Worst negative pattern I've seen is hyper defensive programing. E.g. try: something_that_should_not_fail_and_if_it_does_our_assumptions_are_all_wrong()

Worst negative pattern I've seen is hyper defensive programing. E.g. try: something_that_should_not_fail_and_if_it_does_our_assumptions_are_all_wrong() except: fallback_that_will_not_result_in_correct_behavior_but_make_failure_hard_to_detect() But, of course, it depends a lot on which models you're using and how you instruct them.