> However, it's important to note that every clue you view will have an impact on your overall score. Exercise caution and strategic thinking as you navigate through the evidence and interrogate the suspects to piece together the ultimate solution.
That sounds great! But what actually happened was I clicked through clue after clue about person X being in room Y at time Z until I eventually got told what the murder weapon wasn't three times in a row. I didn't feel like I could exercise any caution or strategic thinking; it was just "Nope, this isn't enough information" over and over until "Yes, this is now enough information".
At the point where I thought I had enough information, I realised I hadn't been consistent in how exactly I was using the notebook: does a tick in the "9:15" box means "They entered at 9:15" or "They left at 9:15"? The boxes really need to correspond to the times they're in the rooms, rather than the instants they move between (so either labelled with a range: 9:00-9:15, 9:15-9:30, or move the labels so they sit in the gaps in between).
Also the "How to Play" link isn't openable in a new tab (because it's handled by javascript), and if you open it normally and go back to the main page you lose all your progress
> That sounds great! But what actually happened was I clicked through clue after clue about person X being in room Y at time Z until I eventually got told what the murder weapon wasn't three times in a row. I didn't feel like I could exercise any caution or strategic thinking; it was just "Nope, this isn't enough information" over and over until "Yes, this is now enough information".
I need to think about this, but for sure the clues will be refined to encourage this. Right now, there are only a couple of types of clues and they don't feel very organic.
> At the point where I thought I had enough information, I realised I hadn't been consistent in how exactly I was using the notebook: does a tick in the "9:15" box means "They entered at 9:15" or "They left at 9:15"? The boxes really need to correspond to the times they're in the rooms, rather than the instants they move between (so either labelled with a range: 9:00-9:15, 9:15-9:30, or move the labels so they sit in the gaps in between).
I need to think about this. English is not my first language and it could be possible that some sentence are misleading to native speakers.
> Also the "How to Play" link isn't openable in a new tab (because it's handled by javascript), and if you open it normally and go back to the main page you lose all your progress
Related to the above, one thing that wasn’t clear to me was whether the first three clues should be sufficient to identify the killer. From a game play perspective, I don’t think that matters because the theoretical high score (assuming you solve it without randomly guessing) could just be lower if you need extra clues for that problem due to the first three not being sufficient. However I couldn’t tell if I was being dumb or I didn’t have enough information.
You are right, it is very very unlikely that the first three clues are enough to identify the killer in any reasonable scenario. The ranking is still very preliminary and it will be improved very soon.
> I need to think about this, but for sure the clues will be refined to encourage this. Right now, there are only a couple of types of clues and they don't feel very organic.
Yes, with some more varied clues I can see how it could be very fun. Maybe an extra type of clue could be associating suspects with potential murder weapons (e.g. "Alice said she never touched the rope" or "Forensic testing found traces of Bob's DNA on the poison bottle").
Or maybe another type could be like: "Alice heard someone in the bedroom when she was in the kitchen at 09:30". In the right circumstances that could give you _just_ enough information to deduce something.
Dave said: "I saw Bob when I arrived to the kitchen at 9:30"
also
Dave said: "I saw nobody when I was leaving the bedroom at 9:15"
How could he leave the bedroom at 9:15, but arrive to the kitchen at 9:30? Where he was from 9:15 to 9:30? I assume that each transition takes 15 minutes, so that he can't step directly into the bathroom or the living room from the bedroom.
I agree a greater variety of clue types would make it more interesting.
And then maybe each clue you reveal increases the chances of some failure outcome (e.g. murderer escapes, kills again, etc)? That would definitely prompt more caution and strategic thinking before revealing the next clue.
If you like solving mysteries like this, I suggest getting a copy of Cain’s Jawbone. It consists of 100 pages out of order and multiple murders. To win the game, you must not only solve each murder but correctly order the pages.
In the same vein, I highly, highly recommend playing Return of the Obra Dinn. Slowly piecing together the fates of all the people on the ship is deeply satisfying, sorta like a sudoku puzzle.
> Slowly piecing together the fates of all the people on the ship is deeply satisfying, sorta like a sudoku puzzle.
Return of the Obra Dinn relies completely on the fact that you can make guesses and the game will confirm for your benefit whether or not those guesses were correct. The information you're "piecing together" is not actually present in the game - you have to guess it, and sometimes the game will tell you that that's what the game developers were thinking too.
That's just not true. Most of what you do is guessing. There is nothing in the game, for example, that will tell you who the second mate's steward is. You assume that it's the guy who you frequently see standing near the second mate.
There is also no indication in the game of who killed John Naples.
> You assume that it's the guy who you frequently see standing near the second mate.
Men wearing steward uniforms frequently hanging around officers do tend to be stewards.
> There is also no indication in the game of who killed John Naples.
Probably the Swedish-speaking man being hauled away from the bloody sword to the lazarette while the captain shouts that he was “twenty years my steward.”
I guess it's implied that people only travel between rooms on 15-minute boundaries, or something? And maybe it's implied that the murderer always lies and the non-murderers don't? I dunno. The clues are really confusing without understanding the meta-rules behind them, which don't seem to be written down.
> I guess it's implied that people only travel between rooms on 15-minute boundaries, or something?
Yes
> And maybe it's implied that the murderer always lies and the non-murderers don't?
Right now, clues are always true, but the murderer will omit saying something very incriminating. But since the website is a prototype, this will most likely change in the near future (e.g. murders can lie, but the other suspects will always tell the truth).
> I dunno. The clues are really confusing without understanding the meta-rules behind them, which don't seem to be written down.
I think you are right, the website needs better (and more) explanation, as well as an expanded tutorial.
(I still think it's really, really cool, by the way; please don't get discouraged by the criticism. With some UX / intuitability improvements it'll be awesome.)
Is this just randomized? It makes no logical sense and seems unnecessarily confusing:
> The killer took the murder weapon from one of these rooms:
> The gun from the kitchen
> The knife from the living room
> The poison from the bedroom
> The rope from the bathroom
Poison (of the pharmaceutical type) is usually found in bathrooms. Knives are found in kitchens, never near upholstery. Guns are usually found in bedrooms. Living rooms usually have something ropelike. If this is random, it's a bit fantastical, and all of it is suspicious in itself. Maybe limit randomization to shuffling 2 of the weapons around, because some red herrings are expected.
Poison is kind of a crappy weapon for this game if you don't account for it being able to kill without the suspect and victim in the same place at the same time, only them both being in the same place at some point in time.
Also:
There isn't much reason to use the "?" flag on the chart as it stands. Having a witness give vague, "don't remember exactly" or excessively-timeboxed clues ("I was in the bathroom from 9:00 to 12:00, leaving only to drop off some vinyl covers to the living room so I wouldn't shart on them, but don't remember when that was") would add some challenge. The murder solves itself if you just follow the clues, plot the whereabouts of each suspect and look for overlap.
- poison in the bedroom is someone’s bedside medicine that would be fatal to someone that wasn’t prescribed it.
- some people have ceremonial/sentimental knives on display, above a mantle for example.
- a gun could easily be stored in a drawer near a door, like a police officer taking off their gun and badge when they get home, or a farmer keeping a varmint pistol for east access when going out to a field.
- the rope could have been in a utility closet in the bathroom, like where laundry machines are sometimes installed.
Poison could be put on a toothbrush (bathroom), cologne bottle (bedroom), tv remote (living room), or a plate (kitchen) and kill without having to be present.
Part of the fantasy is imagining why those things would be where they are; it doesn’t have to be realistic.
Really fun game. From a visibility standpoint, I would recommend improving the visibility of the notebook. Once you're deep into it and a lot of boxes are filled, it becomes harder to keep track of checks and crosses. Native dark mode would be nice to have as well.
Another nice to have feature, quick clearing the notebook would be nice. I didn't need it this time and I guess refreshing the page does it. but it also closes the clues.
Gameplay wise, I would like it if you were more explicit about how people travel from room to room and how long they spend in each room. It's still solvable but I can definitely see how people would struggle to understand.
All this being said, really great job on the site and the mystery. I absolutely love it and look forward to playing it more.
Good fun. I think, though, that precision in language might be a challenge. Some previous comments I concur. Over and above those, I was very strict not to infer anything outside the minimum of what was said. For example, “I was in bedroom from 10:00 to 10:15” does not imply that “I was not in the bedroom before or after that time”. Or, “I didn’t see anyone when I arrived” only means I saw no one in the destination room, not that there wasn’t someone in the kitchen (that I must have walked through) or the source room. Illogical that the murder could have happened up to 11:15, at exactly the same time that the police arrive—unless the victim phoned it in. These rules left ambiguity.
Thanks for the feedback! I agree that the rules and explanations needs to be improved. I don't like the ambiguity and I think the deductions you did are indeed, correct. I should focus on improving the language in general, since English is not my native tongue.
Do people always spend increments of 15 minutes in a room? Can they run through a room to get to another room? That is, if the map is A-B-C, can they be in A from 9:30-9:45, run through B, then spend 9:45-10:00 in C?
Dave said: "I saw nobody when I was leaving the living room at 10:30"
Does this mean no one was in the living room when he left it? Or does it mean no one was in the room he entered when he entered it? Or both?
What is to stop a person from being in multiple rooms within a 15 minute period?
If a person exits a room at the same time someone else enters it, does either see the other?
I really like these types of puzzles. Here's some feedback:
I think I either missed a clue (i triple checked) or it came down to a guess on the time. I hope it's the former and not the later.
I also had a couple of waste clues. Once I had the murderer, I had a few possibilities of the murder weapon but then clues narrowed that way down. Out of 4 weapon possibilities, having 3 clues say what the weapon wasn't is 1-2 too many, especially that late in the game.
The graph was a little confusing. It's not clear that the relation of the rooms matters from the game/clues, but the graph makes me think it is. (this was written before I saw the how to play link, see below)
I got a constable score. I have no idea if that's a good score or a poor score, probably the later since I went through all the clues.
The how to play link should be at the top in the first set of directions on how to play. I think another commenter said that it takes at least 15 minutes to change rooms. IE in the example if Alice is in the kitchen at 10:15, she couldn't have been in the bathroom at 10:00. She could only be in either the living room or bedroom at 10:00 to be able to be in the bathroom at 10:15. I didn't even make that leap until I had read that comment. Call that out specifically!
> Frida: I saw no one in $ROOM when I arrived at 9:30
> Frida: I saw no one in $ROOM when I left at 9:45
to me, the third clue felt wasted. I had already marked off that nobody was there during that time. but does this also mean that nobody arrived at 9:45, and that nobody left at 9:30? in other words, do these clues tell me that nobody was in $ROOM between 9:15 and 10:00?
anyway - pretty interesting game. I think it would be a lot better if the user could ask specific questions. "Where was BOB at 10:30?", and then receive the clue, "BOB was in $ROOM from 10:00 to 11:00" (because if you're investigating a murder, it would be unthinkable not to ask "for how long" as a followup... it should be a freebie). or maybe just "BOB was in $ROOM at 10:30 and left at 11:00" if you want to be trickier. this would give the player some agency and hopefully allow the chance to unlock a large amount of information with a well-chosen question.
also... does it really take 3 clues to determine the murder weapon? maybe that's a legacy of the older Clue game (which I haven't played), but I can't imagine showing up and having no idea which of (poison, knife, gun, or rope) was used to kill a person. by default, it should be obvious IMO (maybe there could be confounding issues, like they have stab wounds but actually died of poison and were stabbed afterward as misdirection), but I guess that takes away a big part of figuring out who did it by placing them in the room with the murder weapon. it would take some work to rethink that part, I guess.
Feels like a rough, very early prototype of a cool idea. I think it would be much more interesting if each clue was a question, like, “Frida, which room were you in at 10:30?” etc. Then I would actually need to think about how to use as few clues as possible. Now I don’t think there’s any way to get the answer right except by guessing until all the clues eliminating murder weapons.
The UI for filling in who is in which room isn’t great, it would make more sense to say which room each person is in than who is in each room - one person can just be in one room after all.
Thanks a lot for the feedback. The website is actually a prototype, with rules subject to change.
> I think it would be much more interesting if each clue was a question, like, “Frida, which room were you in at 10:30?” etc. Then I would actually need to think about how to use as few clues as possible. Now I don’t think there’s any way to get the answer right except by guessing until all the clues eliminating murder weapons.
This sounds a little bit like Clue, right?
> The UI for filling in who is in which room isn’t great, it would make more sense to say which room each person is in than who is in each room - one person can just be in one room after all.
This UI is similar to the one murdle, which supposedly allow to make deduce (e.g. if someone is not in room A or B, then it must be in room C). I need to think on how to improve it..
> This UI is similar to the one murdle, which supposedly allow to make deduce (e.g. if someone is not in room A or B, then it must be in room C). I need to think on how to improve it.
You can simply flip the hierarchy - the top-level headline should be the person, and the subheading should be the room. That would make exactly the example you give easier, as all the information is collected in one place. And you don’t risk filling in that the person is in two rooms at the same time.
I've long felt that the mystery game genre is cool in theory but doesn't work in practice on computers. Most games end up being menu clickers. Or you have to add puzzles. Which is fun, but a puzzle game isn't the same as a mystery game.
Generative AI changes things. It's possible to make a more open-ended game where you can act like a real investigator. Draw up lists of suspects, order forensic tests, call experts, interview witnesses, whatever you can think of.
You won't have to wait long, that space is red hot right now. People are rushing to commercialize LLM powered NPC dialoge. nVidia's ACE might cut everyone's lunch though we will see.
There is one tech demo which has you piecing together what happened at an accident by talking with the NPCs.
There's also a Skyrim mod that replaces all the NPCs with LLM actors.
I think the turning point for that technology will be locally run LLMs that can be very finely tuned and trained on your world data.
I think it will be a long long time before we see an LLM generating a genuinely novel, interesting and fun to play mystery game.
Good mysteries involve providing just enough hints so that the player can make the fun leaps of logic and experience "Aha!" moments. People don't want to play the job of a real detective (collect endless witness statements, construct extremely detailed timelines) they want to be the hero in a detective story, making clever deductions.
LLMs might be able to generate characters and dialogue but they're going to struggle to come up with completely new ideas and mystery mechanics like Her Story
> People don't want to play the job of a real detective (collect endless witness statements, construct extremely detailed timelines)
You underestimate how many people would be really into what you described. Lots of wildly popular games are work in a fun-looking package. Factorio is just programming. Football Manager, and every other sports manager game for that matter, is spreadsheet after spreadsheet with some cool graphics.
That is what the tech demo does, each actor has a prompt that includes what that person knows about the situation and how they should dicuss it but the story itself was designed by a human. The prompts for the actors were written to make the discussions interesting and useful for deducing what happened. So perhaps Actor A saw a mysterious man, Actor B heard the noises and has a better idea of the timelime, and a different detective has already examined the physical clues so you can discuss them with that actor.
It's really interesting but it's also of course really easy to break out of sensibility, and hard to properly gamify. Like you said people don't actually want the tedious part of detective work, they want the fun parts that make them feel like Sherlock.
The game Shadows of Doubt is a great example of how the information flow of detective work is really difficult to balance, I think they did a great job but if you run out of really strong clues, clues that link things together, the game becomes a brute force exercise in talking to everyone and asking the same questions.
Mhm interesting, didn't know that it is seemingly just a German thing, only with some editions that are translated. The stories are often very hardcore, but nonetheless we played these as teens after School together or even in school. It's an analogue "card game", one side of the card is known to all who participate which contains the story and the other side is only known to one person which has the solution. The participants that don't know the solution only can ask yes or no questions to the one who knows the solution. I don't own one personally, and the only things I found on the internet are some images of these cards, but in german. You could try to translate one with Google Translate. If you search for black stories on the official store you can see that there are 135 different versions of it (some are called different colors/names, e.g. pink stories, which are more harmless stories for children or "pferde stories" which are about horses)
> After the meal, the queen died.
If the guests had been more boring, she would still be alive.
Translation (solution):
> A beauty queen (like e.g. Miss America) was invited to dinner every night after her election. To maintain her weight, she stuck her finger down her throat after each meal. At a reception she was involved in many interesting conversations and forgot to retire to the toilet after the main course. Her overfull stomach ruptured and she died as a result.
This is really cool! I had some of the same points of confusion as others did, e.g. what do times mean, does “I was in $place from $time to $time” imply being somewhere else before and after, etc. I think being more explicit with what you _can_ and _can’t_ conclude would go a long way.
I also echo the suggestion to add more clue formats.
But I think it’s really nice and shows a lot of promise. Good stuff!
I don’t see it explicitly mentioned in the instructions, but I’m assuming that the room web/map means that someone can only get to certain rooms by going through other rooms?
For example, in today’s challenge there’s only one way to get from the living room to the bathroom by going through the kitchen.
Thanks for the reply. I originally started filling out the notebook without assuming this since I was getting Clue/Cluedo vibes where you don’t need to enter one room to get to another technically. But this is cool.
Thanks a lot. I'm planing to add the exact description to each rank very soon. The "constable" is the lowest rank, motivated by this definition:
> a British police officer of the lowest rank
However, since English is not my native language, I don't know how easy to understand this is.
Regarding the data structures, the engine generates a complete program represents the mystery. Then, it uses fuzz testing to find a solution (and minimizes the number of clues). The engine to generate the mysteries will be open-source soon!
I'm not sure if anyone is using this this way, but this seems like a great way to test LLM reasoning performance as it is created anew each day and there is no chance it is included in the training data.
> However, it's important to note that every clue you view will have an impact on your overall score. Exercise caution and strategic thinking as you navigate through the evidence and interrogate the suspects to piece together the ultimate solution.
That sounds great! But what actually happened was I clicked through clue after clue about person X being in room Y at time Z until I eventually got told what the murder weapon wasn't three times in a row. I didn't feel like I could exercise any caution or strategic thinking; it was just "Nope, this isn't enough information" over and over until "Yes, this is now enough information".
At the point where I thought I had enough information, I realised I hadn't been consistent in how exactly I was using the notebook: does a tick in the "9:15" box means "They entered at 9:15" or "They left at 9:15"? The boxes really need to correspond to the times they're in the rooms, rather than the instants they move between (so either labelled with a range: 9:00-9:15, 9:15-9:30, or move the labels so they sit in the gaps in between).
Also the "How to Play" link isn't openable in a new tab (because it's handled by javascript), and if you open it normally and go back to the main page you lose all your progress