All of the tasks that were used to train and test the microtask detection had a duration generated by the crowdworkers. So, not all tasks will be applicable or completed in the same time manner as the crowdworkers reported.