Why do you say "without repeatability and predictability"? Social studies must be reproducible. When they are not, they are discredited (where are you getting your information about social science?) As we speak, there are academic organizations being formed to do exactly that: to reproduce more social science results. Prediction, as I explained, is generally impossible, just as it is impossible in computer science and biology in the general case, but in simplified situations, it exists and it is expected. Not only that, it is being used as a technology. Many manipulation techniques employed by large websites employ social science results to great effect.

I think you're attacking a strawman image of social science that's been constructed by people who really have no clue. That's not to say that the social sciences aren't plagued by many methodological problems, but we actually do know quite a lot.