>>13413770By all means, if you are able, run as many variations as you'd like with how many parameters you want.
I definitely don't disagree that there's many ways to make a more solid/credible/accurate model of human society. Things like scale, limiting interaction by proximity, choosing whether to interact or not (i.e. a more complex chart of possible interactions than simple Betray/Coop), variable number of interactions based on past experience, rate of misunderstandings, etc.
I'd be interested in the results of all simulations you can think of.
Where I disagree is that it would SIGNIFICANTLY change the overall result. More options would skew the result to make it less black or white, but the underlying simple truth would still shine through: instead of a crushing victory, an overwhelming majority. I'm also willing to be wrong on this.
But you can see in the simulation at
>>13413605, even when you add misunderstandings ("6. Making Mistakes"), the optimal but simple strategy that emerges in response isn't far off TFT: it's a slightly more forgiving TFT, he calls it copykitten, which retaliates after two betrayals. As I said in
>>13413553So it's wrong to say that TFT "isn't that good after all".
Again, "nuance" can't make all the difference by itself, it can only nudge things towards the other end. Even misunderstandings don't dethrone TFT.
As there is continuity in concepts, there is continuity in solutions. You might say, "how come iterating prisoner's dilemma gives a completely different result?", and I'll tell you that you have forgone continuity by making a big jump into "iterativity". But the less iterations you have, the more it looks like a single-iteration IPD. This is also included in the simulation at
>>13413605.
If you find a striking factor that completely turns the table on the simple solution, I'll find you the middle point where the situation flips over and I'll take my conclusions from that fact. There's your nuance.
