In most games the possibility space extends both before and after the actual start and end points of the game chosen by the designer. Take something like Settlers of Catan. You don't have to stop at 10 victory points, you could ignore the rules and keep playing. You don't have to start at the game start, with everyone having two villages and some resource cards, you could step backwards and start with just one and no resource cards.
It's obvious why we don't. The "one village" stage of the game is somewhat degenerate, as you need four resources to get past your first settlement and it can only naturally claim three. It'd be possible for two players to hedge a third out of the game entirely. Similarly, once you've got lots of points, the game is rapidly approaching being too crowded for people to do much and since success brings success the winner is likely already apparent anyway.
Setting the parameters of the game such that you cut the most interesting possible chunk out of the play space can be the difference between a good and bad game. The problem is that player's preferences for where these cuts go can vary dramatically between players.
I know lots of people who love Betrayal at House on the Hill. Personally I love the second half of that game and view the first half as a price that's paid to get to the bit that's fun. To my eyes before that point I don't know which side I'm on so there's no point taking any action to help or hinder any other player, so it's just walking around randomly encountering things that may or may not help in the long run. If I could start the game with a bunch of house explored and some random items with the betrayal happening on turn three I would.
However the players I play with hate that idea. They see it as building a story, it doesn't matter if decisions have no tactical relevance. The things that happen to people in their meanderings set a context which makes the theme of the betrayal come out more strongly. "Ted became a ghost" doesn't have the emotional impact of "Ted got trapped under a rockfall and none of us could be bothered to dig him out, but how he's back for ReVeNgE".
As is always the case with subjective opinions on the enjoyability of games, nobody is right or wrong, we just want different things out of the game. Generally a game can't please everyone and it's better to excel for one group than to do kinda okay for everyone.
Which is exactly what I didn't do with Wizard's Academy. I had a lot of great responses from playtesters who really enjoyed the full experience. Starting with an empty academy, trying to build resources to be ready for the worst, seeing a tiny flame grow into a raging inferno. That's great and that shaped a lot of the development of the game.
However I've also come across players who see it the way that I see Betrayal: They love the second half of the game, once they've got some resources available and there are some serious problems that must be faced. I like it when people are happy, so relatively late in the development added the "expert scenarios" which start with half of the spells locked and some serious problems down on the table.
Now this is pretty great in terms of opening up the game to another section of players, but it's also problematic in that starting at a critical moment requires immediate competence. That leads me to mark the scenarios as expert, which makes sense on one level, but doesn't do a good job of communicating that the choice between regular and expert is one that might depend upon the nature of the player's preferences more than their proficiency at the game.
Furthermore, the dichotomy between the game modes means that everything does not always function as intended. When a game depends upon a single random outcome then it can be heavily luck based, when it depends upon the cumulative effects of lots of random outcomes then an "edge of the bell curve" result becomes less likely and strategy matters again. Some elements balanced around a full game, while functional, don't remain perfect in the half game setup.
Now the result is still pretty respectable. More people than not rate Wizard's Academy as a solid 8 (22 vs 19) but reading some of the comments it's apparent that there are some gamers that have not been served by this approach to flexible start and end points.
I do hold that in general it's better to achieve excellency for a small group than doing okay with a wider sample. There are thousands of games, each person may only want to own and repeat their top 5% of games, so a game that's not somebody's 5% belongs in nobodies collection.
That being said, I also hold that games should bring joy and more joy is better than less joy. A flexible approach allows something to be enjoyed by more people with different preferences and that's fundamentally a good thing.
Flexible conditions have been around for a long time, it actually seems to be more common in older games than newer ones. Going back to the old bookcase games, they're full of games which say things like "The winner is the first player to reach $X. Choose an X based on the game length you want." along with a few guidelines as to what that should be. I remember thinking that the old Merchant of Venus was informative with values like "4,000: Ship and engine upgrades will have time to pay off if you get them early" but seemed a bit full of itself at "10,000: For people who can't get enough of a good thing".
I'm in two minds as to whether we've lost something in moving away from that. There's certainly room for tighter design and much closer engines in more rigid systems. On the other hand flexibility lets an experience be more tailored to the people who are playing it.
I don't think that I have a firm position on this subject yet, I can see reasons to fix the start and end points of a game firmly at particular anchors. I can see reasons to give players the levers to move them around a little bit in response to their preferences.
I think the main thing I've learned through creating Wizard's Academy and seeing how it's received is that there are good and bad ways to do this.
It's good to have a very small number of fixed points for players to choose between. This makes it possible to playtest and tighten within those parameters and gain a lot of the advantages of fixed conditions. That's enough to let players make meaningful decisions.
It's bad to be unclear about where the levers are and what the point of them is. I muddied the waters by labeling the mid-game-start scenarios as "expert". A more straight forward description of what they were would've served better. Perhaps a half page early in the rulebook describing the various setup choices players can make and why they might want to choose different ones would've helped a lot.
Still, we live and learn. I'd be interested to hear any thoughts you might have on games with variable start and end conditions - or games that don't offer them that you wish did - or ones that do offer them that'd be better if they didn't. As I mentioned before, while I see a lot of the advantages and disadvantages inherent in different systems, I find myself undecided on the question of whether it's a generally desirable thing or not.