Programming questions (and other similar types of questions) from SO, ServerFault, Unix.SE, SuperUser, etc. are verifiable, by simply taking the answer and deploying them.
Some cryptography questions and answers, as well as some advanced math, physics, and computing science questions from their respective sites are theoretical in nature. When I encounter them, I have difficulty deciding whether to up-vote them. I have once answered a question of this type, and said explicitly it serve as "foundation for further reasoning, development, and refutation".
This is an open-ended discussion for meta: Should we vote on questions and answers that we're not confident about? Solely based on their content merit? Ask the answerers to add disclaimer if they can't provide references?
