Personally, I feel like the largest issue we'd face in asking for more interesting dungeons isn't the "community" (whose opinions, as mostly sometimes reasonable sentient beings, change as the circumstances or object of their opinions do), but the
flat roulette rewards. So long as the roulette rewards do not scale with the content in question, or do not share a certain bonus cap (liekly with some day-to-day roll-over) with other roulettes, you always want to roll the quickest dungeon to make their bonus that much more efficient. If a dungeon A takes 20 minutes for 40 tomes, dungeon B takes 30 minutes for 60 tomes, and C 40 for 80, that just allows for variety and choice. But seeing as most will only run the one roulette-bonused dungeon per day, we instead have 20 minutes for 90, 30 for 110, or 40 for 130, and voila, there's a clear most efficient choice. And thus, to those who want to make the most of their time, there is no choice, and all designs but A become superfluous.
In practice, though, there are of course further complicators. For instance, mere dungeon length would not give sufficient variety. We'd also want to allow for differences in difficulty, in which case those times will vary.Let's say we have Players 1, 2, and 3, at differing skill levels.
For a party all of whom are, on average, at the skill level of Player 1, with no outliers so great as to solo-carry, A takes 30 minutes, B takes 50, and C takes 85+, if even possible. For that player, if rewards remained as above, A is faintly more efficient than B, and C would be something of a non-option, at least on a daily basis.
For a party all of whom are, on average, at the skill level of Player 2, A takes 20 minutes, B 30, and C 40. Thus, they have every option.
[Cut due character limit]