Eric, in the thread boredom, suggested that someone devise a ruleset for playing quirky mysteries a la Miss Marple or Jessica Fletcher. I'd prefer to expand it to include Columbo and Sherlock Holmes, and Tommy and Tuppence, and Nick and Nora, and Hercule Poirot, but let's see what we can devise.
In thinking about mysteries, my usual advice is that the writer, or the referee, must start with what actually happened, then from that determine what clues are immediately available to the observer and what else can be discovered by investigation, and making sure that there is enough information in all of this to reconstruct the crime. That's the way I handled it in Mystery of the Vorgo.
However, I'm aware that Seth Ben-Ezra, author of Legends of Alyria, has a very effective little game out there called Dirty Little Secrets which approaches the Noir Mystery genre from the other end: the players create characters and facts along the way, led by mechanics which ultimately determine who committed which crimes. I've not played it, more's the pity, but the concept is intriguing.
Our idea also suggests that the player character, a Verser from another universe, is neither perpetrator nor investigator nor victim, but an outside interested party who might or might not be suspect (as the investigator told Peter Davison's Doctor in The Black Rose, just because you're from another planet does not eliminate you from suspicion of murder--it just means we're a lot less likely to find your motive) but in some way has an interest.
Not so long ago I ran a murder mystery in Harry's Mary Piper Alpha setting; but then, Harry was a Junior Investigator in the Durnmist City Watch on duty at the time the murder was reported, and so he was roped into the story already. With that one, too, I had most of what happened worked out before he reached the scene of the crime, and thus it was easy for me to answer his question by thinking through what I knew had happened and occasionally rolling the dice to determine whether the outcomes made things easier or more difficult (e.g., did someone involved in the fight drop his weapon such that it is unclear who held it?).
I was once in a game of twenty questions in which the person answering the questions did not know what the object was but thought he would answer randomly yes or know and see what object fit the answers. In the end, there was no such object, and that's the risk of any mechanical method of running the mystery: if the resolution is mechanical, it might yield an impossible result (Peggy killed Sue at two o'clock in the afternoon, and was herself killed by Chuck an hour earlier). You have to have some way of making the outcome plausible.
Also, one of the tropes of such mysteries is often that the detective knows some seemingly irrelevant bit of information that solves the case. This trope is particularly challenging, because if the player is to solve the case he has to be made aware of the seemingly irrelevant information without being clued to the fact that that information is critical to the solution.
I will mention Ron Edwards' Moving Clue at this point, a mechanical solution to the problem that arises in mystery games when the players fail to ask the right question of the one character who knows the critical piece of information. By the moving clue, the critical pieces of information are known by one of the characters likely to be questioned, but the referee does not know which one--he only knows that one of them knows, and when the player character asks the right question it will be addressed to the right character.
That, anyway, is a start on the problem. I need to turn my attention to a lot of other threads tonight, and perhaps some others here will contribute to the project with ideas that will make such a scenario workable.
--M. J. Young