I wasn't really planning on making it agent-based, since the world would span multiple countries with each a number of cities in them (each the size of a Capitalism II city) and resource regions in between them. A world like that would require some kind of supercomputer to run agent simulations. My idea was to employ a set of detailed statistical models for the citizens, but have these statistics employ a whole range of different characteristics (gender, age, wealth, income, family size, origin, ...) and have these influence each other. This in itself is going to be a very resource-heavy simulation to run, but it will be manageable even with a huge number of citizens.counting wrote:If writing agent-based simulation taught me anything, is that you can't simulate every detail, at some point abstract concept and boundaries need to be set
Working from the bottom up is a good way to give each part of the game great depth, but 'letting it emerge' so to speak can make the game as a whole come loose. Especially for a graphs- and numbers-heavy game like this I think it would be better to try to create and optimise a 'framework' game interface where each part would fit in, and then work out the details on each part. I really want to avoid making the game feel cluttered.counting wrote:started from the actions and action scripts for players, sort of working from bottom up in "city simulation" perspective, and let the macro-level interaction "emerge" out of the micro-level procedures.
Not exactly. I do want to expand the simulation to provide a more detailed worker level and more varied demographic in cities (because in a city not everyone has the same needs, so there should be more than one 'needs fulfilled' graph in a residential building), but the main focus both for the player and for the simulation should be on the business management level.counting wrote:I sense you really want to simulate at worker/labor level actions, but for player's "main focus" is reside at city-management level, it's a huge gap you need to cover
To give an example of that last part, something that's always bothered me in Capitalism II is that you have a product, and you've got one 'demand' bar graph that symbolises the demand for that product. But there is never just 'x amount demand' that goes up or down depending on price/quality/brand (product rating). Wether you build a jewelry store on the outskirts of town or in the center, the only difference is the amount of traffic you receive depending on the residential buildings around the shop. This isn't really realistic. Expensive jewelry will be in very low demand e.g. in the ghetto where in lots of people live but they don't have much money. But, it will be in much higher demand in the center of town, even if the population density there is much lower because most of the room is taken up by shops, offices, large villas and penthouses, ...
I hope the above illustrates what I want to achieve with the more detailed and layered simulation I'm suggesting. Doing something like this agent-based would be extremely resource-heavy, whereas making the simulation statistical but layered and with a lot more parameters will hopefully prove achievable. As soon as I find a developer to create these server simulation algorithms (I myself know a lot about programming in general, but lack the specific knowledge and experience to pull something like this off) I will discuss the possibilities in detail and I'll have a much more realistic view on what's doable and what isn't.