When running multiple studies on the same topic, researchers usually want to ensure participants are naive to the research purpose. Super Exclude accomplishes this feat by excluding even workers who started a past study but did not finish it.
Most Exclude features work by ensuring that workers who took and were approved for a past HIT do not also accept a future HIT the researcher does not want them to take. The Super Exclude feature goes a step further by ensuring that any worker who merely accepted a previous HIT does not also take a subsequent HIT the researcher does not want them to take. Super Exclude is useful when you're running several studies on the same topic and you want to ensure workers are unfamiliar with your measures and manipulations.
Super Exclude ensures that workers are completely naive to your study by excluding any worker who may have started one of your past studies but not finished it. This means a worker who accepted a previous HIT and completed say 50% of the study—enough to see your manipulation and some of the measures—would also be excluded from your subsequent study. Super Exclude is most useful when running multiple studies that investigate the same topic with similar manipulations and measures.