We show that interaction in any zero-knowledge proof can be replaced by sharing a common, short, random string. We use this result to construct the first public-key cryptosystem secure against chosen ciphertext attack.
Blum et al. (Fri,) studied this question.