That would be true if the error is not random;even if that were the case one could rather easily observe the objective and arrive at the correct spot. Usually you are dealing with a swinging needle that goes on either side of the true azimuth. In practice, one does pretty well to be within two degrees.