I rencently read a book that touches on such social simulation - "An Introduction to Multiagent systems / Michael Wooldridge".
He mentions p.214-218 such a grid like social simulation that was done in the EOS project, undertaken at the University of Essex in the UK. The goal of the EOS project was to investigate the causes of the emergence of social complexity in Upper Palaeolithic France and they were using a 10000x10000 grid.
You could also be interested in "Simulating Societies using Distributed Artificial Intelligence / Jim Doran"
I remember reading in an old AI textbook that some researchers had done an experiment in genetic algorithms where the goal was just for the program to stay in memory. So, taking all the artificial rules out of the game. And supposedly they observed symbiosis and all sorts of stuff. Can't find that anymore though.
However, to be honest, the part about how behaviours works at the beginning of the explanation linked, is not.
Society and sociological behaviours (not to mention economical) are more complex than "whatever behaviours sticks around is the fittest". This stems from the fact that one is living in a modern world and is (to a certain extent) well fed, well educated, ...
You only have to look at the multitude of society that have survived until now to see that that kind of thought is not valid. For example, in some tribe, war is stopped when one life is lost (I believe this was Palau), or in others, wives eat the brains of their deceased husband. In some, matriarchy is the norm. Etc.
It is very difficult to conclude a simplistic view of society/sociological behaviours.
> You only have to look at the multitude of society that have survived until now to see that that kind of thought is not valid
Thinking longer term, it could also be that not enough time has elapsed or that there was/is too little social interaction between all players for one dominant behavior to emerge.
It might also be that there is no one dominant emergent behavior, but several, all contributing to the overarching goal of survival (think Nash equilibrium).
> You only have to look at the multitude of society that have survived until now to see that that kind of thought is not valid. For example, in some tribe, war is stopped when one life is lost (I believe this was Palau), or in others, wives eat the brains of their deceased husband. In some, matriarchy is the norm. Etc.
How does this invalidate the thought? It's like saying survival of the fittest isn't valid because of all the different animals that have existed.
I meant that sociological behaviour does not have "one true adage" (i.e. survival of the fittest, or whatever-else), and cannot be simply concluded that way. ("It is very difficult to conclude a simplistic view of society/sociological behaviours.")
The examples I gave were meant to show that the behaviours that one believe to be "fitting" for a certain worldview, is suddenly not "fitting" in another group.
This is due to the key concept that most students learn in Sociology 101, being in a group changes your behaviour.
Therefore to study "behaviours", if you conclude a simplistic idea, you have to define its boundary. Maybe in, say, the Napolitan culture of Italia, a certain behaviour explains a certain reasoning holds true, but to generalise that to "behaviour" (in general) is hand-waving away many complexities.
> The examples I gave were meant to show that the behaviours that one believe to be "fitting" for a certain worldview, is suddenly not "fitting" in another group.
I think this is where we have different viewpoints. To me, a guiding principle that exists like "optimize for survivability" isn't something that people would actively agree to, hence to "belief" is required.
It is like saying "not everyone believes in gravity, yet it still exists"
In evolutionary terms, "fit" means like a piece in a puzzle and doesn't an ethical connotation or anything like that. More like a model fits, eg in the above game related to ML.
I think you're missing the point. Fitness in this sense has nothing to do with right or wrong or validity. Those examples you gave could be quite "fit" relative to other choices, and thus survive for a good while.
Basically we're talking more about genes and memes, fitness is whatever occasional mutation actually perpetuates, in the particular context/environment. Social behavior is partly memetic, and partly genetic.
Whatever behaviours are fit are just the ones that stick around. In, say, the developed world, we have 2 main competitors: have lots of kids, and try to teach them to behave as you do; or have fewer/no kids and try and spread your particular mind-virus to those not carrying it yet.
Certainly, if one generalises anything far enough, everything holds true in broad strokes.
It's like saying if we look at history by the billions of years, what we do has no impact whatsoever (what you do today has no meaning since your imprint or anything that you affect, would have no effect when considering the history by the billion-years-scale).
Yet concluding that would be avoiding the issue of looking at the reality that, yes, what you actually do today does matter, it matters to your family, your loved ones, to your friends and colleagues, your neighbours and so on.
So that was what I meant by saying: "Hey, it's fine to model things and have fun with it, it's another thing to believe that your modelling explains anything ... it does not."
Having just read the readme I think this is perhaps missing two layers to model human behavioural 'evolution'. It's a super cool project though despite that and I wonder if you could extend it to put the two missing layers in...
The way I see it you have the evolution of individual humans at the genetic level - so this gives them built in desires and instinctual behaviour.
You then have the learning of the individual over the course of their life. Attachment theory is basically a load of research on how this can go wrong if the very critical early stages of emotional & interpersonal development aren't well supported in childhood.
Finally humans are social creatures, individual humans aren't well adapted to surviving 'in nature' so most of the ~500k years of evolution for humans going from anatomically modern to behaviourally modern there was selection pressure for genes and memes that resulted in survival at the level of human cultures.
It would be interesting to model open source contributions using something like this. Perhaps it would reveal something about the economic paradox about why people contribute to open source.
He mentions p.214-218 such a grid like social simulation that was done in the EOS project, undertaken at the University of Essex in the UK. The goal of the EOS project was to investigate the causes of the emergence of social complexity in Upper Palaeolithic France and they were using a 10000x10000 grid.
You could also be interested in "Simulating Societies using Distributed Artificial Intelligence / Jim Doran"