Essays are inherently problematic in most fields, because they're effectively always at least partially graded on the quality of writing, which doesn't have any connection with how well something was learned (which
should have a connection with grade). Some profs separate it out better than others, but students usually have only vague or zero knowledge about who falls where, so the default approach is that, regardless of what the prof states, writing quality matters. Copying from someone who is a better writer than you is a sure-fire way to sound better, which usually leads to a disproportionately-higher grade than an equivalent amount of copying simply the
information.
If I write extremely poorly, the pain of reading through it taints what data exists. The irritation lowers the perception, which lowers the grade. It's a pretty well known phenomenon, lots of studies supporting it, probably most classically known as: sitting in a hard chair makes you more "hard" in your dealings with people.
Note that your example with the presentations only supports the belief that the delivery of the proof-of-learning matters. Not that there's a way to fix it, of course - awesome results are awesome, and should be encouraged. It's simply that I wish more professors were (more acutely?) aware of it, so it would (hopefully) have less impact. I would think that there would be a bit less cheating if delivery didn't matter.