Full write-up here (not gonna do a summary every time, sorry) : https://docs.google.com/document/d/1TD1PCW1IG2BlBQ27GPe5Nqi5phjg4qXwhpgEfJAgbBw/edit?usp=sharing
Since the exact positions of the events and the agents seem to be quite important, I wondered what would happen if their minimum and maximum wealth was limited. For example, a very talented agent who's unfortunate enough to have a run of bad luck might end up so impoverished that recovery was impossible. Maybe if he had some social security to get him through a rough patch, he'd be able to go on to bigger and better things. Perhaps that would help reduce scatter in the talent-money trend. And maybe if agents were forbidden from becoming dangerously wealthy, it would improve social equality and the underlying talent-money trend would become clearer.
Both of these things are very easy to do in this magical simulation land : I simply set agent's wealth back to the starting value if it ever drops below it, and restrict it to some maximum if it ever exceeds it. The maximum was chosen based on looking at the typical wealth distribution that results.
I also realised that until now I've been concentrating on the richest and most intelligent people in the simulation, and not given much thought to the poorest and stupidest. Time to stop neglecting those who need the most help ! Here, the 20% poorest are shown by faint black line while the 20% least talented are shown with a faint red line.
The first row in the figure uses the standard conditions : talent affects whether lucky events will increase wealth or not, but that's all. Capping the maximum wealth doesn't change very much. Limiting the minimum wealth doesn't help the most talented very much (maybe a little bit) but it does prove useful for the talentless and poorest people. The wealth of those demographics evolves in a very similar way. So even though there's still no (obvious) talent-wealth correlation, it has improved things, in a sense : the poorest people tend to be the stupidest*. Hooray, I guess, if crushing the idiots is your thing.
*I need to check that more rigorously though, I'm just basing this on the similarity of the curves rather than comparing agent numbers.
Oddly, capping both the minimum and maximum wealth splits those two demographics apart again : the most and least talented people basically never get any more or less share of the total wealth than at the start. The poorest people, on the other hand, no longer lose quite as much as they did previously.
The second row shows the case of allowing talent to affect both whether unlucky events will cause harm and whether events are lucky or unlucky to begin with. As previously, this causes a clear talent-wealth correlation whilst maintaining the power-law wealth distribution (not shown here). Under those conditions, regardless of whether we cap wealth at all, the stupidest and poorest people tend to be one and the same. Which made me feel guilty : rewarding the most talented is one thing, but crushing the least talented is quite another... I'm so sorry, agent no. 532, forgive me ! It's better than giving the stupid people all the money, I guess, but actively punishing them for being stupid wasn't what I had in mind. Though, if both upper and lower wealth caps are used, all the demographics plotted are much more equal than in the other cases.
I've also found that sometimes these plots look a lot more chaotic, especially if the simulation is given more timesteps. What I think is happening is that a few agents are able to amass vast amounts of wealth, much more than any others, which skews the statistics. I could try removing these outliers, but a wealth cap - a relatively high one - would probably be an easier solution.
poverty may impact the intellectually challenged, but what about the highly intelligent/talented clearly mentally ill
ReplyDelete