Describe the issue linked to the documentation

In the docstring of balanced_accuracy_score, it is mentioned that best score is 1 when adjusted=False, but the opposite is stated in the adjusted argument below.

Is this a mistake or am I missing something?