It seems at this point that deep learning language translation is pretty solid.
But it also seems that Google in other countries doesn't index machine translated versions of webpages that are written in English.
That means that we seem to be unnecessarily limiting the convenient access of the non English reading world to lots and lots of knowledge, which seems bad.
How can we make this better and easier? Should Google include machine translated versions of webpages in every language in the search results? Why don't they?
To my knowledge, large companies don't deliver unreviewed machine translated strings to customers because of the legal implications. In any machine translated interface, the customer has to take active steps to enable the machine translation. The customer is taking ownership of the page and it's content by doing so.
This is for several reasons: 1. Machine translated strings are sometimes literally correct but actually wrong or misleading in implication in the new language context. 2. For legal or technical terms the machine translation just isn't there yet. 3. For some phrases, the machine translation is just gibberish.
Another thing I noticed is that machine translated output atleast from google uses a very formal tone. People just don't write or talk like that.
It turns out that machine translation is not yet 'solid' as your post assumes. It often generates misleading results and sometimes it is outright incorrect.
Google et al will likely do what you're saying when they get translation working better.