I had sent a manuscript of mine to one of the IEEE Transactions journals a while ago. After the first round of revision, I received some comments of two reviewers (let's call them A and B). But after the second round of review, the new comments I received obviously indicated that the two new reviewers (say, C and D) were not A and B (both C and D had explicitly noted that in their comment letters). The comments of C and D were mostly contradictory to those of A's and B's. The weird thing was that both C and D complained about some changes which I had applied to the manuscript according to what A and B had requested. Yet, the most perplexing thing, about the most-recent comments I received after the next round of revision, that is, the editor apparently sent the new revision yet to a new pair of reviewers (here, E and F)! Again, their comments are no where close to something harmonious with what the former pairs had asked before.
Now, I am confused to understand the strategy of the editor. To me, a review process has to be closed just like a loop, in that those who comment on a manuscript should assess the resulting revision. Otherwise, anyone else would pose new sets of comments and ideas which may simply contradict the former ones.
Can one explain how this way of review moderation yields convergence to a decisive result?