Every evaluative agent deployed into a design independently reports a measure of whether the design fits its criteria. The better the design, the lower the reported unfitness, with 0.000 representing the best possible design.
An agent's evaluation is then averaged with all others of the same type. The resulting average is weighted by the "Weight" value when it is averaged into the total unfitness. A higher weight means that an agent type's opinion is more highly considered. To disregard an agent type's evaluation entirely, set its weight to zero.
Weights are relative only to each other. Setting all weights to 50, for instance, has the same evolutionary effect as setting them all to 10. However, remember that you are only seeing the weights for a single zoneType at a time. If you set the weights of all agent types to 10 for a single zoneType while keeping the weights at 50 for other zoneTypes makes that zoneType's characteristics 5 times less important to the overall evaluation than the other zoneTypes'.
Therefore if you want to concentrate the system's evolutionary pressure on a single zoneType, set its agents weights high and all others low. The system may respond by trying to minimize the number of shapes of this zoneType because it is difficult to achieve a low unfitness with highly weighted agents. In that case, use the AgentNumShapes to indicate that it is important to keep a presence of this zoneType. You may also want to edit the AgentNumShapes settings of the other zoneTypes to keep them from becoming prevalent.