The first graph with advice. This could have been hoped for, if i'd been looking for specific results. Blue is lying, pink is believing, and yellow is an indication of how much advice is being taken. I have 10 kinds of advice. Lie 10% less, lie 20% less....lie 100% less. Notice that ultimately the reason lying converged on 0 and believing 1 was because people started taking the advice. Untimately everyone but one or 2 took the advice 'lie 100% less' I don't know if this happened in stages or whether 100% less advice was the advice being taken around generation 14101. Another thing to note is that without advice taking it converged on 99% and 1%, but that the convergence was not permanent. I think the advice taking is permanent or at least more permanent but I'd have to run many more to make sure.

I have genes that determine the strategy and also genes that indicate that an agent is an advice taker. I'll need to argue that the former is more practical and the latter is more like following a moral rule. Again, I don't think we can argue from this what the rational strategy is, it is clearly different in the middle than at the end. A person who never lies in the middle of this run will die out.

I am hoping that all of the graphs do this, but I have been terminating all the programs when they got to around 99% believing 1% lying, except for this one, so I don't know if they might have later got unstable again. Some of those converged because of strategy and the rest converged because of advicetaking. All converged below 1%. The ones that converged because of advice giving got very close to 0 lying and sometimes had 0 exactly.

A final thing to suspect is that the beginning of taking of 'advice' will not occur when things are already close to optimal. This will have to be checked also.