The next game will be different in feel. It will also have a mystery though it'll be more personal and I think in this one I'd like to add mini-mysteries too. Sort of like...hmmm...say in Buffy where she fought a different monster each week but each season there was a big bad. Something like that. So it'll have an episodic feel with an overarching narrative.
The problem I see with a lot of sandboxes and I think it's what you are referring to and what burns people is that a lot of them resort to repeating events. IE "go to this place, click on NPC, have sex with them and it's the same exact scene as next time. Add +1 to corruption." I suspect devs do it so they have content to fill the sandbox that doesn't take a lot of work. I told myself in my game I wouldn't have very many, if any, repeating scenes because I wanted my game to feel immersive. A side effect is that a lot of places in the game you go to and see something like "so-and-so isn't here right now." because I haven't added new content for them.
That design for the next game sounds pretty interesting!
Repeating events are IMHO an attempt to step away from worst cases I mentioned, in which the Sandbox adds nothing, but arguably make the game overall worse because of bad game design.
Events/things in a game can be fun intrinsically (for example reading a new scene, or shooting an enemy in an FPS), or fun because of what they mean in context - for example, clicking on an enemy to shoot them in an XCOM-like isn't peak fun, but you enjoy it because you have an idea - A PLAN - flank there, eliminate that one, great... oh, that wasn't supposed to happen, but I could counter by... great! - making those decisions and seeing how/if they accomplish your goals is what's satisfying, as is seeing progress towards your meta-goal. Both of those examples are skill/mastery, but obviously they don't need to be either - placing blocks in a Minecraft-like to build some large project you imagined is another example of "context-fun". And obviously, the fun of most things in games is a composite of both, and what is fun is highly subjective - your very intrinsically fun thing might be boring or frustrating for me, and the context that makes another thing fun for me might be pointless to you.
But as subjective as it is, I'm pretty certain that the amount of people for who viewing the same erotic scene over and over because you have to to raise a stat to advance a plotline, with no real decisionmaking except "I want x, therefor I got to click this 15 more times", is either kind of fun is extremely small.
So events in a sandbox should be either be fun intrinsically (as in, be interesting/fun each time you do them, for example, as a minigame; or just by avoid repeat events, like you do) or fun because they serve a fun goal of the players - which usually means that event the is the result of some decision making process - either frequent and fun (XCOM for example), or long-term and deeply personal (Minecraft), and because it's hard to do "build whatever you want to build" fantasy fulfillment in an eroge sandbox, that usually means that there should be many decisions regarding what events to do, with mutilple good decisions depending on what your goal is as a player. (Ideally, they should be fun in the moment, be the result of an decision between multiple interesting options, and in service of a long-term self-set goal, but that's hard to accomplish.)
UA is a good example - first off, combat, sex and dialogues are minigames, not static text, and thus can be fun even on repeat. They're also influenced by the stats of the characters in the metagame, deeply connecting how the minigames work with metagame. Trainings are just short 1-page passages awarding experiences, but like the previous events, they result in experience changing the stats of the characters. And there's limited time and competition by the NPCs, which makes decisions of what stats to increase an interesting decision. Then there's a potential meta-game, akin to Minecrafts "what do you want to build" - NPCs act according to a logic that takes into account their relationships and drives, so you can set yourself an overarching goal (like "everyone gets along with everyone", or "I want everyone to serve me"), the fulfillment of which you can then strategize for.
The problem with all of this is of course that it is much easier said than done. A minigame that is fun x times can become just tedious if it needs to be done 2x times; what you think is motivating context might not be for others, and the x might vastly differ between persons - I chose both UA and SC as positive examples: In both cases, there are some people who feel that the elements are annoying or tedious - the art is to find something that feels right, to an at least large enough audience and yourself.
An art which also requires iteration and experience in an additional skillset to the usual skillsets of writing and rendering, game mechanics design.
Recognizing when a certain mechanic works or not is hard, and finding the correct touches to make it work is even harder - there's a number of cases where the fix to "too slow" was moving the camera down a bit and some optical effects.
(And I don't mean to say that game mechanics is harder than rendering emotional/attractive scenes or writing interesting characters/scenes - those are also damn hard. It's just another, separate set of skills, which shouldn't be taken for granted, same as writing shouldn't be taken for granted (*cough* Bioware Management *cough*))
...I'm honestly not sure if my rambling is of *any* use, but I guess there's no harm in publishing it.
