Interesting concept. I only skimmed, so maybe the rationale is explained in the article, but I'm a bit surprised by the "reverse subtitle" design aspect. That

Interesting concept. I only skimmed, so maybe the rationale is explained in the article, but I'm a bit surprised by the "reverse subtitle" design aspect. That is to say, rather than seeing the original language by default and getting some side-bar with translated text to help you over bumps, you play the translation with a gloss of the original text in a side bar. So by default you kinda need to put your eyes where the translated stuff is. Isn't that noticeably less efficient if the aim is learning the original language?