All difficult conjectures should be proved by reductio ad absurdum arguments. For if the proof is long and complicated enough you are bound to make a mistake so

All difficult conjectures should be proved by reductio ad absurdum arguments. For if the proof is long and complicated enough you are bound to make a mistake somewhere and hence a contradiction will inevitably appear, and so the truth of the original conjecture is established QED. -- John Barrow