As such, I generally consider an undefined symbol to be a syntax error.

Well, it all depends on the grammar, and natural languages are a good deal more permissive than programming languages.

On the issue of such simple and obvious substitutions making such a difference, I agree that it does seem pretty surprising. Syntax / semantic issues aside, I find it an interesting phenonemon.

On the matter of my questioning the seriousness of your post, my sincerest apologies: I misinterpreted flippancy where it wasn't intended, sorry.