1. Perhaps FISA really does rubber-stamp all requests.
2. Perhaps NSA really is surveilling almost solely targets of real security interest.
3. Perhaps the FISA process and the documentation it generates forces NSA to be overly conservative about surveilling targets, to the detriment of security.
4. Perhaps NSA only uses the FISA process for targets of real security interest, and uses some other process for other targets.
In fact, a 50/50 success rate at FISC would be disturbing for other reasons: it would imply that the USG was aggressively targeting people of no security interest, and only the FISC process was preventing that from happening.