: Two difficult issues for the logic of knowledge have been logical omniscience and common knowledge. Our existing logics of knowledge based on Kripke structures seem to justify logical omniscience, but we know that in real life it does not exist. Also, common knowledge appears to be needed for certain real life procedures to work. But it seems quite implausible that it actually exists in real people. We suggest two procedure based semantics for knowledge which seem to take care of both these issues in a relatively realistic way. What this suggests is that if we really want to understand knowledge, then existing customs and plans must play a greater role than we are used to assigning them.