Coincidentally, I was at a congress last week where I was first introduced to the concept of pre-registration. After the speaker was finished, similar questions were raised in the audience as mentioned in the course  such as limitation of creativity. I do not really see this problem, as it sounds like you can still introduce new/additional/other data with sufficient backing. The speaker also showed how the paper results published from pre-registration showed a higher percentage of H0 hypothesis (aka non-significant results), which shows that it is working to some degree: not only publishing shiney, exciting results, but also results that are not as exciting. But this way that same research will not be repeated, because it has been published and people will know! I find pre-registring an interesting development towards fairer science.