The important thing to change would be that the consumers of research need to care about reproducibility. Currently, that is only the case in very few instances and there reproducibility is typically checked quickly.