I am working on the design of a Game development tycoon style game, and trying to figure out how to effectively simulate consumers. The player will be competing against, and cooperating with AI controlled companies for the sale of consoles, games, and game engines. I want the market to flow similarly to the real market. An example of this is Halo for xbox, it kinda started the boom of FPS games becoming mainstream. Another example was Final Fantasy VII for RPGs.

My current Idea is creating consumer objects to represent a smaller part of the market. Each consumer would have a tech and value score for deciding which consoles to buy. It would also have a budget for each month to purchase software, and have a list of liked genres, and minimum game score. if the player or AI release a game and its above the game score and in one of the liked genres list the consumer will buy the game, if they still have room in their game budget. Multiple games with a really high game scores in the same genre being made in the same year or two can influence the liked genre list for consumers.

The market as a whole will trend from great, consumer growth at 80 to 100% all the way to bad consumer decline at 5 to 10%. I was thinking that each consumer object would represent 1000 actual consumers, with some randomization so all sales numbers aren't exactly 1000. My problem is the size of the consumer base. I mean its probably not too bad when the game begins and there are only 1 or so million and I need 1000 consumers objects. When the market grows to something like it is today with 300 million or more consumers.

My current Idea to deal with market growth would be to boost the number of actual consumers each consumer object represents when the actual base reaches certain milestones. At 10 million actual customers the objects would then represent 10000 customers, and at 20 million objects would represent 20000 customers. This would keep the max consumer object number at 1000.

Does anyone else have any ideas for representing the consumer market in a semi realistic way?

Maybe do things such that you track various regions and their populations. Within each population, the people fit into different age groups. Age groups tend to embrace certain sub-cultures which then tend to gravitate towards certain types of games. Since people can also have a variety of tastes, it could be as well that individuals might subscribe to more than one sub-culture so if you were to total the population in each sub-culture you get a value that's greater than the population of the region.

I don't really know how the calculations would go for handling growth and changes over time but at least this way you'd be handling a smaller set of numbers.

Hi moneal,

We cannot make long-term forecasts in the real market with any more accuracy than dart-throwing monkeys. An accurate long-term simulation would be an entirely random market with somewhat predictable seasonal changes and unpredictable recessions and booms. We can make short-term forecasts based on current momentum in the market with better accuracy than random chance, however.

Given that so many people even in the face of evidence still believe they can accurately predict the stock market, a completely random market is likely perfectly fine for your game. Trends, hits, and flops will establish themselves “naturally”.

I recommend this article on Forecasting: Myths and Reality for some more information. And since your game seems to be about economics, I highly recommend the book Thinking, Fast and Slow by Daniel Kahneman, winner of the Nobel Prize in Economics, that goes deeply into decision making and behavioral economics, and might give you some great ideas for your game.

Cheers,

Chris

http://iopscience.iop.org/1367-2630/14/6/063018/article

The 'hit' phenomenon: a mathematical model of human dynamics interactions as a stochastic process. Akira Ishii et al 2012 New J. Phys

Maybe this sort of modelling can help?

I am planning on using a consumer object because I want it to change likes and dislikes depending on the games being made by the player and AI. Like I mentioned in my first post, if a transcendent game, like Halo, gets made by either the player or ai, it will start a boom for games within that genre. I also plan to use events like you say for somethings but not really for genre popularity.

Honestly, from what I'm reading your premise seems to be that the market is largely predictable when in reality the market is largely unpredictable. The best simulation of an unpredictable distribution is a random distrubition. So your starting point should be a random distribution that you add just a tiny bit of non-randomness to (e.g., seasonal trends, highly marketed games being just a tiny bit more likely to succeed, and so on).

Like I mentioned in my first post, if a transcendent game, like Halo, gets made by either the player or ai, it will start a boom for games within that genre. I also plan to use events like you say for somethings but not really for genre popularity.

As I said before, you will inevitably get exactly these kind of trends with a random distribution. Genre popularity and game quality are not even large predictors of success. FPS can be the most beloved genre at the time, and you can make the best FPS you have ever made, and it can still sink like the Titanic on the market because World of Mariocraft was announced the same day as your game, so no one really cared about FPS #135323.

If you want an accurate simulation of the real market, start from randomness. If you want a mostly predictable market, don't try simulating the real market.