Government Model v.2

[Mark_Everson]

Hi everybody:

Man, there has been a lot of action since I was last here... I'm going to respond to the coding debate in the other thread.

Unless I'm missing something, the only other thing that I really need to respond to is Rodrigo's question about the problems in the Econ model when there is only a lower and upperclass. (And also taking as the definition that the lower class cannot control any capital or land.) I have already described the problem in my post of July 5. I will elaborate on it a little bit here. The main problem comes near subsistence. If we assume that the UC controls all capital and land, it is natural to assume that they get a part of the profits proportional to the capital and land value in the production process. So I have assumed in trying to match the economic and government models that the class that controls a particular input to the production process gets a proportional amount of the production to call theirs. To figure out the value of each input to the production process (labor, capital, and land in the farm sector for example) I assume that the marginal productivity due to that resource indicates its value in the production process.

To make this concrete, the values that are at the top of the Econ spreadsheet for the very poor low-tech province show the extra amount produced if adding a single unit of each of labor, land, and capital in the farming sector. These amounts are 3.68 for labor, 0.93 from land, and 0.81 from capital. So if we assume that the UC gets all the food produced due to its contributions (both land and capital) in the current model it will get (0.93+ 0.81)/(3.68+ 0.93+ 0.81) = 32% of the food credited to it. Only the remaining 68% would be what the LC could call its own. So even Before Any Taxation the food left to the LC would be 68% x 5.27 = 3.6 units of food. This is on the brink of catastrophic starvation! (4 units of food per head are required for minimal nutrition and no net population growth) At this level of food the population will be declining approximately 1% per year! And this is a province with pretty good agricultural land, it would be much worse in a less well-endowed province.

So that is why I think we need one of two things, either an LC that allows for some control of capital, or a middle class. I think Axi's suggestion about calling free subsistence farmers middle class is basically a good one... We could also mess around with the way the Econ model works to avoid this problem, but the Econ model also has a lot of work in it, and is somewhat difficult to balance. I have already cited numerous other arguments for why a middle class is, I think, a good idea, and I won't repeat them here.

[axi]

Mark: Thanx for sending me the econ spreadsheet. There are some big issues about it's compatibility with the govt model. I will discuss them presently. Hang on for a couple of hours...

Ok, I sent my work to Mark and Rodrigo and I posted my comments and explanations in a new thread. I am going to be away for the weekend. I hope that you can cope without me for a while.
[This message has been edited by axi (edited July 14, 2000).]

[roquijad]

I don't see a real need for the Middle Class, but I guess we can just go for it so everybody is happy. I find more interesting Mark's proposal to find a way to build class' behavior in a more general way to increase flexibility. I think it can be done and I'll see how to do it.

As for the discussion about ideologies, I believe the concept is right, so we need ideologies in the model somehow. Ideologies fix the problem of classes caring only about themselves, so you can find a happy WC in a regime with low WC pol.power if they actually believe that's good. Ideologies are IMO the most important ingredient we added compared with the old model and they provide several other good features such as a true-to-life modeling of revolutions. Maybe actual implementation can be changed, preserving the concept, so more classes can be added without destroying the core of what they're giving us now. I'll think about other ways to implement them. If you can give it a thought too, Axi, it'll be appreciated.

[roquijad]

Hi F_Smith,

I tried to play with the beast last night and windows crashed while doing it. I'll try again tonight to give you more feedback.

As for the questions you rise now,

1)It's perfect how you have it now. We haven't invented a way (in terms of equations) to make a class choose a desired tax rate. Probably it won't be ness. either. Keept it how you have it.

2)There're some good reasons for the ranges I chose. It has to do with the outcomes needed. Preserve ranges as they're.

3)Job? I don't know what you mean, but if it's something you'll keep hidden and I can still be thinking of our regular social classes, I don't mind.

4)Yes, they have to sum 100%. The numbers come from a math process considering cultural values, ruler's preferences and ideologies values. If for the moment you prefer to keep it editable so we can put whatever numbers we want, that's alright. Implementation of the math process can come later.

[F_Smith]

Hi, Rodrigo:

Oh, god, I'm sorry it crashed Windoze. That's not suppose to happen with Java, altho using IE as the platform can create problems. If it happens again, please let me know. Do be warned, the GUI stuff is *very* slow to load . . . my webspace is the freebie space that comes with my cable modem, and it's poor. Altho on my machine here at home it loads in about 15-20 seconds (at work it takes about 30-40 seconds).

The downloadable zip (which doesn't have the newest stuff yet, maybe I'll put that together tommorrow) has an application that alleviatest that problem, if you don't have a cablemodem yourself.

I am dying to get your feedback. I'm sure I've missed the boat on a few things, and I'm anxious to finish this baby and get on to other models.

Just f.y.i. -- for 'filing' purposes I've seperated the 'economic classes' (uc/mc/lc/slaves) from the 'job/social classes' (religious, military, govt, etc). They'll act the same, from ya'lls standpoint.

[F_Smith]

Okay:

The beast is about ready to be put to work. After some cosmetic cleanup, I'll be spending time on the turnhandler code tonight. Which leaves me with a few questions:

[*]Tax Rate -- Right now, the govt's tax rate policy automatically becomes the ruler's tax rate pref after one turn. I was wondering if the majority EthnicGroup (in the proportions of the govt's power structure) should have a say in taxation.[*]Numbers -- As I look at and play with this, it is slightly confusing for slavery and a few others to not use the 0%-100% range. Might just be me, but should we standardize the numbers?[*]Economic Class v. 'Job' -- it made most sense, from a data storage viewpoint, to seperate these two. But this will be seamless and invisible to the player. The values will cumulate and look exactly as ya'll have laid out.[*]Political Power Structure -- right now, the player can set the %'s. I assume that they have to add up to 100% total (might have been in the text and I missed it)? I think perhaps I'm suppose to have a selection of govt types (the ones in the 'ideologies' section of the govt post) for the player to chose from? If so, anyone want to come up with values there?[/list=a]

[roquijad]

GOVT MODEL UPDATE


For the last weeks I've been considering how to increase flexibility and scalability concerning social classes. This was triggered by F_Smith and others who wanted to see more classes than the ones I and Axi considered originally for the model. Unfortunately, I don't bring good news. I decided to change a couple of things to allow some more flexibility, but it's far from being what some of you wanted.

In order to let you all know why adding more classes is no easy task, I'm gonna enter in some details. Hopefuly this will help understanding the problem.

From the computational point of view, it's quite easy to add any number of social classes. You just have to add more objects and that's it. The problem comes in modeling classes' behavior. Objects hold two things. Data and procedures for the data. Procedures take data and process it with some equations to produce something we call "class' behavior", which is nothing but another set of numbers we interpret as classes' decisions. In order to have an arbitrary number of classes AND their behaviors, we need equations (procedures) that are common to all of them. Having common equations for all social classes is what makes possible scalability, because if you need to code new procedures for new behaviors each time you add a social class, then you have to admit there's no real scalability.

Having a common set of equations for all social classes assumes their behavior is similar. In that case it's the data (class' attributes) that produces the small differences. This is FE what we see in the military area. Each time we want to add a new unit, we just specify its attributes (attack strength, speed, etc) and the game engine uses the same procedures with these new numbers to generate the unit's behavior in the battle field. And this is good because military units are indeed very similar. Having a common set of equations for all units (of most of them) makes sense and therefore you have scalability.

Unfortunately this is not the case with social classes. If one starts to imagine classes like "farmers", "scientists", "religious warriors", "land-based aristocracy", "trade-based aristocracy", "freed slaves", "greenpeace fighters", "jedi knights", "elves", "robots", and a very long "etc", then it's pretty obvious that behaviors can get to be very different. If you try to create a set of equations valid for all of them it becomes a monumental task. Having a set of equations common to all classes was very problematic when I was developing the model and I only had 5 classes! I actually didn't achieve it. I found that to have a reasonable and believable class behavior for each of them I needed class-specific procedures in many parts of the game. Generic (or parametric)equations are hard to create, specially if you want social classes to have a sophisticated behavior like the one we're trying to implement.

Equation sophistication is just a part of the problem. Data input you need for behavioral equations will not be entirely held in classes' attributes. Scientists will be interested in tech variables, aristocrats in economic data, greepeace in ecological data, etc. A generic class behavior equation would actually have to access all available info in the game!

It's the nature of human behavior that complicates things. It can be so complex and different from group to group that it's ambicious to resume it all in a few equations, even if you are focusing only in some aspects of behavior.

Coding will allow us to create any number of social classes we want, but we have to realize this coding flexibility doesn't mean game flexibility. When a new social class is added, it will use the same procedures older ones used and so, class' behavior will be almost the same. The new social class is useless and senseless. It will behave as the others do, which means the game hasn't really changed. You can add 20 more classes and the same happens. We'd be fooling ourselves if we trust that adding more social_class objects will bring us new and different scenarios.

So, I'm sorry to say it, but the model doesn't have the capacity to scale up. One solution is to change the model and create a new one with a much simpler behavior for social classes. Simple enough so we can have generic procedures for all classes. But I wouldn't take that road. On one side the model will lose flavor and in the other it implies a more homogeneous behavior for all classes (as it happens with military units), and homogenity is the greatest enemy to social classes, because the fun in having more classes is having new behaviors.

The other solution is to allow scenario designers to code new behaviors. Add a class and code whatever it feels good for the class behavior. But, well, if players can get into coding, the word "game flexibility" takes another meaning...

This ends my explanation. Now I present to you the update to the govt model. It will allow having an arbitrary number of the so-called "economic classes". The model originally included the Upper Class and the Lower Class. With this update we can (if we want) include the Middle Class and as many other economic classes we want without reducing behavior flavor we have till now. The model, then, will have N economic classes, plus the Warriors Class (formerly known as Military Class), the Religious Class and the Bureaucratic Elite.

The update replaces UC and LC behavior equations with a generic set of behavior equations for "Socio-economic classes" (I hope the word "socio-economic" exists in english). I prefer the term "socio-economic" to simply "economic" because class behavior will be given by economic role AND cultural info. The scenario designer and we (when creating the "by default" game setting) have to choose a number of socio-economic classes and for each of them need only to specify 1)the degree in which the socio-economic class controls kapital and 2)the degree in which it provides labor. The exact measure in which both variables will be handled is still TBD depending on how they best serve the econ model. Political and economic behavior for socio-economic classes will be given using these two variables and cultural info from the social model. Equations for socio-economic classes' behavior will wait until we decide how variables for provision of kapital and labor will be measured. Equations shouldn't be a problem once that is solved.

One of the problems of adding more classes was ideologies. Ideologies were meant to have a list of all social classes and their respective political powers. Adding more social classes would make the list longer, which isn't a problem, but would complicate defining how any govt type should look like regarding pol.power distribution and would also encourage the introduction of an important amount of extra ideologies. To solve this, instead of a list of all classes, ideologies will posses this info:

Ruler's pol.power
People's pol.power
Capitalists pol.power
Religious Class pol.power
Warrior Class pol.power

(if you don't know why the Bureaucratic Elite doesn't appear here, refer to the model document)
This means Upper Class pol.power and Lower Class pol.power were replaced by People's pol.power and Capitalists pol.power. People's pol.power is pol.power granted to the masses in terms "one man, one vote". This means a social class has access to this pol.power proportionally to its demographic share. The RC and WC are very small because they're respectively the religious leaders and the high officers of the army. Then, demographically they're assumed to have a share of 0%. Then, only socio-economic classes have relevant demographic sizes and therefore only them have access to People's pol.power. Demographic shares for each socio-economic class are given by a TBD procedure currently in discussion in the govt-econ thread.

Capitalists pol.power is the pol.power granted to socio-economic classes that provide kapital. These classes have access to this pol.power proportionally to the amount of kapital controlled.

Having ideologies defined like this implies that socio-economic classes have a total pol.power computed as the sum of the People's pol.power granted in terms of demography (which may be interpreted as "votes") and the Capitalists pol.power granted because of its control over kapital (which may be interpreted as aristocracy privieleges).

Just as an example, the ideology for a democracy would look like this:
Ruler's pol.power___________20%
People's pol.power__________80%
Capitalists pol.power_______0%
Religious Class pol.power___0%
Warrior Class pol.power_____0%

Even though Capitalists pol.power is null, socio-economic classes providing kapital have pol.power anyway through People's pol.power, but only proportionally to their demographic share. In this way, you can see why Capitalists pol.power can be considered a priveledge. It gives more power than the demographic share.

Finally, this change in how ideologies are defined implies a change in how Warrior Class' mentality is computed. It was formerly computed as partly influenced by the Upper Class and partly by the Lower Class. It will be now computed using People's pol.power, Capitalists pol.power, demographic shares and kapital shares in a way I won't detail now.

One advantage of the new way to specify ideologies is that computing demographic shares can be (if desired) totally independent from ideologies themselves (that is, independent from the economic variables ideologies specify). In the old system, it was very "advisable" to compute demographic shares using ideologies info to achieve consistency.

It should be said that in order to make this update useful, the govt-econ interactions must be able to manipulate an arbitrary number of socio-economic classes. Since that thread is still under development, I guess we're still on time to get that done the right way.

Comments?

[Lord God Jinnai]

I think there is a need for the scientific class. Either that or drop the religion class because you are unfairly treating those 2 by giving one the ability to have a ruling class and the other not. This is very important as we move into the future because there are many cases where scientists who would have much political clout in a system could increase spending for new technologies and that. Not doing this you are saying its not possible for the scientific elite to have power like their usual rivals in religions. This i believe will be very unbalancing as there are things only scientific elite that have control of part of a society can produce or direct the civs progression into the future.

Bottom line: Either drop the religion class (which i doubt you will) or add a scientific one. This is more than me just asking for more flavor, but actual gaming mechanics of how the model will effect stuff.

[F_Smith]

rodrigo:

I like it. I like it a lot.

I have to disagree, what you've built scales nicely and would allow a large number of different types of 'social classes' to be defined that acted very differently. Varying the values of the ideology produces a very different social class behavior.

I like 'ruler', 'people', 'capital' as an object design for political power. Then including any collection of 'social classes' defined for that game is fine.

We can define yours as the 'basic' social classes, for now -- in fact, I'll do that soon, in the beast.

And other 'social classes' with their own ideologies can be defined. Adding a 'Scientific community' social class can be a matter of defining it's ideology.

Or am I wrong?

[Mark_Everson]

Rodrigo:

Obviously you thought about this a lot. Great solution!

I don't want to muddy the issue any, but I had one vague thought. For arbitrary class behavior of the 'scientist' type we might be able to use the tags that the tech system already has. If mostly using the tags some simple system that fit with your existing equations could be designed, then all you would have to do is interface with the 'hard science' tag (don't think there is one...). We are trying to keep the list of tags down to 10-20 or so, but still that would give potentially 10-20 additional classes that could be added based on tags. FE farmers... The Elves are clearly on their own

I don't mean to push, this is more in the line of brainstorming

[Richard Bruns]

Mark: How would you use the tech tags to determine class behavior? They would be a good guide for determining what the class would invest in, but I cannot imagine how they would determine all of the behaviors of a group of people. They were never designed to do that.

[roquijad]

All:
It seems my proposal for having an arbitrary number of socio-economic classes is finding support. That's fine. But I was also seeking your agreement about the imposibility of having an arbitrary number of social classes (elves, etc). Did I convince you?

About the scientific class: If we accept the number of classes is a key decision, then we can start discussing about it. If, as a team, we feel a sci class is needed, then I have no problem generating equations for its behavior. Just please let's focus first in the arbitrary vs fixed number of social classes. IMO this is a more important subject and relevant for coding matters.

F_Smith:
Based on your comments above and some you gave in the object builder thread, I believe you have a little confusion with ideologies. I'll give a brief explanation below about them. If this confusion of your was only in my imagination, just ignore me.

Class behavior and ideologies:
Ideologies are a description of how the govt should work, regarding 1)who has the political power, and 2)what economic regime to implement. Variables describing an ideology are the same for all of them. Only numbers change from one to another. FE, a capitalist democracy ideology and a communist dictatorship ideology:

Variable__________capit.democr______comm.dict.
Ruler's pol.power__________20%_________70%
Capitalists pol.power______0%__________0%
Religious Class pol.power__0%__________0%
Warriors Class pol.power___0%__________30%
People's pol.power_________80%_________0%
Private Property___________90%_________0%
Social Policies____________15%_________100%
Economic Planning__________X%__________Y%

(I didn't put values for EP because we're not sure anymore how to handle this variable and it's likely to disappear from the game)

Just like the two examples above, we'll have around 20 ideologies in the game covering all the govt types we've seen in history. The values for each ideology are fixed during gameplay. Ideologies will be discovered as techs, just how in civ2 you discover "monarchy", "communism", etc.

Each social class has a behavior (or mentality). A social class has NOT an ideology. A social class' mentality covers several things, like its preference about slavery (not covered in ideologies as can be seen). As a part of social class' behavior, class members choose the ideology of its preference (among those discovered). FE, Lower Class members in western countries would be inclined to choose democracy as their preferred ideology. The model actually allows classes to be divided. Some class members may choose democracy while others choose communism. FE, the following is possible:

Class______communism_____democracy____monarchy
WarriorClass___5%_____________85%________10%
LowerClass___26%____________71%________3%

which means, FE, that 26% of the people within the lower class supports communism.

Class preferences for available ideologies vary depending on things like culture. Because of this, as culture changes, preferences do too. People can support monarchy in ancient times, but dislike it in modern ages.

The above shows that behavior/mentality is not equal to ideology. That's why a phrase like "Adding a 'Scientific community' social class can be a matter of defining it's ideology." is not correct. The sci class has a behavior which may include things like a preference for scientific investements. The sci class can choose, as only a part of its wider behavior, an ideology. This choice represents the sci class preference for govt type. In fact, several classes can choose the same ideology and still have different behaviors in other aspects.

Class behavior/mentality and the concept of ideology are different things.

[Mark_Everson]

Rodrigo:

I was not suggesting the scientists class as something that should be in the basic game. Rather it was just an example of something a scenario designer might like to do. I did not mean to throw off the discussion of the basic system you outlined, which I think is really good, with my comments about flexibility. My idea was simply one that to model a "landed aristocracy" class that one could basically use something that extends the behavior of the UC. So our landed aristocracy class would simply be a class that behaves like the UC in most ways, except that some fraction of its happiness with the government would involve increases in economic behavior related to "farming" tags in the technology model.

I agree with you completely that the classes that you outlined should be all there is for the basic game. I just thought I would get my suggestion in about the flexibility part well you were still thinking about it. That way if the suggestion worked, we wouldn't have to backtrack yet again.

[Richard Bruns]

Elves would be a seperate race, not a different social class. The population model is designed to create and model nonhuman races. The Elven race would probably contain the same social classes the Human race would.

I think I understand the difficulties involved with creating an arbitrary number of social classes. In addition to interacting with things like the economic model, the classes interact with each other. Thus, the complexity of the system increases exponentially as more classes are added.

However, I think we should include the functionality anyway. The ability to design custom social classes is a powerful design tool. Perhaps the designer of a modern scenario would want to take out the Military Elite and replace it with Entertainer Elite. It would be hard to do it properly, but if the good Civ 2 scenarios are any indication, some people are willing to put lots of time and care into scenarios.

My General Policy: If the programmer says it is feasable and the model lead can add it to the model without much trouble, I say go for it.

[Lord God Jinnai]

My view is we should stick with what we have (except for adding a scientific Elite, which i stongly think we need...i'll explain why later), but allow those really adventuresome people out there the ability to make new social classes. This doesn't haveto be something an average Joe can do, but someone who is good at working with the scerio already and knows a little of scripting language perhaps. Anyway such things aren't uncommon for varying levels of complexity in designiong scerios in games, in fact its quite common because it gives new people a chance to explore the process of making a scerio while more advanced users can get more into the guts and add stuff others never thought of.

But if that can't be the case, I'll go for a fixed set vs more arbitrary (so long as there is the sci elite)...besides most of the stuff was either ethnic groups or races which are handled elsewhere.