Having been cheerfully awarded the sleuth-o-matic rank of Barely Conscious (the murderer was was clear, the precise weapon+time wasn’t) I was wondering: is the mystery generator implemented in such a way that there’s enough “metagaming cues” in the clues to allow someone clever to get a good score on ALL generated mysteries?
For context: my partner correctly solved today’s mystery seven clues earlier than I did, and she explained that her answer was based on the fact that that 1. In literature certain weapons are predominantly associated with killers of particular genders, and 2. Killers will try to minimize their time spent at crime scenes to reduce the risk of detection, so if the murderer was found there, it’s probable that the murder was recent.
I was suitably impressed, but I’d be curious to know if human psychology and behavior patterns in literature are actually taken into account for the mysteries.
Well, the mystery generation is stochastic (where some of them will be more complex and difficult, while other ones are simpler) and the scoring is still under development (it usually very difficult to get a good score). Expect some changes in the near future to make the score more adaptative and fair.
> For context: my partner correctly solved today’s mystery seven clues earlier than I did, and she explained that her answer was based on the fact that that 1. In literature certain weapons are predominantly associated with killers of particular genders, and 2. Killers will try to minimize their time spent at crime scenes to reduce the risk of detection, so if the murderer was found there, it’s probable that the murder was recent.
> I was suitably impressed, but I’d be curious to know if human psychology and behavior patterns in literature are actually taken into account for the mysteries.
Assumption 1 is incorrect, all the weapons are randomly assigned so it was a lucky guess (however, it is an interesting one to make, so I will think about it to include it in the future..). Assumption 2 is .. actually correct. The mystery generation has a "minimization" procedure, where the actual input (the action from every agent) are reduced while keeping the murder mystery. This will indirectly, tend to make this assumption true.
Thanks for your feedback, it was really interesting.
Might be because I clicked through to either the sleuth-o-meter or the how to play and then hit back? I also filled in the example table on the how-to-play.
(Also, glad I'm not going crazy!)
Edit: I guess an expanded version of this should be in the README
[1] https://mcpshsf.com/ [2] https://github.com/xctf-io/chalgen