Let's put the basics down : An event is set at a point of IG time, and kills the PCs. The PCs then "go back in time" (no duplication of the PCs), and have to alter the course of time to avoid the event. If they fail, the loop triggers again.
"If you don't stop the mad bomber before he sets off the nuke, the city gets wiped out, you die in the explosion, and you wake up 24 hours before the nuke goes off. Again."
At any rate, you need a hard inescapable reset or the first loop in which they flee is the loop in which they have removed themselves from the entire premise of the game.
The conceit of the game demands that you place hard walls around the edges of the plot.
Pros: structure and tension, the audience-level "omnicient narrator" style awareness of the players contrasted with the narrow first-person awareness of the characters.
It introduces a kind of strategy: "what do I tell my next iteration?"
Oh, I remember where I'm getting such strong flashes of this! TNG did it!
"Cause and Effect": The Enterprise crashes and explodes, then everyone resets to 24 hours earlier with no memory of it.
But as the event repeats, particularly sensitive people start to get a bleedover effect from all the timelines on top of each other.
Eventually it comes down to this: what short word will Data send to himself subconsciously in the next time loop to stop the crash?
That kind of tactical choice could be very interesting.
HOWEVER, it comes with a massive con-side for an RPG context, which the Star Trek version overcame through powerful and talented directing and camerawork:
The players have to endure figuring the same stuff over and over again.
It's new to the characters each time, but the players are going to get so. freaking. tired. of it.
I'm not just speaking theoretically here; I've run games where that kind of "rediscovering what we already know" thing happens, and we gloss over it as fast as possible. Which pulls the teeth out of the tactical tension you're hoping for with this choice.
No, literally: if the clues change, it stops being solvable by investigation, and it's not a mystery. Not just "too complicated," but "the players will give it up as impossible."
We've* You are at least 50% of what we've come up with!
Do you think something like a 3-clue tree could be interesting, or something a bit more sandboxy with a relations diagram of the points of interest instead ?
3 clue tree : The players are expected to find 1-3 clues in their environment, which will lead them to a conclusion. this conclusion gets them to another enviroment, with again, 1-3 clues etc
Relation diagram : The environment is defined by objects (NPCs, things, places, events) that are linked to each others. The PCs follow a "string" that lead them inevitably towards the conclusion
Hum, actually none of these might fit the setting...
Since there is no pressure to succeed within a particular time frame, the only limit on your ability to experiment and test is the duration of each loop.
"Cause and Effect" started with the Enterprise main characters at the bridge, and the ship exploding before the credits, then came back after a commercial to a poker game 24 hours earlier.
The reception party is this game's version of the crew on the bridge just before the explosion.
However. Idea.
It's not quite the "realising we're all in it together" thing as you're envisioning it, but...
It's a predictable thing, which means you can safely build on it.
If you tie some crucial part of the plot to investigating the reason they remember the loop, you're basically guaranteed they'll find that part of the plot.
I'll be going for a couple hours soon, can we try to summarize things to know where to begin next brainstorming session (if you're still interested of course)
A cataclysm sends a group of PCs into a recurring time loop bounded at the far end by the cataclysm. Their only escape from the loop is to prevent the cataclysm.
Each PC can remember at least as much as their player can about all previous loops.
It's been suggested that the game begins with the group accomplishing an apparently tangential task --such as robbing a reception party-- just prior to the first iteration of the cataclysm.
The PCs will inevitably investigate the reason only they recall the loop, so that's a good hook to hang plot on.
A special thought needs to be put to : Bring the PCs together to die in first iteration; Get the PCs to understand they are stuck together ; Keep the PCs as a group as much as possible.
@Saffron I would be using some system that has means to measure time with some in-game timer. The first thing coming to my mind is how, in Psi*Run, the characters are chased by the opposition and need to move faster than the chasers. In your case, the chasing entity could be time.