This is the latest graph. Standard blue is lying and pink is believing. In this series I am trying to figure out if '100% less lying' is the advice that forces the changes or whether it goes in stages.

It seems to have stablized a few times near 0 and 100% before falling into advice stability. Dave might find that 30-37% stability area interesting. It seems to repeat but with different believing levels

Here the 10 kinds of advice are monitored. Series 10 represents how many are giving the advice lie 100% less. Series 9 is a combination of how many are giving either the advice 'lie 90%' or 'lie 100%', etc. By definition then Series 1 should always be higher than the rest. And series 2 should be higher than those series above it, etc. But this deserves a blowup around the last spike. First let us blowup the lying and believing around 19321. If you will recall. Each point on the graph represents approximately 2,000,000 interactions, so point 19000 is after 38 billion interactions.

The believers do not respond rapidly to the lying here. As soon as they do the, advice givers come in and run the game. In the next graph I blow up the advice-giving stages

Note here that 'lie 100% less' is not the leader. We have 'lie at least 90% less' establishing itself before that and 'lie at least 80% less' before that. Another blowup is needed around 741 for next graph.

Here we can see that 'lie at least 40% less' is also an early leader and maybe even l'ie 30% less' before that. It is hard to see series 1 because it is mostly covered by 3 and 2.. When we plot the exact advices instead of those who give advice 'at least as much as a level' we get the next graph.

Here we see the 40% advice level getting an early start and then the 80% and then the 90%. The reason everything goes down but the 100% level at the end is if someone has the program 'to lie 90% less''to and also has the program lie 100% less' then the latter rules. This graph is a plot of the ruling strategy. That 100% person might still have the 90% strategy in the background, so that if his children mutate away from 100% then they might still have the 90% as backup. But It is not clear from this how many backup genes we have. In the next pictures we see how many lie genes of each sort we have.

It looks from this as if the 'lie less' backup genes were all established at about the 50% level plus or minus a bunch. But it is hard to tell from this if they were important genes. Sometimes a useless gene gets established because it just happened to be in some guy who had a good gene. After 5 or 6 generations everyone in a population of 200 calls him great-great---- grandpa. So we can't really tell here, but the fact that the percentages are mostly in the same magnitude of 50% might be an indicator that they all played a role. If a bunch were 100% we'd worry about the grandpa effect. This is weak reasoning but food for future studies.