I ran this one game years ago, that was quite popular, even if it lasted just a handful of sessions. The reason for the popularity was, judging by what my players told me, mystery solving. The game was set in a kind of 18th-19th century fantasy world and I ran it with straight 3rd editon D&D with some custom classes.
Now, the truth about that game is there wasn't really any mystery solving. It was a display of the world I had made, a Setting-focused Right to Dream on my part, with some semblance of a plot conducted through Illusionism. I revealed clues when I was ready to reveal them, when I felt it was appropriate. I jealously guarded my dramatic moments, while at the same time pumping out loads of setting Colour that didn't really matter to any sort of mystery, except to contribute to the celebration of the setting, the illusion of "there". It's not something I would do again, technique-wise. These days I personally find it exhausting and it does not satisfy my play.
So I wanted to revisit this game in a more effective manner. First off, I went to re-define and re-watch the source material (chiefly: Sleepy Hollow, Brothers Grimm, Brotherhood of the Wolf, marginally: Vidocq, From Hell etc.) - I needed my insights. Then, I tried to figure out what the players actually did in this game, what was the objective of play. "Solve mysteries" the players said.
So ok, this is a game where you solve mysteries. That's awesome. This brings about two corollaries: 1. the GM does not solve them for you and 2. it's not about morality, or drama, or theme. So the procedure of play crystalizes: The GM creates a mystery, the players try to do their darn best to solve it.
It was clear from the first that the D&D chassis from the first game had to go. It does in no way support that sort of play, even though they share the same agenda. So I went back to my source material and built a game around that. It's good (at least I hope it is), mechanically sound. The players must carefully manage their resources to maximize their clue-gathering. It's like D&D but with monster mysteries instead of monster hitpoints. That's awesome.
That is, until I noticed something pretty interesting that I hadn't realized before. You've got your clues, right? Bits of fur near the body, a smell of sulphur in the room. The problems is, you can't draw any kind of conclusion from those clues, because you've got nothing to compare them against. So we've got a monster that's furry and smells of sulphur. Great. That means squat all. Because it isn't real. If we're playing a real-world mystery and there's a cigarette at the crime scene, great, we're looking for a smoker. Fur at the crime scene in a fantasy game...huh. Do werewolves exist in this world? Do hairy devils? Do talking bears? I can't make a case here.
And if you go check it up in an in-game encyclopedia that just means it's the GM revealing the clues to you again. It's not you finding them, or drawing conclusions. You're being led, bit by bit.
You would need the players to know, very explicitly, what sorts of monsters exist in this world. So as they keep on finding more and more clues, they can narrow down the suspects. This would require an amount of pre-play and setting familiarity that I'm really not willing to accept.
The first part, trying to get as many clues as possible, that's great. But once you've got your clues you need an existing body of knowledge to compare them to. If the players don't know the sgrouglulax is a hairy monster that exists in this world, then they aren't figuring it out, they are being told.
A possible alternative is that you simply gather abstract "clue points" that you then pay to reveal the mystery. But that doesn't really feel like solving a mystery at all, it feels like gathering points. I've got a third option, but I need to think hard about it.
Man, mystery games are hard.