IanStewart
mcelroytutoring
Theoretically, this is true, but after searching GMAT Club (and the rest of the internet) high and low, I see no concrete evidence that scores of V49 or V50 are possible with one counted question wrong.
I don't understand what you think a V49-scoring test taker is doing besides answering one question incorrectly, if you think that with only one incorrect answer you can sometimes get a V48.
To be clear, I don't "think" that 1 (counted) question wrong can result in a V48 score. We have concrete evidence of that, in the form of an ESR (see below):

As far as a score of V49, I think there are a few possibilities. One is that a score of V49 simply doesn't exist (you might skip directly from 48 to 50, or vice-versa). Another is that you get some experimental questions wrong, but zero counted questions wrong. The final possibility would be that you get 1 counted question wrong--presumably a harder question than the 1 question incorrectly answered by the person who scores a V48. But the fact remains that
we don't have any verified V49 or V50 scores with 1 (counted) question wrong anywhere on GMAT Club, so there is no way to know for sure just yet.
Good point about GMAT scores being valid for 5 years--hence, it is true that HBS's class of 2019 admission stats are not necessarily indicative that a V51 is
still possible. It could very well be that a V50 is now a perfect score. However, if we take the
GMAC website at its word, then a 51V is still possible (see below).
UPDATE 1/4/17: SEE
THIS DEBRIEF FOR MORE DISCUSSION OF THE SAME PHENOMENON.