automationgame.com

I'm not sure, but from what I've seen in the dev update video I had the impression consumers are devided into groups by just one dimension; meaning they are either commuters or families or sportscar fans or whatever, and that's that.
Real human beeings are much more complex. They have gender, age, income, lifestyle etc. and can therefore be grouped by many different criteria. My suggestion is to implement at least a few of these different dimensions/criteria.

My thinking is this:
In each dimension a company has the possibility of sticking out by appealing to a certain group. This would give more possibilities to get into a market and make marketing more complex but also more fun in my opinion.
For example targeting young males would probably result in getting you into the budget-sport category, since there would be a strong covaraince between those groups.

So instead of simple normal distribution you would use multivariate normal distributions with covarainces, at least, maybe even some non-normals.

In my studies I've done such things quite a few times (mainly pattern recognition) and since three months summer vacation is coming up next month I'd have a lot of time on my hands, so if you like I could then give it a try.

The only question remaining is, whether the added complexity would actually be more fun or not for most people. I'm afraid I can't answer that one...

While the groups are somewhat one dimensional like you say, the large amount of demographics result in a lot of demographics that are different but similar. For example, the "fun" demographic and the "city" demographics, where the only difference is sportiness being a factor for the "fun" demographic.

Also, about income, that part is already covered as most demographics are already split in 3 pricing groups.



Also, to bring an example of how an other game works, gear city has like you said multidimensional demographics. The thing though, is that since they also have a lot less categories of cars, it doesn't make that much of a difference. And the way the whole thing is handled in automation with the overlap between demographics ends up being much better IMHO.

Exactly that's what I'm unsure about. Maybe it's really not a bit more fun...

Maybe if I checked out gear city and/or other eco-sims I could form an opinion.

The system you are proposing is the first one that came to mind for us as well, but we decided against it in favor of what we plan now.

The only thing I see the "young male in city"-system doing is adding a layer of abstractions that is not helpful in any way. How are you supposed to know what that "young male in city" likes for cars? It would be a distribution across the different categories we have planned right now and that would be the case with all other categories too. It would not be adding anything to the game but make it less transparent because instead of addressing the actual car you are building directly, having vague groupings doesn't tell the player much about the actual needs of that demographic.

1) "young male in city"-system structure:
Group 1: Demographics Size 1, Car Type A (A%), Car Type B (B%), ..., Car Type Z (Z%)
Group 2: Demographics Size 2, Car Type A (A%), Car Type B (B%), ..., Car Type Z (Z%)
...
Group N: Demographics Size N, Car Type A (A%), Car Type B (B%), ..., Car Type Z (Z%)

2) Our planned system:
Car Type A, Demographics Size A
Car Type B, Demographics Size B
...
Car Type Z, Demographics Size Z


Now take a look at system 1) and multiply out the demographics to be able to apply this to cars in the market:
(Demographics Size 1) * (A1%) + (Demographics Size 2) * (A2%) + ... + (Demographics Size N) * (AN%)
(Demographics Size 1) * (B1%) + (Demographics Size 2) * (B2%) + ... + (Demographics Size N) * (BN%)
...
(Demographics Size 1) * (Z1%) + (Demographics Size 2) * (Z2%) + ... + (Demographics Size N) * (ZN%)


and poof... you are now exactly where system 2) is. This illustrates that system 1) doesn't add anything but complexity for the sake of pseudo realism.
I hope that illustrates why we chose to do it the way we did.
Cheers!

exactly, there would simply be some function mapping the demographics-space onto car-feature-space and it would be the players job to tinker around with marketing in order to guess that function as well as possible and exploit it.

I wouldn't say there was "nothing" added this way, but I'm not sure it would actually be fun...

Anyhow, it's not going to happen, so case closed.