I second this, but I guess I'm biased towards replicable experimentation.