Our art culture regards originality as the most important factor in determining artistic value. The rule certainly applies to music. If a composer merely follows a set form, critics and scholars say that his/her music is rated as "second-tier" and is not worth further studies. Such music does not last long. Even if the music has been popular by the time it was written, it would soon be forgotten. We welcome those who add something personal to the set form, those who are inspired to break through rules of the time. We therefore adulate Mozart, and forget about his colleagues and contemporaries.
This rule might be easy to apply to original composition (In fact it is not that easy. Why would there be occasional revival of "second-tier" composers like Janacek?) but it triggers more problems in transcriptions. For example, when I hear a transcription of Chopin's piano solo music for violin, the first thing that comes to my mind is "Okay, yet another transcription for people who only know popular tunes in classical music." Transcriptions are not original, and one essential motive to write transcriptions is to popularize some certain unknown music or to enforce the popularity of already well-known pieces. To put it simply, transcriptions are written for the average listeners. They are not "art for the sake of art." That is why scholars and "learned listeners" despise transcriptions.
Strangely enough, transcriptions made by people other than the composer are usually regarded as lacking sophistication while those made by the composer him/her-self are viewed otherwise. Bach turned his violin concerto into a keyboard concerto, and it is completely welcomed by us. Mozart transcribed some Haydn music into his early piano concertos, and yet that is not appreciated. Contrary to this rule, Ravel's orchestration of Mussorgsky's Pictures at an Exhibition receives much more attention than the original work. Good transcriptions should bring in something new and original about the piece, as Ravel brings in interesting colors and imagination through the use of orchestral instruments. A better interpretation of the rule is that transcriptions are good and acceptable only if they are original and present something new. But how much new stuff is "good"?
To extend the question, one may look into the appropriateness of adapting Bach's keyboard music to the piano. Scholars who work on "authenticity" of Bach's music simply deride playing Bach on the piano. Although it is true that the piano does not produce Bach's intended sound, I argue that his music is beyond the limit of the instrument. This argument brings up another problem. Playing Bach on the piano is just similar to playing Chopin's nocturne on the violin. Adaptation is different from transcription, but they follow the same argument. Does playing Bach on the piano not add anything new to the music? I believe it does, as the piano has much more variations in timbre and volume than a harpsichord. Goldberg on the piano creates imagination that a harpsichord cannot achieve. Can anyone claim that Bach would definitely hate the sound of a modern piano? Even Bach specialists cannot yet agree on all little details about how to "authentically" play Bach.
Maybe rules just don't work as nice in art as in science. A solution is to put aside our pigeonhole and judge case by case. A good musician should develop a good sense of judgment on when to apply rules instead of blindly following those set by the authority.