By that logic, consensus should be the only thing you would trust. Paying people for new findings disincentivizes consensus, unless perhaps those people have several avenues to explore and are willing to close off some of the less promising ones - again in favor of consensus being trustworthy.