💾 Archived View for yujiri.xyz › game-design › saving.gmi captured on 2024-08-31 at 11:37:16. Gemini links have been rewritten to link to archived content
⬅️ Previous capture (2024-07-08)
-=-=-=-=-=-=-
last edited 2024-07-03
Let's talk about saving philosophy. Should we let players save at any time, only out of combat, only inbetween levels, or have only autosaves?
First, let me illuminate the difference between saving against failure and saving against poweroff. If a player has to leave a game for any reason, or their computer powers off, we never want them to lose progress. Letting the player save at any time, even during a fight, and frequently autosaving is a good idea for this reason.
But that doesn't mean we want that in a game design sense. Saving arbitrarily often inside a fight can trivialize almost any challenge. If a player fails, they should generally go back to the beginning of the fight or the level or the checkpoint or whatever the game has. We don't want the way the game is normally played to be impacted by considerations about IRL disruption. It seems like we want two different kinds of saves: saves that you return to if you exit and reopen the game, and saves that you return to if you fail.
Of course, if you have a system like that, the player can exploit it by closing the game whenever they're about to fail, in order to use an exit-save as a fail-save. I'm not worried about that. It's okay if the player can choose to "cheat" at a singleplayer game.
With that in mind, how do we want fail-saves to work?
For most genres, there is such a concept as a "section", by which I mean a self-contained challenge. You start it with full status (health, stamina, mana, etc) and your status is restored after it. Sections can be individual battles (I call this the encounter-section model), or they can be longer, containing multiple battles (I call this the level-section model). These sections are what saving policy should revolve around, because challenges that consist of a series of isolated parts generally aren't interesting. If you've already done A, and the specific way you did A doesn't affect B, why would you want to have to redo A every time you fail at B?
This leads to my first principle: There should be a save point before and after each section.
The before saves should let the player get into their next attempt at the challenge quickly. For example, if there's a cutscene before a battle, it should be skippable, because no one wants to watch a cutscene they just watched. It should also not take 2 minutes to run from the save point to the challenge (looking at you Dark Souls). That's just wasting the player's time.
What about saving inside of sections? Halo is a game that does this: its sections are entire levels, which are very long, but it autosaves inside of them often. However, you can't revert more than one save back. This leads to a problem: you can get an autosave while in a near-unwinnable situation (low health, no ammo for any good guns, just lost the vehicle you're supposed to have) and then you're stuck there. If you can't somehow win from this situation, you have to restart the entire level.
Using manual saves instead of autosaves would not fix this problem, because players might save without realizing how bad the situation they're in is.
But if Halo just let you revert more than one save back, this problem would be fixed: if you realized your last save is in a near-unwinnable situation, you could go back to the save before it and have a chance again, without having to restart the entire level.
This leads to my second principle: If a game lets you save inside of sections, it must also let you revert more than one save back.
I have separate articles on other considerations related to saving: