We thank the reviewers and all suggested corrections will be accounted for in the final version. We only address the major comments in this rebuttal. $ Reviewer 1: We believe similar techniques of weighting will significantly improve the performance for general method-of-moments. We ran generalized moment methods given in Azari Soufiani et al. 2014 paper, and simulations confirm that the optimal choice of weights that we proposed gives improved accuracy in comparison to uniform weights. Section 3.2 discusses theoretically the trade-off between computation and accuracy. The price we pay depends upon position of breaking separator. If each ranking includes top-p items, computational complexity is roughly same for MLE and rank breaking. If each ranking includes first m unordered items and an ordering of next p items then complexity of MLE will be O(m!) whereas complexity of the proposed consistent rank breaking will not not depend upon m. Reviewer 2: Impact of positions of separators: The reason the position of the separators is crucial is because the model itself is asymmetric. Consider a simple case when all items are being compared all the time. If we place the separator at the top, then the least-preferred item will still have a reasonable chance of appearing at the top with probability about 1/(e^2b d). However, if we place the separator near the bottom, then the most-preferred item will almost never appear in the rank-breaking, with probability about (1/d)^(e^2b). Hence, having separators at the bottom results in missing many of the top items, requiring significantly more samples to get high accuracy at the top. Intuition behind the proposed weights: Consider an event (A>{B,C,D}), which happens with probability e^A/(e^A+e^B+e^C+e^D). Uniform rank-breaking would be essentially substituting this probability with e^A/(e^A+e^B)*e^A/(e^A+e^c)*e^A/(e^A+e^d), which can be significantly different. Instead, optimal rank-breaking essentially substitutes this probability with e^A/((e^A+e^B)*(e^A+e^c)*(e^A+e^d))^(1/3), which is a much better approximation, intuitively. The optimal weights treat each sampled poset equally, resulting in a better approximation.