Text this: On the accuracy of aposteriori error estimators