What the site does is obviously fair use. If the lawyers are concerned about content appearing to originate at the parody site, use a google-style header that explains the meaning of parody.
If the lawyers keep rattling their sabers after that then one could provide a tag or a robots.txt-style file which will stop translation and instead put up jerk error. The biggest disadvantage of this is it suggests that you don't have the right to do it in the first place. Maybe explain in ingratiating detail how you provide the opt-out for squeemish webmasters and companies without the confidance to be laughed at.
What about the potential for defining a subset of each language which *can* be reliably translated? I think this is similar in spirit to the trade languages which popped up in places where differently-speaking peoples came together.
Universal Network Pidgin?
The in-converter would make an effort to identify idioms which are ambiguous or don't translate well and either have the author remove them or encode them in such a way that the core meaning isn't lost when they cannot be present.
The out-converter would convert what it could, then if more information remained present it in another form like footnotes.
Still not a easy problem by any means. But I think if we could find a way to write which could be reliably translated by machines it would be worth the investment.
Slashdot Mirror
What the site does is obviously fair use. If the lawyers are concerned about content appearing to originate at the parody site, use a google-style header that explains the meaning of parody.
If the lawyers keep rattling their sabers after that then one could provide a tag or a robots.txt-style file which will stop translation and instead put up jerk error. The biggest disadvantage of this is it suggests that you don't have the right to do it in the first place. Maybe explain in ingratiating detail how you provide the opt-out for squeemish webmasters and companies without the confidance to be laughed at.
Universal Network Pidgin?
The in-converter would make an effort to identify idioms which are ambiguous or don't translate well and either have the author remove them or encode them in such a way that the core meaning isn't lost when they cannot be present.
The out-converter would convert what it could, then if more information remained present it in another form like footnotes.
Still not a easy problem by any means. But I think if we could find a way to write which could be reliably translated by machines it would be worth the investment.