Reviewer: Manfred Georg
Date: 9-8-2005
What is the consensus rating for this paper? top 25%, but not top 10%
How confident are you that the consensus judgement treats the paper fairly? unsure
How would you rate the overall review process for this paper? ok, but uneven
How did the disagreements among the reviewers get resolved? Or if you were unable to reach consensus, what were the main sticking points?
As is common in general and especially at the beginning of the semester, I feel that the ratings on the papers have been excessively inflated. Although the consensus of most reviewers states that the paper is in the top 25%, one reviewer couragiously places it in the bottom 50%. I tend to agree with the latter.
Several points were brought up in the reviews and during discussion. Foremost was the concern about the validity of the assumptions used in the paper. Particularly, the assumption that each flow maintains a consistent RTT (Round Trip Time) seems unrealistic, given that queuing delay significantly effects RTT.
The next major concern was over exclusive comparisons with Zing. That is, the paper only dicusses differences between Badabing and Zing. This is only valid if Zing is the accepted, standard method to actively probe the state of the network. No one could make a conclusive statement as to whether this was true or not.
In general, the paper is not particularly well written. Both the graphs and the proofs could stand to be explained in further detail. There were hidden implications of their results which were not explained. Similarly, there were hidden assumptions that were required by their proof. An example of the latter is that the first proof implicitly assumes a discritization of the packet stream which is not justified.