think i have cracked the design. Im gouing to jot in down quick before I forget ;)

I am going to scrap the idea of persistant voters. The voters are generated at the last minute just before they vote.
This means that the percentage that determine how many unemployed there are etc, now become 'live data' that can be affected by policies.
So if a policy cuts unemployment and upsets the unemployed, then I can upset the neuron that is the voter group "unemployed" but that group contains no voters.
Then at election time i generate the population using the current value for freq_unemployed that the simulation gives me (which is likely to have changed since last election) and then those voters are affected by the voter group "unemployed".
So i can strill have groups of people in a number of different groups at the same time, but now I can vary those groups size as well as thei appreication within the simulation itself.
w00t.
back to star wars galaxies...