Personally I think this is one of the more interesting concepts currently out there. While certainly being extremely ambitiouns the concepts behind it seem quite promising.
Tau is not a Turing complete language (in contrast to Ethereums scripting language) which makes it possible to decide certain properties of a given piece of code. This could prove very useful as one can "ask" questions about implications of a program (e.g. a contract).
I am somewhat lost with the whole "Quad" (context, subject, predicate, object) stuff being able to express programms=proofs=ontologies myself though.
Though I'm not sure I'm convinced by the fundamentals, I invested a bit, since the Bittrex price si almost half of the official crowdsale prices. I'm a little bit frightened that they don't really get the blockchain economics, as Ohad mentioned the idea of sharedropping on BTC owners on Bitcointalk. It won't be done, but coming up with this idea after the crowdsale started doesn't look like a great insight in my view. I would say that it has a good hype potential.