SJ and I did get a chance to follow up with him afterward and suggest that they might considering teaching English literacy directly to non-native speakers (as we did in our Ethiopia pilot, rather than only teaching students in their native language; of course DuoLingo doesn't support a large number of languages, so allowing folks to skip directly to English would greatly increase their reach). They are big on A/B tests and quantification over there, I think we convinced him it was an interesting hypothesis to test.
SJ asked how one could help with the literacy effort and he gave us some contract information. If any of the OLPC crew is interested, drop me (or SJ) a line and we'll try to hook you up.
I think they might actually be able to make a significant dent in the literacy problem.