Detective games

Thanks, that sounds really cool.

No problem. But like I said, it is not am AAA game. The writing/dialogue and voice acting is very serviceable, but nothing special. Focus Home Interactive published it, and it’s very in line with the quality level of the other games of theirs I’ve played.

Make It Good is a 2009 interactive fiction game that the reviews were implying was the modern incarnation of Deadline. Complex characters with state that you need to manipulate by showing them evidence and telling things, and investigation that requires actual understanding of what’s going on with the case so that you know to use the right non-obvious verbs. Since I thought that Deadline had done some of this stuff better than any other game before or after, this sounded intriguing.

The reality is quite different from what those reviews said.

You’re a drunk detective straight out of a Raymond Chandler book, who hasn’t managed to solve a case in a year. You’ve been given one last chance; somebody is dead, you need to make an arrest before sunset or you’re fired. So there’s a somewhat Deadline-like setup for artificial urgency too.

Now, the game has a core conceit that all the reviews left unmentioned. I can see why, since it’s technically a spoiler. But it’s a spoiler for roughly the first 30 minutes of the game, and I find it totally impossible to talk about the game in the context of this thread without it.

It becomes apparent very quickly that the player character is the murderer, and that impression gets reinforced regularly. Your fingerprints are on the murder weapon. Your footprints are in the flowerbed at the escape route. When the vicar greets you, they say they’re happy to meet you “again”. You’re in the possession of a note that the stiff wrote to the person who was blackmailing them. A button that’s fallen off your coat is under the victim’s bed. And more.

Your job is not going to be to find the murderer. It’s going to be to frame somebody else for the murder. Which turns out to be a fairly boring mechanical exercise, where your sidekick will always helpfully mention what category of evidence is missing to make an arrest.

The first time you make an arrest stick, it turns out there’s a complication. Your evidence won’t be enough, and you really need to get somebody to confess to a murder they didn’t commit. That’s a slightly more complicated process, but doesn’t really feel like it requires any kind of deep understanding of the case or the characters. (I had to look at a walkthrough, but it was only since I’d given up on forging a certain kind of evidence, after once being told off by the game and the second time getting totally the wrong reaction from a character. Turns out the second attempt had actually worked correctly, but I just needed to wait for a while.)

I’ve at times thought that a game about framing somebody would be really interesting. Turns out I was probably wrong about that.

Make It Good was a harsh disappointment. As an interactive fiction game it’s technically great; there’s a seemingly endless supply of custom responses to everything. (Basically the first things I did in the game was throw a bottle of whiskey at a dog that was annoying my character, and then try to lick the whiskey off the pavement, and the game handled both of those commands). There was just a little bit of parser frustration, mostly due to the conversations missing some words that I felt were really obvious things to ask about.

But as a spiritual successor to Deadline, or even as an exampe of the genre, it just doesn’t work. The feeling of fitting together the puzzle pieces in order to solve the mystery just isn’t there.

An Act of Murder is another modern interactive fiction detective game that’s clearly been inspired by Deadline (all the way down to having Duffy as a character, though by now he’s been promoted).

A man is dead, in a remote mansion that nobody except the five guests at the dinner party could have entered. You conduct an investigation via a text parser, interrogate people by asking them questions about things you’ve found, and ultimately present a set of evidence to make an arrest that’ll stick.

The games is mostly pretty crude compared to the interactive fiction games I’ve written about earlier in the thread. There are no interesting investigative verbs, it’s basically just going to every room and searching every object. The conversations are just you the characters asking about things you’ve discovered (plus a couple of generic ones, like asking about their alibi).

And proving that you’ve solved the case is really just a case of telling Duffy about the evidence you’ve found. Once you’ve mentioned all bits of evidence that are germane to some aspect of the case, Duffy will explain their significance. As far as I can tell you could just tell Duffy about everything in your notebook, and he’ll solve the case for you. There is no need for the player to actually understand the case.

The gimmick is that the mystery is generated procedurally. The murderer, the motive, the murder weapon, the timing of the murder, the schedules of the characters, etc all change on every run. The first time around the illusion was maintained quite well. The murderer was maybe a bit obvious, but it was still quite satisfying since the game did a lot more than typical with detailed timelines of who-was-where-when-with-whom vs. the timing of the murder. And I just love that shit in detective fiction.

The second time around, the illusion collapsed completely. The details changed, but the process didn’t. Exactly the same two bits of evidence (one derived via absurd adventure game logic) were used to establish the time of murder. It became obvious that you’d every time exclude two suspects via them having mutual alibis for the time of the murder, one suspect via an inability to commit the murder (e.g. lacking a piece of knowledge needed to execute the crime), and one suspect due to not having a motive. And then whoever is remaining can be arrested, and will confess. Doesn’t even matter whether you’ve figured out if they have a motive.

(I might be wrong on the exact procgen structure, but even though I’m pretty sure that further runs would take about 10 minutes each, I can’t be bothered to try it again).

This won 2nd place in the IF comp in 2007, so my expectations were really calibrated much higher than the reality. I think it must have been purely based on some kind of “it’s a miracle this works at all” recognition of how hard this would have been to code, rather than anything to do with the merits of the game play.

I’ve played this one and I suspect you’re mostly right. I think a lot of games place high in the IF Comp when they have some impressive bit of coding or a great Floyd-like character, rather than being great games on their own.

It’s pretty fun the first time through, though.

I decided to give modern interactive fiction one more chance with Color the Truth. Coincidentally this also got 2nd place in the IF comp, but in 2016.

Once again somebody is dead, and you’re the detective tasked with finding the murderer. The game has two gimmicks. One is that when you get somebody’s statement on when they last saw the victim, the viewpoint switches to that character and that time. You then play a flashback sequence through as them, to find out what happened.

The other is that the game explicitly notes when you received some knowledge of interest, and gives that bit of knowledge a name. You can then ask characters about it or, more interestingly, combine two contradictory pieces of knowledge to a new fact.

This new fact proves that one of the people providing these facts must have lied, and you can confront them about it. This makes them update their statement, and you get to play through the flashback sequence again to find out more.

The linking together of facts is exactly what the evidence-providing part of An Act of Murder was missing; i.e. the player should not just know that the gun went missing between 1pm and 3pm, and that Bob arrived at the mansion at 5pm. They should be required to point out that the combination of these facts is significant (Bob might not have an alibi for the murder, but they must be innocent since they could not have acquired the murder weapon).

But everything about Color the Truth conspires to make this a totally ineffective game mechanic:

  • The only way you can combine two facts is to point out a contradiction, which means that somebody has lied. That’s just not very expressive; it hardly even counts as deduction.
  • There are not very many facts in the game. Maybe around a dozen? You need to combine them maybe 5 times or so, and all but one of the combinations are totally obvious.
  • Color the Truth is not subtle at all about what’s important. When a bit of dialogue produces a new fact, it’ll tell you about it very loudly. There’s very little need to interpret the text.

There is no collecting of physical evidence in the game. The actual parser-driven gameplay during the flashbacks is totally on rails, with the game telling you what you’re supposed to do and chiding you if you try to do something else. Playing a game doesn’t get more boring than that.

Combine the lack of actual decisions outside of the combining of facts, the game automatically flagging the facts when they come up in dialogue, and the tiny number of combinations that can be made, and you have an experience that I’m hard-pressed to even call interactive. The plot? The game is so short and the mystery so simple that I was stunned when the game ended. Like, that was it?

Color the Truth is just a baffling game. It’s like somebody took all the parts of parser-based interactive fiction that make it a great fit for the genre, and threw all that away. And then they replaced it with what I think might be the second simplest fact-combination system I’ve seen. (It’s not quite as bad as The Trace). It’s like they totally ignored what the genre looks like outside of IF.

The whole on-rails flashback system has exactly the same problem of playing to the weaknesses rather than the strengths of the technology. If I want a non-interactive bit of exposition, I’ll rather watch a cut-scene. It seems obvious that the only way you make the flashbacks interesting is by making them the method by which you highlight the contradiction. Something IF would be uniquely suited for.

I think I’ll give up on modern IF detective games for now. These three seemed to be the most highly recommended ones, but none of them really worked for me.

Yep, I agree. The first run through the investigation was good, just a bit short. My disappointment only started setting in once I got to the bit where you’re supposed to prove you understand the case, which the game just didn’t have a good system for.

It’s just that it feels like the majority of the implementation effort was put into the procgen, and I don’t think it provides any value.

A real detective’s take on detective games ->

https://www.eurogamer.net/articles/2019-10-20-interrogating-detective-mechanics-with-a-real-life-detective

Pocket Detective is another of the single-deck-of-cards detective games.

Basically you collect clue cards, which will generally have about a paragraph of text + potentially pointers to other cards. On a player’s turn, they choose which card to visit and read it out. Once the group thinks they understand the case, you answer a couple of rudimentary questions on who did it and get a score. It deviates from the usual way these games work in two ways: how you answer the questions, and the handling of time.

The questions at the end are multiple choice, and you resolve the answers in such a way that if you get the answer wrong, you’re not spoiled on the real answer and can continue playing. This is such a minor benefit that it might not exist at all.

This doesn’t work at all. The need to have an explicit result card for every failed question means that they’re severely limited in the number of questions and the number of options to provide. So you only need a very minimal level of understanding of the case to answer them.

Second (and worse), the multiple choice format leaks information. We decided we’d gotten all we could get out of the case, we had a plausible hypothesis, and went to answer the questions. And one of the options made us realize we’d ignored something, we redid our hypothesis of the case on the spot, and got it right.

The handling of time is potentially interesting. For any card, the pointer tells you how much “time” visiting the clue is expected to take and how much “stress” it’s expected to generate. Every unit of time is -1 point, every stress is -10 points. The game tells you that these estimates are not necessarily correct, and gives an idea of just how many units of time is optimal / par.

The game does a couple of interesting things with the time. Some clues are blocked until the players have spent enough time (which is how they can force the players to investigate one part of the scenario before proceeding to other parts). And in other places, you’ll get different results depending on whether you visit the clue before or after a certain time threshold. That’s a rather neat way of focusing people into concentrating on the essential, rather than just doing a breadth first search through the scenario.

Unfortunately the part with the uncertainty of the time cost kind of breaks the game. Turns out there were things you were supposed to achieve in addition to the game’s stated mission of finding the facts. And those things are hidden behind clues that have an outrageously high cost of investigation that’d ultimately get refunded. It just makes no sense: the game actively and dishonestly discourages you from entering these parts of the story, and the heavily penalizes you for not doing it.

The story and writing were competent enough, with a suitable number of red herrings but everything eventually falling into place and making sense. But it’s also pretty slight. Combined with the rubbish question answering system and the issues with the time system, I can’t really recommend it. Sherlock seems like a better bet for detective games in this form factor.

Anybody following along with this? It looks pretty awesome.

Wow, that’s quite the twitter thread preview of the game. Very ambitious. I hope they pull off fun gameplay with all that procedural generation.

It’s one dude! 😱

I’ve had a look at the devlog on Tigsource every now and then, until the updates moved to videos.

I remain deeply skeptical about procgen detective cases being interesting, but can’t fault people for being ambitious.

These pixels please my eyes! @KristiGaines

Whoa. This sounds nifty.

Lk that does look good!

Wow that looks incredible. Even keeping my expectations in check, I’ll absolutely purchase it day-one in support.

Something about the combination of “stealth” with “detective game” sets off my spidey sense. If you guys try it and tell me it’s awesome I’ll try it, but I’m a little hesitant about this one.

I’m keen, for two reasons.

  1. I one starting making a fully procedurally generated detective game, so I’m always keen to try out others
  2. Concrete Jungle was great so I’ll buy the next game no matter what!

This is high on my list. Even if it is flawed, I’d like to support him such that he is encouraged to make a followup.

For a LONG time, I’ve wished I could program in order to create a 40s LA noir detective game, set in the world of Chandler/Marlowe, with some dynamic aspects and proc generated cases. So, for example, if you decide to go talk to the backroom casino boss who happens to have mob connections, and you push him and strong-arm him, he may tell his guys to give me a hard time (at best.) Or you may decide to try to play quid pro quo with him, and he has one of his guys give you an inside lead on some shady character who may have been involved. An entire world of such characters and options, playing out differently every time you play. Yeah, that’s my dream game of this type, so anyone who tries anything remotely close to it will get my dough.