@Eugene, my choice of phrasing “might not exactly be” is not meant to cast doubt on your assertion of “that it is not”, which I do agree with. I meant to express something along the lines of “might not be reflective of the actual ratings used”. I use the word “might”, since in some cases, by chance of rounding, the difference between the displayed ranks will in fact correspond to the handicap stones assigned. For example, if the players actual ranks are 3.0k and 5.9k, that gets displayed as 3k and 6k, and 3 handicap stones are assigned.
I understood your explanation, but I thought it left out a crucial factor, namely the effect of deviation, which appears to also be playing a role here in further impacting the handicap assigned.
In the post that I linked, I try to further explain how deviation might impact things (which is a phenomenon that seems to be confirmed by @flovo and @anoek earlier in that thread), since in that thread, there was a case where a game between a 22k and 18k resulted in only 1 handicap, whereas even the worst case (floor) rounding would still seem to dictate that there should at least be 3 handicap.
The rounding effect that you discuss in your previous post would only explain apparent handicap discrepancies of up to 1 stone. However, the person that started this thread specifically complained about a larger apparent discrepancy here: Handicaps
Note that @flovo alluded to the effect of deviation earlier (here: Handicaps), but I thought it would be helpful to further elaborate on how and why rating deviation could impact handicap, since the poster immediately expressed further confusion in response here: Handicaps
However, when I went to look for my old post that explains this, I also discovered that it was in a thread that was also started by the same user that started this thread. Hence, it seems that we are largely addressing the same questions in both threads to the same person again.