This interview is part two of a series from an interview with Jongwoo Kim, the creative director of Lucifer Within Us, a unique mystery game. To read our writeup on the game, click here. To read part one, click here.
Making Murder Mystery Mechanics
Fritz Wallace: What lead you to be interested in creating a mystery game?
Jongwoo Kim: For me, I was very frustrated at the mystery genre. To start, it’s a very vague genre. Like horror, there isn’t any single mechanic that defines it, like jumping for platformers, or shooting for shooters. There’s no unifying mechanic. Some games take the approach of a visual novel, being very text and menu based. Some are just adventure games, but with a mystery aesthetic.
So what I wanted to do with the project was to create mystery mechanics. To design tools we can give to the player that can be meaningful and applicable across multiple cases, and that the player can build skill around.
As a counter-example of what I’m talking about, take something like Frogwares Sherlock Holmes games. Sure, ultimately you accuse someone with their systems, but it’s bloated with a lot of mini-games. And if I’m gonna be blunt, they aren’t fun to play.
In one, I’m being hunted with a hunter with a rifle, and dodging bullets. It doesn’t have anything to do with the mysteries. And when you fail at these, it’s an awful feeling. “I’m Sherlock Holmes, why am I doing this?”
And so that’s where the timeline mechanic began. The initial prototype of the game was really focused on getting the timeline right. But it quickly led to other questions, such as “Okay, the timeline is good for temporal and positional parts of the mystery, but it’s not so good for relational information.” Such as, who likes who, who hates who. Who owns what, who saw this?
So at an earlier stage in development, we had a mechanic I believe we called the evidence map. And so any piece of evidence you found would show up on this evidence map, but it would also be linked to the other items on map. Among games that are actually out there right now, I think Murder Mystery Machine bases its entire gameplay on a similar mechanic.
But as this relates to cut mechanics, there was a version of Lucifer Within Us with a mechanic with the same idea. And it was compelling, but players found it to be almost too interesting.
Fritz Wallace: In what sense?
Jongwoo Kim: It wasn’t meant to be a fundamental aspect of the game. Actually, even the timeline wasn’t intended to be the fundamental aspect. All of these mechanics were intended to serve the tools for the player to understand what happened to the best of their ability.
But even though it was just informational, players tended to engage with the evidence map with the assumption that they had to use evidence map to progress in the case, even though it didn’t actually have the ability to do that.
There was an additional feature that tied the timeline and evidence map together called the hypothesis system, which allowed players to tie things to together, and map assertions about what they believed about happened.
So if you found certain things in a character’s testimony, and certain things about evidence, you could say things like “I think X did this with Y,” and it would test to see if that lined up with your current understanding of the case.
And all of these features were functional at a core level, but they became very clunky. The amount of UI needed to get them to do something simple was difficult. And secondly, it was difficult to make content for them at any reasonable pace.
If you’ve ever heard the term information haystack, that’s more or less the issue we ran into. The idea is that if you’re trying to find a needle, and there’s so much hay, the more hay you add in relation to the needle, it gets harder to find the needle.
Now obviously there’s a balance, and want some things to sift through for a compelling mystery, but at a certain point, if there’s just too much information, too much hay, it becomes impossible to invalidate certain possibilities.
So the inclusion of these features and creating content for them would in retrospect add too much to both the team’s haystack, and the player’s haystack. As a designer, though, I feel like the full mystery experience needs all these things.
Unfortunately, it’s hard to know what’s essential until everything is assembled. Also, to know if something is an information haystack problem, or a UI problem. Could we have just improved the UI in such a way that the players would know how to use it intuitively? Or was it a fundamental problem with the content or mechanics?
And if you don’t know, even just playtesting these mechanics is hard
Fritz Wallace: It seems like a big challenge to let the player do all of that, and trust them to figure it all out.
Jongwoo Kim: Right, so on that note, there were a lot of questions around that. “Why don’t we just script this part? Why don’t we just tell the player this part?” And the answer was in terms of the vision for the project, I wanted the player to be the detective, not the developer. I didn’t want twists people wouldn’t see coming to show up.
The design is to leave a case open enough that there’s nothing stopping a player from getting the information they need to reach the correct solution.
But that means you have to keep in mind a lot of possibilities. What if the player talks to this character first, or finds this evidence first?
If I look at the Steam reviews, some folks complain it’s too easy, some complain it’s too opaque, and I have mixed feelings on that. I do feel confident someone can solve every mystery without resorting to a walkthrough, or brute forcing everything.
You can see a suspect, the evidence, the alibi, and it’s more likely someone did something rather than not.
But something I noticed during playtesting was that players get attached to early theories. If they don’t like a certain character’s personality, or they seem suspicious for some reason, there’s a level of confirmation bias that motivates a player to stick with train of logic.
On the one hand, I didn’t want the game to be a sudoku puzzle: i.e., eliminate all other possibilities, it can only be this. But if the solution is too blatant, there isn’t much for the player to do. But some people would argue LWU is like that, and honestly, there’s more a degree of the obviousness of the solution than I first though when starting the project.
In early development, we had cases that were pointlessly complex relative to the average player’s desire to solve them. There was motivation loss at a certain point because there was so much to consider. And while I’m glad we streamlined them to a point, if I went back in time and talked to myself at the start of the project to say “By the way, at the end of the project, here’s the level of complexity on most of these cases” I’d be kind of disappointed.
Fritz Wallace: So there was some compromise on the levels of difficulty in the cases? How do you feel about that?
Jongwoo Kim: It is a bit of a bummer, but especially with the premise change, the project became more than just “Make a compelling player-driven mystery solving game.” We needed to execute to the best of our ability with the team we had, with the strengths and weaknesses present.
Fritz Wallace: Even with all that, how do you feel about the end state of the game?
Jongwoo Kim: I am quite proud of how the project came out, and the overall quality. I like Ada’s characterization. She’s a strong, confident protagonist in a murder mystery game, in a male dominated society. And I think that comes across in the game itself.
I think in terms of character design, I’m happy with how we balanced technology and religion. Given the number of characters and complexity of the cases, I think the cases are satisfying to solve. And I’m really pleased with how we got the smoothness, and level of detail with the timeline mechanic.
This concludes part 2 of the interview. Part 3 will go up tomorrow, and talk about the technical challenges involved in making a game where information is critical, and the player can rewind time whenever they want.