in the province are controlled by the central govt. For now we'll let it entirely to ruler's control.
Administration Effectiveness Level (AEL): 0-100% (0-1). How effective the local govt is providing administration. That includes, FE, providing a justice system (police, courthouses, etc). No equation for it yet.
Province total population: Simply the sum of all populations for all ethnic groups in the province.
Province majorities population (PMP): Sum of all populations for ethnic groups for which SIGN(E_ED+E_RD)=0.

At the Civ level:
Empire's Stability (ES): 0-10 (real valued variable). ES is handled by the govt model.
The number of active revolutionary (military) units. Let's call this NRU
The number of active guerrilla (military) units. Let's call this NGU

Now we're ready to go.
The riots model should be implemented like the following (my best guess with my little knowledge of OO-coding):

Each available ideology needs:
A measure of the difference between it and the current govt profile. We'll call this variable IGD (Ideology-Govt Distance). Range for IGD:0-100
A measure of the difference between it and the ruler's govt profile. We'll call this IRD (Ideology-Ruler Distance). Range for IRD:0-100
The ideology total support (ITS: 0-100%). This is, what share of the total majorities population supports this ideology.


Each ethnic group needs 3 PAFs (that will be called "ethnic PAFs"):
NRF (Nationalistic Rebellion Feeling)
ADF (Anti-Discrimination Feeling)
IF (Independence Feeling)


Each social class needs 2+m PAFs (that will be called "class PAFs"):
RRF (Replace Ruler Feeling)
BAF (Better Administration Feeling)
RF_1 (Revolutionary Feeling for ideology 1)
RF_2 (Revolutionary Feeling for ideology 2)
...
RF_m (Revolutionary Feeling for ideology m)
and...
A variable measuring the difference between class preferences and ruler's preferences for Directly Negotiated Policies. We'll call this DNPDR (Directly Negotiated Policies Distance with Ruler). Range for DNPDR:0-100
A variable measuring the difference between class preferences and the govt profile for Directly Negotiated Policies. We'll call this DNPDG (Directly Negotiated Policies Distance with Govt). Range for DNPDG:0-100
N dummy constants to control events, where N is the total number of class events in the game. We'll call these CDEC (Class Dummies for Event Control). The role of these dummies is to define if an event can or not be triggered by this class. In this way events are ad hoc to classes and, FE, only the military class can make a military coup.

"m" above is the number of available ideologies.

Each ethnic PAF needs:
A value (PAF value). The procedure to create the PAF value is unique for the particular PAF.
A list of ethnic events. Events included depend on the PAF as shown below, but all PAFs include by default these 3 events: Verbally Exposing Discontent (VED), Streets Protests (SP), Riots (RIOTS).


Each class PAF needs:
A value (PAF value). The procedure to create the PAF value is unique for the particular PAF.
A list of ethnic events. Events included depend on the PAF as shown below, but all PAFs include by default these 3 events: Verbally Exposing Discontent (VED), Streets Protests (SP), Riots (RIOTS).


Each ethnic Event needs:
A probability built using the PAF value to which the event is related (Event probability). The procedure to compute event probabilities is generic to all ethnic events.
3 constants to be used when calculating the probability. Let's call them E1, E2 and E3.
Instructions describing what to do if the event takes place.


Each class Event needs:
A probability built using the PAF value to which the event is related (Event probability). The procedure to compute event probabilities is generic to all class events.
3 constants to be used when calculating the probability. Let's call them C1, C2 and C3.
Instructions describing what to do if the event takes place.


Ethnic Events included in ethnic PAF's:
NRF: The default ones and Guerrilla Forces Formation (GFF)
IF: The default ones and Declaring Independence (DI)
ADF: just the default ones.


Class Events included in class PAF's:
BAF: just the default ones.
RF_x: The default ones and Revolutionary Forces Formation (RFF), Attempt to Murder the Ruler (AMR), Army Betrayal (AB) and Military Coup (MC).
RRF: The default ones and Attempt to Murder the Ruler (AMR) and Military Coup (MC).


Computing PAF's:
All PAF's are computed like this: PAF=Base+100*(100-Base)*Mod/(1000-Base) and what changes from PAF to PAF is how "Base" and "Mod" are calculated.
All PAF's lie in the range 0-100.
All "Base" values lie in the range 0-100.
All "Mod" values lie in the range 0-1 (0-100%).

Ethnic PAF's:
ADF:
Base=AEL*(Enslaved*30+E_ED*ED*2+E_RD*RD*4+SIGN(E_E D+E_RD)*(10+SIGN(Slavery)*20))
Mod=0


NRF:
Base=(1-NCHK)*100*LandConnection*(1/(1+EXP(6-11.5*Nationalism)))*(1/(1+EXP(2-11*EGDS)))*(1-0.035*ES)*(1-0.1*AEL)
Mod=(1-NCHK)*(0.7*(ADF/100)+0.3/(1+EXP(6-1.5*NGU)))

where EGDS is the ethnic group demographic share in the province, that is, the ethnic population divided by total population in the province. ADF here should be replaced by the whole equation that allows computing ADF.


IF:
Base=NCHK*100*MAX(Centralization,1-Centralization)*ProvinceIsolation*LandConnection/(1+EXP(5-12*PMP))
Mod=NCHK*(1-Nationalism)*(1-ES/10)

Class' PAF's:
BAF:
Base=0.5*(Sum_over_i_of IGD_i*CSS_i)+0.5*DNPDG
Mod=0.6*(1-ES/10)+0.4*(1-AEL)

where IGD_i is the IGD value for ideology i and CSS_i is the Class Support Share for ideology i, that is, the fraction of people in the class who supports ideology i (the latter computed according to the govt model).


RRF:
Base=0.5*(Sum_over_i_of IRD_i*CSS_i)+0.5*DNPDR
Mod=0.5*(1-ES/10)

where IRD_i is the IRD value for ideology i and CSS_i is the Class Support Share for ideology i.


RF_i:
Base=IGD_i/(1+EXP(8-19*ITS_i))
Mod=1/(1+EXP(6-1.5*NRU))

where IGD_i and ITS_i refer to IGD and ITS values for ideology i.

Computing probabilities for Events:

Ethnic Events:
All event probabilities are computed as
EXP((E1-0.01*Aggressiveness)*(PAF+E2+E3*CivilRights))

where PAF is the corresponding PAF value to which the event is related.

Class' Events:
All event probabilities are computed as
CDEC*EXP((C1-0.01*Aggressiveness)*(PAF+C2+C3*CivilRights))

where CDEC is the dummy associated with the class to which the event is related.


Computing Ideological Distances (IGD and IRD)
Both variables measure the difference between two govt forms. IGD between a given ideology and the current govt profile and IRD between a given ideology and the ruler's govt profile. Both are computed the same way, calculating differences in values for every present variable and then "summing" all of them to have a single outcome. The calculation is:

IGD (or IRD) = 100/(1+EXP(7-14*D/Q))

with D being equal to this sum:
A1*(1+10*ABS(X1-Y1))^2
+A2*(1+10*ABS(X2-Y2))^2
...
+A8*(1+10*ABS(X8-Y8))^2

where
Q is a constant equal to 100.
X1...X8 are the values for one of the two govt types being compared corresponding to the 8 variables present (ruler_polpower, people_polpower, capitalist_polpower, religious_polpower, warrior_polpower, privateproperty, socialpolicies and economicplanning).
Y1...Y8 the same, but for the other govt form being compared.
A1...A8 are constants associated to each variable. We'll make them (for now) all equal to 1 except the one corresponding to SocialPolicies that will be equal to 0.7.


Computing Distances for Directly Negotiated Policies (DNPDG, DNPDR):
It's the same system used for IGD and IRD, but for policies civilrights, foreignaffairs, slavery, ethnicdiscrimination and religousdiscrimination. In this case it is compared the class preferred values with the ones in the govt profile (to compute DNPDG) or with the ones in the ruler's govt profile (DNPDR). Q in this case is equal to 190 and constants A1...A5 are:
A associated with civilrights: 3
A associated with slavery:0.03
A associated with ED: 0.004
A associated with RD: 0.02
A associated with foreignaffairs: 1.5

Final Notes
Note1: The effects of economic unhappiness are not included. They'll be added in a later time.

Note2: In all these equations it is assumed ethnic attributes such as nationalism are all in the 0-1 range.

Note3: The event "Legally Replacing the Ruler" initially considered in the model is not implemented for now.

Note4: About the "Instructions describing what to do if the event takes place": I don't know what's the correct way to implement them, but for a "Riots" event, FE, instructions could be something like "display message 'riots in province X!!' and then destroy a random amount of infrastructure in the province". This is a simple event. Others have wider repercussions and the "instructions" can be more sophisticated. Even though I have in mind what should happen for each event, I'll leave that for a future update.

Note5: The original model didn't included the effect of cultural differences in aggressiveness for computing event probabilities. Now it does.

Note6: The effect of Political Representation (representation values) is not considered in this update. It will be added in a later time.

Values for Constants E1...E3 (ethnic events)
Event E1 E2 E3
VED 0.02 -250 150
SP 0.02 -300 150
RIOTS 0.05 -110 0
GFF 0.05 -103 0
DI 0.09 -105 0


Values for Constants C1...C3 (Class events)
Event C1 C2 C3
VED 0.02 -250 150
SP 0.02 -300 150
RIOTS 0.05 -110 0
RFF 0.07 -115 0
AMR 0.08 -117 0
AB 0.07 -118 0
MC 0.07 -110 0


Values for CDEC (Class' Dummies for Event Control):

Event/Class Upper Middle Lower Religious Warrior BE
VED 1 1 1 1 1 1
SP 0 1 1 0 0 0
RIOTS 0 1 1 0 0 0
RFF 0 1 1 0 0 0
AB 0 1 1 0 0 0
AMR 1 0 0 1 1 1
MC 0 0 0 0 1 0


Do all this look coherent for coding it, F_Smith?
[This message has been edited by roquijad (edited September 09, 2000).]

[F_Smith]

Rodrigo:

Yeah, that looks great.

I can start merging this in tonight. I'll upload the newest stuff tomorrow. It's running pretty well now.

[axi]

WOW! I'm impressed! And think that here there are not all the riots model features! I have many (naive) questions to ask:

1) The modifiers that you are providing with your formulae Rodrigo are too specific to be arbitrary. This means that you have tested the formulae somehow and you have ended up with these numbers as a preliminary attempt of balancing the game? Is any of this stuff by any chance implemented, in spreadsheet f.e? If yes, I would be very glad to see it, along with the word document of the update. Also, as you will see below, I am eager to discuss the formulae and their behavior, if you have time (we could discuss privately if you don't want to complicate things too much in here)

2) A question on object structure:

quote: A civ has provinces A province has ethnic groups Each ethnic group is divided in social classes The riots model was slightly changed to fit this structure. Also, PAF's and Events are now divided in Ethnic PAF's, Ethnic Events and Class PAF's, Class Events.

This means that, since classes are under EGs, there won't be just 6 class PAFs for one civ, but 6*#ofEGs PAFs? Or, furthermore, since EGs are under provinces, there will be 6*#ofEGs class PAFs per province, thus 6*#ofEGs*#ofprovinces class PAFs per civ? If this is so, doesn't that mean too much redundant information (not necessarily for the game itself, but definately from the player's point of view), which will mean that we will eventually need to do some aggregation between PAFs, chiefly for interface purposes?

3) Another question on object structure: How about the ideologies? Will an have a PAF (RF or RRF) that is the aggregate of all class PAFs for the certain ideology? And what will trigger the related events; this aggregate PAF or one of the class PAFs? How will groups from different classes, who bear the same ideology cooperate to start a revolution or replace the ruler? This is important because it will judge the outcome quantitatively and qualitatively.

4) I think that for now, it is a wise decision not to add interaction with economic data into the riots model. The beast is for the time being exclusively political; the govtecon model should be included first and then we can expand the riots model upon economy, using it's preexistant structure, as it is laid out here. Nevertheless, it is not bad to plan ahead.

5) What about the Exploited Class? I am not refering to minorities, but to the part of the minorities that are economically exploited (whatever value there is for slavery). Of course, at the govt model they are disregarded as a class, but at the riots model, they are entitled to some class-related PAFs and events. They must have a special, economic-type, ADF and an RF and they should be allowed to cause RIOTS and RFF, plus of course other, economic events. On the XC, Rodrigo, Mark and I had reached an agreement in the govtecon thread.

6) Could you enlighten us with some explanations about the calculations of PAFs? What is the difference between "Base" and "Mod"? How did you decide what variables go into each? Finally, why are the formulae so complicated? What is the special effect of using exponentials in the denominator?

7) On event probabilities: Why is aggresiveness subtracted in all formulae? Shouldn't probability rise with aggressiveness? More aggressive people mean more probable events, don't they?

8) On ideological distances: I can't understand the purpose of and the complexity of the formula:

quote: IGD (or IRD) = 100/(1+EXP(7-14*D/Q)) with D being equal to this sum: A1*(1+10*ABS(X1-Y1))^2 +A2*(1+10*ABS(X2-Y2))^2 ... +A8*(1+10*ABS(X8-Y8))^2

Again I don't get the meaning of the exponential. What naturally comes to mind is to handle ideologies as vectors, as I have done in my proposal of the election-type system:

quote: D(Ia, Ib)=sqrt(2)*sqrt[(Ia1-Ib1)^2+(Ia2-Ib2)^2+…+(Ia8-Ib8)^2], so that 0 < D < 4.

Maybe, if F_Smith codes up my election system for ideologies as an alternative to the default, we should use your formula for the distance?

9) We should bear in mind the need for flexibility that made us change so many things in the govt model. First, we must provide for the N Middle Classes that you included. Then if mixed and complex classes, along the lines of contributions are implemented, as I have proposed, we must provide for them too. Luckily, the riots model is so structured, that it is mostly independent of the class structure. I think that only the CDEC matrix needs to be rendered flexible, to allow the non-standard classes access to the proper events.

10) My previous comment, on how dependent the riots model is of the class structure made me realise a possible drawback in your system: Assuming all other conditions are equal, should two classes have always the same probability of causing a certain event, triggered by a certain PAF? F.e., If the UC and the RC have the same RRF, would they have the same probability to AMR? I believe that the UC would be a more probable murderer. Or if the MC and the LC have the same BAF, would it be equally probable for them to make SP? I feel that the CDEC should consist of more than just 0's and 1's: it should be a generic probability per class matrix rather than a yes/no matrix.

11) Another important feature, that was discussed long ago between us, but which I don't see implemented now, is one classe's PAFs or Events influencing another classe's PAFs. What if the RC is so pissed off against you that it decides to excommunicate you. Will the other classes attitude remain unchanged? What if the UC dislikes you; won't it try propaganda, to raise the discontent of other classes too? From my comments #10 and #11, it seems that, while trying to create failsafe and waterproof equations, you have ditched some important interconnections between the various PAFs and Events.

12) This model controls the procedure that creates events that correspond to the ruler's policies. This procedure has certain steps:
a. How much people want to do things against you. (PAFs)
b. How possible it is that they will do them. (Events probabilities)
c. How serious these hazards will be for you.

This last step is still missing and we should start considering about it. F.e, if there is a revolution by ideology X in province Y, how strong will be the revolutionary forces? If there is an AMR from class A, how possible is it's success? (and also, if this is a Military Coup, what is the possibility of the ruler getting killed in the process?). How much will a riot cost in terms of infrastructure? How much will street protests drop your international image?

Please excuse me for the flaming. My criticism is not aimed at you personally, but at your model, so that we can make it better.

------------------
"In a time of universal deceit, telling the truth is a revolutionary act."
George Orwell

[roquijad]

F_Smith:
I'm glad it sounds good at least on overall.

Hi Axi,
Thanx for the detailed analysis. Here are some answers:

1) Yes, I made experiments with a spreadsheet. I can send it to you if you want, but I have to warn you it isn't as "friendly" as the one for the govt model. All numbers (and equations) are subject to change if they don't have the desired outcomes, but I believe in most cases it won't be more than changing the constants in them. Current parameters guarantee at least an overall "correct" result. Note you yourself can implement as few as only one PAF in a spreadsheet and play around with it. Contrary to the govt model, here equations are not all related to each other, so analysis of equations one by one is possible and I suggest you to do it for those you feel aren't ok.

2) Absolutly right. When I said "the model was slightly changed to fit [F_Smith's implementation of the govt model]" I was refering to the absence of a "majorities" object. Therefore, PAFs now appear as many more than I initially planned. Currently we have a code that provides a lot more social units than planned, so PAFs have grown accordingly. I have mentioned before that this big number of social units could be too complicated for the player, but I preferred to stay with F_Smith's implementation and "just see later".

3) I don't understand much of your wording here. The road I took is to measure the difference between an ideology and the govt profile as the RF. Then, for each class, the share of people supporting it takes that value as the PAF. In other words, the PAF value for RF is the same regardless of class, which means all people who support a given ideology is equally pissed off about it not being implemented. The same happens with RRF. There's no "cooperation" for action, although it may seem like that for a player, because different classes can at the same time be triggering events with the same goal. "Cooperation" sounds like something attractive to add, but I don't know exactly how it should be implemented.

4) Agreed.

5) I don't like minorities having political behaviors as majorities have, but in this implementation they'll act exactly as you want. That happens because even if the ethnic group is a minority, in F_Smith's coding the group is still divided in classes, and therefore all class PAFs exist for them.

6) Sure!
"Base" is the essence of the unhappiness. Base *is* the PAF. "Mod" is the things that can reinforce the Base. Mod works like a multiplier, so in its "purest" sense, the equation for PAF is something like Base*(1+Mod) and FE it could be 30*(1+50%)=45. The equation is more complicated only to ensure the final number not being greater than 100, which is the highest value a PAF could take and to avoid Mod transforming Base too much.
As a corollary of the above, it's the Base value that interest us the most. No matter how high Mod is, if Base is null, PAF is null. That's why I call Mod a "reinforcer". This tells you what variables you would put in Base and in Mod. Take NRF for instance. Nationalism must be in Base, because the PAF *is* a nationalism matter. But the number of guerrilla units currently active in the civ must be in Mod, because it isn't true a priori that "the more guerrilla units, the more rebellious you get". NGU has an effect only if you already have some thoughts about starting a nationalistic rebellion. You then would say "Look!, nationalistic uprisings are happening, therefore, we, who want to fight for our nationality, should take advantage of this and join the struggle". But if Base is zero (nationalism is not a matter to you), no matter how many guerrilla units you see, you won't start up a fight.

Is that clearer?

About the exponential, in general in all three models I use in many cases 3 equations with an exponential form because they have shapes that fit many real life behaviors. They're:

EXP(x) in the range (-inf,0]

1/(1+EXP(-X)) and 1/(1+EXP(x))

The last two are interesting in the range [-6,+6] approx. where they present "saturation" effects (asintotic behaviors). I suggest you to plot them in Excel to see what's the effect I'm putting.

Just as an example, in "Mod" for RF, expression 1/(1+EXP(-x)) is used for variable NRU. "x" here is equal to variable NRU but with a linear change of scale: x=-6+1.5*NRU. When NRU is zero, x=-6 and the expression takes a value near zero too (which is correct). As NRU increases, Mod increases too, slowly at the beggining and then quicker, but when NRU is around 8, Mod tends to find a stable value in 100%. Here it is:
NRU Mod
0 0%
1 1%
2 5%
3 18%
4 50%
5 82%
6 95%
7 99%
8 100%
9 100%
10 100%

So, combining this effect with the interpretation of Mod, what I'm saying is:
a) When NRU is null, PAF is only its "essence" value.
b) When NRU is small (<=3), the revolutionary feeling gains a small momentum because of the existence of already active revolutionary units.
c) The momentum is much greater when 4<=NRU<=7
d) When NRU is >=8 additional revolutionary units have no effect. The momentum has full strength already.

This is good because even though you expect the formation of military units can give momentum to the revolution, you don't expect this element alone to continously increasing the feeling for revolution endlessly. At some point there's a sufficently large number of units that people don't get any more excited if a new one is added to the revolution.

Of course constants "6" and "-1.5" define how rapid this saturation of the effect is achieved and for what NRU level saturation is got. We may decide that a "sufficiently large" number is not 8, but 15, FE, and in that case we'll modifiy constants. But that's for testing. For now we have at least the main effect of the variable.


7) It is working as you say. More aggressive, higher probability. The substraction is confusing you. Plot it and you'll see it.

8) I tried to use it like we agreed, as an euclidian distance computation, but I couldn't get the effects I wanted. Playing with numbers, I found ideologies one could consider "different enough" wouldn't give me a high D value. So I used this other form. If you wanna insist in using the euclidian distance, I'm all for it. Just do some testing in a spreadsheet and give me a precise way to do it because I couldn't.

9) This model is flexible only to what regards the number of classes and ethnic groups. You can have as many as you want of those and just specify here the matrix for CDEC. I'm affraid the number of PAFs and Events is not flexibile. A scenario designer won't be able to create more PAFs , because, as seen, each PAF is computed differently, so another PAF means another equation. And for events, it's the "instructions concerning what to do if the event takes place" that kills flexibility. What happens in an event can be as unique as the vent itself, so there's no way to generalize instructions to have any possible event designers could imagine.

10) You're right. That's better. I'll generalize it.

11) That was a fine feature I totally forgot about when I took away all unrest activities from the govt model to create the riots model. We should consider how to implement it in this structure.

12) It's true. In the "events" section in the riots model (part IV.2) I describe what should happen in a general sense for each event. According to that, we must start to describe precisely the "instructions regarding what to do if the event takes place". Having that, we can start to think what's the best way to present the info to the player. That's the next step in this model.


Thanks a lot for your comments, Axi!

[roquijad]

Riots Model Glossary, Brief Summary and Current Status (September 24th, 2000)

BRIEF SUMMARY
The two types of social units in the game, Ethnic Groups (social model) and Social Classes (govt model) have a list of Pro-Action Feelings (PAF's). A PAF is unhappiness or discontent that becomes a clear and specific intention of doing something. It's the goal you want to achieve that differentiates PAF's.

List of Ethnic Groups PAF's
NRF (Nationalistic Rebellion Feeling). The intention of a controlled nationality to form its own civ.
ADF (Anti-Discrimination Feeling). The intention of a minority to stop being so.
IF (Independence Feeling). The intention of a province to become independent and form a new civ.

List of Social Classes PAF's
RRF (Replace Ruler Feeling). The intention to replace the ruler.
BAF (Better Administration Feeling). The intention to get the govt doing better.
RF (Revolutionary Feeling). The intention to overthrow the govt to replace it with a very different style.


PAF's can take different actions. With the same goal, you can do several different things to achieve it. This is made with events. The higher a PAF is, the greater probabilities for different events ad hoc to the particular PAF.

Some of the events are Street Protests, Riots, Formation of Revolutionary Forces, Attempts to Murder the Ruler, etc.

The Riots Model is fed by the social, govt and economic models.


GLOSSARY (MOST IMPORTANT TERMS)
Pro-Action Feeling (PAF): Unhappiness taking the form of a specific goal.

Event: How a PAF becomes a precise action in the game with implications.


CURRENT STATUS
Equations for PAF's and Event Probabilities recently released.

Equations still need to be refined. In particular, no info is yet used from the economic model. That's the main line of work currently.

Specification of what exactly should happen when each of available events occur, is missing.

Almost no debate has been around this model.