"Sir, gamma fleet have radioed in, they've taken the planet, bringing our total to 43. Our port back home has used the extra material to produce an additional three battleships that are approaching our position now. The scientists have declared the new prototypes battle ready, they should be able to fly in atmosphere and most troops they kill will rise again to fight on our side. We've got eighteen billion credits and the last threat to our great civilisation has only two planets left and is in range."
"New information, they just developed their third transportation technology. It's over sir, we lost."
Our first couple of games of Twilight Imperium typically ended this way, with someone winning on victory points in some suitably un-dramatic way regardless of the state of the board. This occurs less often as you get better at the game and once the expansion changes the working of the Imperial card, but it is a feature of a lot of games. In a similar vein, Pandemic is prone to a lot of pyrrhic victories, in which the game ends in victory because you have cures for everything in the lab, but everyone in Asia and Europe is dead.
In the past I've questioned the design decisions that lead to games having the sorts of end conditions that can result in a victory that does not appear to be supported by the board state. Playtesting my own game has done a lot to convince me of the importance of these things. An important problem is that that board state can change in both directions; A particular player (or group) may find that their fortunes wax and wane over a game. This makes for a bad game ending mechanic, as progress towards the end of the game is uncertain, the length of the game varies dramatically and a game could theoretically go on forever. By contrast, games which offer a win by victory points and offer the players options to accumulate but not lose points will march inexorably towards an ending. Games that simply have a turn limit or a time limit can achieve this more directly.
However, this isn't the most important issue. Game designers often comment that a game should stop while it's still fun. Clean up actions are rarely enjoyable for anyone, there's no challenge in removing threats once you have every conceivable tool for doing so, there's no fun in fighting battles that are so one sided that no decision either player makes could possibly matter. Ultimately the goal of an end game condition should be to detect that the game has been won and to end it without players needing to do a whole bunch of boring fiddly housekeeping.
I think that the problems players have creep in where the end game condition is so divorced from the game as a whole that it simply doesn't make any sense for the game to end. Pandemic and Twilight Imperium are both great games, that have succeeded in so many ways, but are slightly weakened by how the end of a game can feel. If the game has been successful and drawn players into the theme and got them caring about it, then it feels odd for a game to end where thematically it's apparent that the winning side was doomed.
In Mage Academy this provides a tricky design challenge. With new threats appearing on a regular basis even once players have acquired lots of runes and know every spell in the book it could still take a long time for the game to end if they had to remove every problem. On the other hand if the game end condition ignores the problems on the board there's every chance that a game could end while the building is on fire and overrun by demons.
I've tried to address this by having victory conditions that would naturally deal with the remaining problems in the academy. Most of them center around the players generating some new power that could be used to deal with their other problems, such as building an army of golems, binding the power of a demon or improving the spell book as a whole. These solutions are okay, but they don't always work, in particular I've been asked how binding the power of one demon enables the players to survive the three other demons running around the academy and I don't have a great answer. To an extent this problem may be something I can only mitigate rather than avoid entirely.
It does work really well for one scenario though, the one in which an inspector shows up and must be convinced that the academy is safe. The visual image of him walking out of a burning building with the wool thoroughly over his eyes works brilliantly.