This is a really good question that I also think about frequently.
I’ve been of the mindset that scaling in object mode, and then applying the scale before entering edit mode is the best approach by default — just to avoid some of the potential pitfalls that Omar mentioned in his post.
I also found the video a tad misleading at 6:32 where it says “often times you’ll want to scale something in edit mode instead.” To me, that suggests there’s something inherently bad about scaling an object mode — when the crucial point is to just ensure that the scale transform is applied. (That said, it is useful to know that Edit Mode scaling is ultimately what we’re really trying to express with either approach)
