Both the GUI and script-based indicator configuration requests invoke a class IndicatorFactory. The indicator factory fires up separate processes for each indicator. So the indicators for a given year (or up to a given year) can be run as soon as the flt files for that year have been written out from the simulation, even if the simulation is still running. Doing this in fact can be quite useful for long simulation runs spanning several days -- run some indicators on early years in the output to see if things look OK, and if not, stop the simulation. Of course it is also fine to run any or all of the indicators after the simulation has completed.