True, but I have argued for several years elsewhere that percentage correct is not nearly as meaningful a statistic as some have claimed it is, and not just in Geography. Selection bias strongly inflates the percentage correct number, especially in niche categories. Single questions are much less affected by this sort of bias, since the New Question Game exposes these questions to a fairly representative cross section of the FT public. But in certain niche categories, the base quizzes from which the Duel questions are drawn are significantly distorted by selection bias. The kind of people who play "Diseases" quizzes (such as me, obviously indicating a serious departure from the norm, just from that fact) is very different from the kind of people who would play "Diseases" as a Duel category. Maybe 75% of the questions may be correct in the source quizzes (even assuming completely standard method of play), but the % correct for those same questions in the Duel setting would be significantly lower as many players who would not normally play "Diseases" as quizzes will play it as a Duel, especially Gold Members who regularly play all 15 Duels a day.
By the way, the increasing number of photo quizzes and match quizzes does help prevent the "loophole" I mentioned in the overall numbers, but that is totally useless with regard to Duels, since questions from photo quizzes and match quizzes are excluded from being in Duels. I am less worried about inflation in Geography quizzes than you, as only a small minority of Geography questions are actually traceable to quizzes used in the Bus Ride. I happen to love Geography, so I play it not just in Duels, but in the World Quiz, in Pot of Gold (where it regularly dominates one day of the week) and in the Global Challenge. Geography is a much broader category than just the Bus Ride (non-photo, non-match) quizzes.
We should be suspicious of high percentage correct in *any* type of quiz-based question. That stat should not be afforded the significance people routinely attribute to it.