Do you think it would be possible to build a consensus algorithm for scientific consensus?