Traduttore, traditore! For translation is at best but a pis aller for the original—approximation rather than identification (L. A. Willoughby, p. 11), there is no synonymy between words of different languages (J. Lyons, p. 458). Therefore, as a technique of teaching meaning, translation is in the long run unsound (David Arthur Wilkins, p. 130). However, there are cases where translation has to be used, for example, if the students fail to understand, the teacher may ask for a confirmatory translation or translate briefly and move on (L. G. Alexander, R. H. Kingsbury, and M. C. Vincent, p. 16). The fact is, there is usually no full equivalence through translation (L. A. Samovar, R. E. Porter, and L. A. Stefani, p. 132). Used in L2 instruction, the efficiency and effect of translation has to be viewed with care. Borrowing ideas from information theory and complexity theory, this presentation will analyse in detail when translation has a spot to be applied in L2 instruction, and when translation is better stopped towards the tops as soon as possible. The reason is that L2 translated into L1 is either known already, which provides no new information at all but incurs double-learning at least, or unknown, which creates new words or new ideas in L1, thus proving that L2 is either better or more advanced linguistically or pragmatically.