That's true if determining whose kneecaps you need to break is either trivial (e.g. open dissidents and protesters) or can be easily achieved using social engineering (e.g. anonymous print pamphleteers). If doing so is a /technical/ rather than social task, and your technical cryptanalysis isn't up to par, breaking kneecaps is much less powerful of a deterrent because your hit rate will be very low. Breaking kneecaps to deter the use of encryption itself is another matter though.