We get this question a lot but I think the focus is in the wrong place. It is less important WHERE or WHO the rules come from than that the rules are structurally coupled with the desired emergent properties. This is in part because all communities are dissipative structures so the question of who/where the rules come from is theoretically an infinite regress problem (but I digress).
The more important thing is, like i said, that the rules are structurally coupled with the desired emergent properties. An example might help here. The simple rules for ant colony foraging behavior are
1. randomly search/find food
2. if find food, return to base
3. if find food, shoot pheromones out of butt
4. never cross a pheromone trail
The first 3 rules are pretty obvious if its a food foraging strategy. But the 4th rule is the most curious. I often imagine two neoliberal/conservative ants debating whether they should follow the 4th rule. One particularly rebel ant is trying to convince others they are being oppressed by rule 4 and why shouldn't they be able to cross a pheromone trail ("we're bein' oppressed! You can see the tyranny inherent in the system! he decries.) But if this rule were jettisoned, it would dynamically alter the desired emergent property of the system.
So, we must look at how the rules play themselves out statistically in the collective dynamics. Not at the ant-to-ant level but at the global level of collective dynamics. Without rule 4, ants would not stay on pheromone trails, thereby decreasing the probability of non-random searching/finding, thereby decreasing the probability of stumbling into a pheromone trail, and thereby decreasing the probability of a surprising and intelligent emergent property. The evolutionary benefit of these rules (hard won through millions of years of evo) would be lost and the ant colony would be selected for, eventually causing the extinction of the species..
So, when we think about purposefully generating simple rules for complex adaptive systems as humans (communities etc), we must look at how the rules play out in the collective dynamics statistically and how they are linked to the emergent properties we desire.