I’m toying with the idea of programming a game based on The Murder Hobo Investment Bubble. The short version is that Old Men buy land infested with monsters, hire Murder Hobos to kill the monsters, and resell the land at a profit. I want to make something that models the whole economy, with individual agents for each Old Man, Murder Hobo, and anything else I might add. Rather than explicitly program the bubble in, it would be cool to use some kind of machine learning algorithm to figure everything out. I figure they’ll make the sorts of mistakes that lead to investment bubbles automatically.
There are two problems. First, I have neither experience nor training with any machine learning except for Bayesian statistics. Second, it’s often not clear what to optimize for. I could make some kind of scoring system where every month everyone who is still alive has their score increase by the log of their money or something, but that would still only work well if I just use scores from the previous generation, which is slower-paced than I’d like.
Old Men could learn whether or not Murder Hobos will work for a certain price, and whether or not they’ll find more within a certain time frame, but if they buy a bad piece of land it’s not clear how bad this is. They still have the land, but it’s of an uncertain value. I suppose I could make it so they just buy options, and if they don’t sell the land within a certain time period they lose it.
Murder Hobos risk dying, which has an unknown opportunity cost. I’m thinking of just having them base the expected opportunity cost of death on the previous generation, but then it would take them a generation to respond to the fact that demand is way down and they need to start taking risky jobs for low pay.
Does anyone have any suggestions? I consider “give up and do something else instead” to be a valid suggestion, so say that if you think it’s what I should do.
Edit: I could have Murder Hobos workout expected opportunity cost of death by checking what portion of Murder Hobos of each level died the previous year and how long it’s taking them to level up.
The player can be an Old Man or a Murder Hobo. They make the same sort of choiced the computer does, and at the end they can see how they compare to everyone else.
if they buy a bad piece of land it’s not clear how bad this is. They still have the land, but it’s of an uncertain value. I suppose I could make it so they just buy options, and if they don’t sell the land within a certain time period they lose it.
You could charge a periodic “property tax”; that way, the longer a player holds on to a property, the more it costs the player.
I’m toying with the idea of programming a game based on The Murder Hobo Investment Bubble. The short version is that Old Men buy land infested with monsters, hire Murder Hobos to kill the monsters, and resell the land at a profit. I want to make something that models the whole economy, with individual agents for each Old Man, Murder Hobo, and anything else I might add. Rather than explicitly program the bubble in, it would be cool to use some kind of machine learning algorithm to figure everything out. I figure they’ll make the sorts of mistakes that lead to investment bubbles automatically.
There are two problems. First, I have neither experience nor training with any machine learning except for Bayesian statistics. Second, it’s often not clear what to optimize for. I could make some kind of scoring system where every month everyone who is still alive has their score increase by the log of their money or something, but that would still only work well if I just use scores from the previous generation, which is slower-paced than I’d like.
Old Men could learn whether or not Murder Hobos will work for a certain price, and whether or not they’ll find more within a certain time frame, but if they buy a bad piece of land it’s not clear how bad this is. They still have the land, but it’s of an uncertain value. I suppose I could make it so they just buy options, and if they don’t sell the land within a certain time period they lose it.
Murder Hobos risk dying, which has an unknown opportunity cost. I’m thinking of just having them base the expected opportunity cost of death on the previous generation, but then it would take them a generation to respond to the fact that demand is way down and they need to start taking risky jobs for low pay.
Does anyone have any suggestions? I consider “give up and do something else instead” to be a valid suggestion, so say that if you think it’s what I should do.
Edit: I could have Murder Hobos workout expected opportunity cost of death by checking what portion of Murder Hobos of each level died the previous year and how long it’s taking them to level up.
Is it a game or is an economic simulation? If a game, what does the Player do?
The player can be an Old Man or a Murder Hobo. They make the same sort of choiced the computer does, and at the end they can see how they compare to everyone else.
Are you missing a word there?
Fixed. I messed up the link.
You could charge a periodic “property tax”; that way, the longer a player holds on to a property, the more it costs the player.
That would make it even more complicated.