I think overall it's a good habit to go through your proofs again, and remove unnecessary double negations, and make proofs 'more constructive'.

Reductio ad Absurdum makes coming up with proofs easier (you have one more information to use, you can work from both ends of the problem and try to make 2 half long proofs meet, instead of one normal long), but in the end it is often unnecessary, you can remove it, and your proof reads better.

I don't share his view that Generalized Pigeon-hole Principle makes the normal version unnecessary. The normal version is used a lot in the form it is formulated.