To increase the reliability of computations result, a natural idea is to use duplication: we let several computers independently perform the same computations, and then, if their results differ, we select the majority's result. Reliability is an important issue for quantum computing as well, since in quantum physics, all the processes are probabilistic, so there is always a probability that the result will be wrong. It thus seems natural to use the same majority rule for quantum computing as well. However, it is known that for general quantum computing, this scheme does not work. In this paper, we provide a simplified explanation of this impossibility.