I run every code sample before it goes in a video, a post, or a PR description.
I learned this from comments pointing out the ones I didn't run. Each broken example costs a little trust, and people remember the broken one longer than the ten that worked.
