Evidence is a shaky thing on prospects. Lars Anderson's case is very much an exception - one of the weirdest of them all, usually guys who are performing at levels they are young for are very high probability guys. In the long run, betting on guys like that works. (just like, as Tampa would attest, despite the volatility of prospects, drafting very high still works quite well)
The Japanese guy is worth signing if he is better than your alternatives in a meaningful way. The history if Nippon league starters would lead you to believe that this is in fact not true - Darvish's style is much more "American" than the guys they have generally produced. The more typical examples have been the Nomos or Matsuzakas, not pitching to contact, wasting a lot of time etc.