CE Under the Hood compared to Other City games

Thinking of the (awful) simulation model of Simcity (Online), compared to Tropico, Anno series games, Simcity 1-4, Children of the Nile, etc has there been much discussion on the game architecture?

For example - will individual sims be fully simulated going about their daily activities (Tropico, CoTN), the general populace statistically modeled (simcity 1-4), or some hybrid inbetween where individual sims are shown walking around, but are actually modeled as a group (Anno series)?

Another big question I have is the services - will a service provider (church of the cog warding off demons, firehouse, bars, etc), push out its effects in a radius (like Simcity 1-4), send out individual walkers to provide the service to anything they encounter (CoTN, Anno), pull in individual sims (like Tropico), or something else entirely?

I'm sure I can come up with other technical questions - but I think these are the big two. Its fun to think about the experience part of the game, but I think what choices are made regarding mechanics are fascinating as well.

 

Cool, thanks for the quick response! Reading the weekly updates here helps take away the bad taste I get from the new Simcity

Are you guys up to the point where you've prototyped/tested the service providers yet? Or still musing on the design for it?
Walkers can be... annoying... sometimes (never making it to a particular building, circling around, etc), even if they are a tried-and-true city building mechanism.

 

The higher the madness the more random the turns?
Oh, the good ol' days where you had to put roadblocks everywhere to keep your walkers from running off...

 

While you can hate SimCity for it´s awefull start with the awefull servers, I would say it´s simulation is actually quite good. We are talking about the new one, right?

Because it actually simulates the lifes of the agents in my city, with personal money and happines value that depend on the jobs and actually work across multiple cities.

The simulation of CE will of course depend on the AI of the citicens. In DF my main problem was that certain jobs weren´t done and I didn´t know why. Please make it so that I know why my people do or don´t do certain things.

 

The more attention deficit the citizen, the more random turns made, sometimes not at intersections and into buildings. Not through the door, more against the wall while reading the post and trying to fish around for some coins for the urchin who delivered it.

Attention deficient citizens make for better poets, primarily because their lives are much shorter.

In all seriousness though, sending walkers around randomly probably is not a very bad strategy for evenly covering an area. Is the smarter system based on inferring and servicing patrols? Individual route segments form a category (for two routes from A to B and from B to C, they can be composed into a route A to C, and there exists a "do nothing" route for each possible place P from P to P), tuples of two stops a semigroup iff there exists at least one viable route for all possible pairs of stops (for two routes "A to B", and "C to D", there exists a route that is the sum of those two), individual stops that need merely be supplied with a "source" destination that has to be returned to form a monoid (exactly the same as the semigroup, but there is a unique destination None that can be seen as "stay where you are and do nothing at all" which never changes any result when you add it to other stops), and this whole sentence forms something horrible (at least to my eyes)..

The last structure could be termed an itinerary, and as a monoid it supports singleton, appending, and zero operations. Very intuitive to work with. In my senior engineering project for the local school district this property was used by me to easily generate random routes for a harmony search optimization of bus routes. Patrol routes could be similarly selected for based on how closely they match the provider's intended radius of effect, how much their cover, how little they overlap, etc. The computational cost of this can be constrained to a constant amount as with any kind of iterative randomized "genetic algorithm" such as a harmony search, and can be made deterministic through a fixed random seed, and harmony searches can be done with in place destructive updating of a 2D array. It might actually cut it in a game.

 

It does at the surface - but a closer examination shows they took a lot of shortcuts to make it seem like the agents are real. They are not really agents like you see in CoTN or Tropico. Its like a simulation of a simulation.

---

Walkers are tough to get right. Make them random, and places right next to a service structure get ignored. Make them intelligent, and they actively avoid some area for reasons you can't figure out. Best I can tell as a player is that increasing the number of walkers helpd - and that cutting down the range as well. Assuming you don't want to do something cheesy like make your town a labyrinth/spiral, or make roadblocks to herd them where you want to go.

 

Which shortcuts would that be?

I mean the only things where it matters, employment, money transfer and traffic, seem to be simulated good enough. At least I can´t see a difference between how I think the agents should behave and how they behave. Even if it isn´t a true simulation, what different does it make if it get´s the same results as a simulation?

Newtons laws also don´t represent reality right if oyu consider quantum mechanics, but they get the same results on a nonmicroscopic scale, so we can use them without error. Same goes for simulations I think.

 

It's only getting the results if you play casually. A lot of people hacve noticed some pretty bad flaws with services and traffic as cities get larger and they put more hours in

This euro gamer article is a good roundup - there are a lot if others out there
http://www.eurogamer.net/articles/2013-03-15-simcity-review

 

Suffice to say that in the newest simcity, a lot of the population is actually fudged, with a few actual agents circling around, but even them are actually totally simplified in their behaviour and their behaviour isn't even remotely making any sense if you look closely (the most obvious one is that no agents own housing, they just go to a "free" housing unit when they "leave work", obviously that cause many issues down the line, but it's not the only thing).

People on reddit have made a graph to explain the .GetFudged formula that Sim City uses to pull out population numbers out of nowhere :

 

So what is on the x-axis..?

 

Also population, but at a different scale.

 

Found a nicer one;

 

I made this one since the discovery of the code: https://www.desmos.com/calculator/zsfmsiiltr

Red line is GetFudgedPopulation; Green line is real population; Blue line is to help see the exponential curve better.

Having an artificially inflated population is not the problem, but how Maxis calculated it boggles the mind.

 

Well i'm sure it's the sort of numbers people would like to see if it involved money .. Or steam.

 

Hopefully this point of comparison will yield some differences.

 

The Sim City main music will forever be associated with ludicrous situations in my mind because i keep seeing that kind of video ^^

 

I guess in my mind, the CE world is actively trying to destroy your colony and will ultimately prevail over your attempts at survival.

 

I hope so, Dwarf Fortress can get extremely easy real fast after establishing basic needs production unless you actively handicap yourself.