You should
absolutely design an alternative to Save Anywhere (really, it's Load Anytime that's the problem, but we'll get into that). While unrestricted saving may be an option you wish to provide for some players, you should consider your game as having unfulfilled potential for truly coherent player incentives and tightly designed challenges and consequences if that's the only way to play. I'm going to get on my soapbox here
RatTower, so don't think I'm accusing you personally of not understanding these things, I just want to make the argument in full.
My personal philosophy is that a keenly designed restricted save system is always superior to a Save Anywhere system. Being able to save and load your game is not "meta", nor does it exist apart from player incentives, strategies, and experience -- it's deeply embedded within a game's mechanics and structure. We're talking about rewinding time here -- try to imagine how you'd do things differently in every realm of your life if you could rewind time at will. The player will take a different approach to gameplay depending on what they stand to lose. If permadeath (in a longer game such as this), the player will feel increasing levels of tension as the game goes on and may feel it necessary to employ extremely conservative low-risk strategies to ensure they don't have to redo hours of the game. With savescumming it's the exact opposite with almost no stakes for failure, and players will have a strong incentive to employ high-risk strategies and simply "reroll" even marginally bad outcomes. Neither is particularly fun -- the former is tedious and limiting, and the latter is devoid of excitement and challenge. The purpose of a good save/load/death system should be to strike a balance between these two extremes.
With Save Anywhere, the developer is basically offloading the work of designing intelligent and fair checkpoints to the player. How should the player know when and how often to save in order to keep the game from becoming tedious, while also maintaining an acceptable level of tension and challenge? They have no knowledge of the obstacles in the level, nor perhaps of their own potential and ability to rise to meet certain challenges (the thing that makes e.g. Dark Souls' hardcore checkpoint system work). People often retort "just don't abuse the saves, then!", but this is like giving the player infinite medkits and requesting that they not ruin the game for themselves. You're asking players to
design their own experience, which is a dangerous thing to do when players are conditioned to optimize their performance in games, which is at direct odds with the sort of metacognition they have to exercise to strike this balance. Granted, in some cases this may be preferable -- in particular, when designing checkpoints for a highly open-ended game is simply too difficult, or when there are large disparities between your most hardcore and most casual players. But this shouldn't be the bar you set for yourself as a designer, and it is possible to create flexible options in and outside of the game that manage this. So let's talk about some solutions.
This discussion has been happening in certain corners of the Immersive Sim community, which obviously has some parallels even if you don't outright consider Monomyth as explicitly following those design principles. I have a few guidelines and recommendations to offer of my own, but I thought I'd highlight a video I found where someone stumbled onto the same key points:
- In games built around systems-driven interactions, much of the fun comes from forming mental strategies about how to overcome obstacles, executing them, and then adapting according to the situation as it actually unfolds
- In these games, there are typically many states between failure and success (as a basic example, enemies in Thief hearing a sound and becoming suspicious but not yet hostile, rather than murdering you instantaneously)
- With the ability to Save and Load at any time, you rarely feel the consequences for your mistakes, and can miss much of the actual gameplay loop from these intermediate failure states as well as much of the feeling of accomplishment when you do succeed
- The player could solve this problem for themselves by designing their own rules around manual saving or relying on autosave checkpoints (that's what I recommend for Prey), which seems natural for games built around player freedom
- However, the incentive to use saves strategically still remains, and you shouldn't expect players to behave in a manner that the game rules actively discourage
- As such, it would be nice to have some kind of "Adaptive" mode where you can only save at certain checkpoints, ramping up the tension and aligning the player's incentives with the game design goals of encouraging good planning, execution, and improvisation. These checkpoints can further incentivize exploration, and if they are are single-use, their overall effect can be multiplied
- The argument against this could be that these games involve a lot of player experimentation, as the player learns through trial and error how to manipulate the game systems and better predict the consequences of their actions, and manual saving helps the player do this. Hence, it may be appropriate to restrict this "Adaptive" mode to a higher difficulty setting, something you try after you've already played through the game once
As it turns out, this is not just a thought experiment. The
GMDX mod for Deus Ex already features restricted saving in its Hardcore difficulty mode, matching exactly the design specifications that the above video's creator independently hypothesized. There are autosaves between level transitions (this means that the more nonlinear and explorative civilian hub areas are more lenient with checkpoints), and single-use save points scattered within the levels, often as rewards for exploration and as a way to balance certain paths. Hardcore is unlocked after you complete the game once (or through a config file), ensuring that players understand the game systems before they have to deal with the consequences that restricted saving imposes on failure. The feature is a resounding success among the mod's fans, and is the preferred way to play the game for the masochistically inclined (such as myself). Traditional manual saving is still available, but the design problems stated earlier are more or less solved.
This is not the only option, of course. I'm quite fond of System Shock 2's (and to a lesser extent SS1's) resurrection chambers, which more or less act like Dark Souls' bonfires. Once you've activated one of these, you no longer lose temporal progress when you die, but you pay a resource penalty and lose geographical progress as you have to trek back to your objective, encountering newly spawned enemies on your way. I wish I could say the same about Ultima Underworld's Silver Sapling, as it obviously later became the resurrection idea in Shock, but it really just functions as a tool to let you run roughshod over the level, to a worse extent than savescumming (I'm not opposed to the concept, I just wish there had been some penalty like enemies healing up and maybe some marginal respawning when you resurrect). With contiguous player respawning, you can decouple death from saving and loading, and instead have the game saving all of the time, and/or whenever you quit the game.
That should be my last point, which is that restricted saving/loading doesn't mean you can't quit the game to go make a sandwich, because I'm so sick and tired of that rebuttal. SUSPEND SAVES, people. The "sandwich problem" has had a solution for ages. Even if you had restricted saving, you can have an additional suspend save system which allows you to suspend and resume the game at any time, but loading the suspend save also deletes it so you can't keep it as a personal checkpoint. Now go and enjoy your stupid BLT.