Pretty easy. You wouldn't even need Google - just store a list of the thousand or so most used words in the script.

I wrote bookmarklet once that replaced every word in a webpage with 'blah'. Rather amusing when used on news websites.

Yeah, definitely not hard at all. Although, I think the harder problem is figuring out a specific word's complexity so it corresponds to the relevant immersion factor. For instance, having a lower immersion level shouldn't just mean lesser words (which is what is right now) but also more common and memorable words. Maybe a type of sentiment analysis problem?

OH! And it works on other sites too, but that's "experimental" since it's not tested on all sites (which is why we put the on/off slider). Might be interesting to build a DOM parser- in the future -that figures out where relevant, "could be translated" content is in the webpage.

It was not very clear to me what the slider meant. Good thing that I read this comment section I guess. The slider 'on' means that the translation happens on all websites, whereas off it only happens on the selected sites?

Yeah, I'm going to make that clearer in the next update. It's much simpler, 'ON' means make the page translatable and 'OFF' means to leave the page alone. Because sometimes you just want to read the damn thing