To do this, we adopt a simplifying assumption – agents should disagree at the same rate as they agree. We already know one end of this spectrum -- the amount of agreement when agents are aligned (gap = 0). We establish the disagreement side (gap = 4), by assuming it to be the inverse of agreement