Authoring detective games

I think this should definitely be an important driver in picking a system: how much fun am I going to have working in this system? For some of us, we may have more fun building than playing. I name no names (cough)

For example I don’t like Javascript but I can live with it when I am just plugging in small functions into a framework and when those small functions drive something cool, it’s fun :slight_smile:


Nobody mentions Phoenix Wright games? I imagine a text version of it can be easily implemented either using Twine or Inform.

1 Like

This is exactly what ClubFloyd does, albeit virtually, not physically. They pick an IF game and play it collaboratively together. After checking out some of the games, maybe look at the corresponding transcript and see what you think of the experience. It’ll give you a good idea of what happens when multiple people play a single-player IF game.


Thanks for posting that. you beat me to it.

Secret they dont want you to know: This is true in _any _ system. :slightly_smiling_face:


Hello, thanks for the ping @manonamora.

Having read the thread, as to your initial question @ricm I think Twine might be helpful, because you can build a more board-game-like UI that allows you to see and compare pictures, text, and other clues which you might want to cross-reference, as opposed to the linear text form inherent to Inform which makes it harder to easily reference past content. That said, I think you’d need to do a lot of UI work.

However, going back another step, why do you want to make a video game, instead of making a board game? I think if you’re aiming to have it played in a group where people are physically there, as you’ve indicated with the idea that they’d play around a tablet or on a screen, you may as well just make a board game. You mentioned modeling your prototype off of 221B Baker Street having a limiting design, but there are also a lot of different detective-style games in the board game space which aren’t strictly based on boards, locations, and rooms, and in the context you seem to be describing the the game design seems agnostic between digital and physical forms.

In any case, even if you’re dead set on building a computer game as opposed to a board game, I’d recommend actually prototyping your game as a board game, if you can. If you’re not already conversant in Twine/Javascript/UI design, it can take a lot of effort to get up to speed, as you seem well aware.

Anyways, if you haven’t already, I think you might want to give some other tabletop stuff a try, like Sherlock Holmes: Consulting Detective, a text-only gamebook (Sherlock Holmes: Consulting Detective (gamebook) - Wikipedia, Perspectives, a collaborative text/image multiplayer detective game (Perspectives | Board Game | BoardGameGeek) or Deception: Murder in Hong Kong, a Werewolf-style social game with detective theming (Deception: Murder in Hong Kong – Grey Fox Games).


You can throw directions out the window and use Inform for abstract locations if you like. There are ways in which choice-based is a better fit for conversation than parser-based, but parser-based games often mitigate it by using a choice-based conversation mechanic. Twine has less built-in support for a world model, which might or might not matter depending on the details of what you want (and you can implement something world-model-ish, with work).

With apologies for veering from the question as asked, here are some links of interest regarding mystery IF:


I tried those games and what a revelation. Initially I was leaning towards parser-based games believing this would be most appropriate for interviewing witnesses and suspects. The experience of asking questions left me floundering and progress, if any at all, was slow and hard won. I didn’t finish any of those games. When I say “hard won” I don’t mean because the level of inference/deduction was challenging but because I couldnt find obvious ways of asking questions or directing actions. Responses like “I don’t understand x, y or z…” or “That won’t help you solve this game” were overall unsatisfactory: incredibly frustrating and diverting energy from problem-solving towards grappling with the mechanics of the game play.

The best experience was with Erstwhile. The options both confine the range of what to do next and also support the player by providing potentially useful approaches/questions and lines of enquiry.

My feeling about Erstwhile confirmed to some extent my earlier worries about providing actions that players might want to do. Why, at one point, would I want to possess someone and make them confess to a murder? I had no idea then who did what to whom and, to be honest, I’m still not sure “who did it”. There is a real danger of option-based games leading players by the nose. I think that’s quite challenging.

And the finish was terribly unsatisfactory: at what point was the mystery solved? What did I have to do to prove I knew? Linking “facts”? Only they weren’t facts. There was no evidence, nothing to weigh and attach importance or significance.

Erstwhile had many fine points, however, and it taught me much about how to keep up the pace of the game.

Another lesson: rich narrative can help an individual player become immersed in the story. That same narrative could be a source of frustration to people playing collaboratively because they would more than likely demand “Get on with it!” “Get us to the clues!”. The target audience is important.

The goldmine, however, was “the-case-of-the-thinky-game-jam”. The videos were a delight to follow and constitute what might be termed ‘essential reading’.
I will re-run those videos perhaps many times before I am able to take on board the key lessons about the activity of detection versus role-playing a detective in what might otherwise be a logical puzzle.

I cannot thank you enough for providing that link.

I started out believing I could do better than this or that - only now to realise how difficult it would be to craft a game that gives players that “Solved it!” rush. The hill is a steep climb; the prospect daunting.

Much more reading and research to do.



Yes, that’s my instinct too.

Given html/css/javascript game platform there would be no reason why individuals should not each start a game at the same time and co-ordinate between them the various lines of enquiry. Each instance/player/team could then contribute their findings to the whole group.

The one thing missing would be the lack of interaction at the software level that would change the game status according to the individual enquries and choices made.

Time for the old “evidence board” beloved of the TV detective genre to co-ordinate/summarise findings…? Hmm…

I’m going to try more games.


1 Like

Hi, vaguely related, but I originally made my first game (LUNIUM - Details) to be played as a group (it even worked well as a team-building exercise). Playing as a group gave the game a very different dynamic and really encouraged team-work. This was also before I’d heard of this strange thing called ‘interactive fiction’! Although it’s essentially a detective game, it’s more of a virtual escape room format rather than anything resembling a board game. It was played collaboratively either together or online using Zoom/Teams – and makes use of images to set the scene and puzzles. It was created in Twine, so if that sort of choice-based interface works, then I can recommend that format.


As a parser lover, I feel compelled to defend parser games. It takes some time to learn their conventions, what the shortcut commands are, and the general way they go about things. You are not going to appreciate them after trying a few with no experience for one day. I see people all the time who don’t know the ropes, try commands that really aren’t appropriate, and then write parsers off as bad or frustrating. I think choice-based might be exactly what you need for what you want to do, but please don’t write off playing parser games after such a little bit of exposure.


I’d also add that the parser examples are really old (Ballyhoo, Deadline) or idiosyncratic in their interfaces (Toby’s Nose), and modern systems and more standard designs are generally more intuitive. It might be worth looking at Color of Truth, which has menu-based dialogue as well as a clever structure, and Happyland, which is a modern take on the classic Deadline-style mystery.

(With that said, there are a lot of cool choice-based mysteries too! But some of the stuff you’ve flagged as your design goals makes me think a parser approach might be worth considering in more depth).


You are right. My exposure to parser-based games is minimal. But I didn’t reach my assessment lightly nor arbitrarily.

The first time yesterday I played “Toby’s Nose”. I didnt get anywhere. So today I tried again and managed to get further only to arrive once more in a position where nothing I typed moved the game forward.

There is clearly a mind set that allows some people to get inside the verb-noun constructs more easily than others. It’s a genre with its own way of phrasing problems in a way the parser can accept. And this is the stumbling block for me.

To make a poor (and I wont object if you think unfair) analogy is it’s not just throwing dice between moves in a turn-based game. But learning how to throw dice - and to do that every move.

My experience with the other parser-based games was similar. Maybe I’m rubbish at that mode of thinking. Heck, I’m rubbish at lots of things, so discovering one more won’t hurt. But. I do need to focus and I do need to converge on a solution that is the best fit to the overall aim.

The responses here, collectively, have been amazing and I have learnt much by following the various suggested lines of enquiry. It’s inevitable the field would narrow.



I am making notes as i work through the suggestions that have been made. It’s been a bit of a helter-skelter ride but slowly i think a more precise “wish list” is emerging.

I’m very happy to try your suggestions.

It may be that, having tried other types of game, I’m brought back to parser-based games as the solution of best fit. But I haven’t yet made that journey.

One thing I do know: I have to try writing games in both mediums. I need to experience what it’s like. Playing ready-made games is instructive but it’s a final chapter. Crafting a game comes much earlier.

Thanks again,


Aside, that’s why I switched from mainly authoring in parser to choice, because the authorial control is greater, beta-testing/bug-fixing is easier and there’s less “what do I do now?” for the player. Parser fans mark that as a “limitation” and enjoy the freedom of being able to stack boxes on a table while a suspect is sitting in the chair waiting to be questioned and wondering what in the Benoit Blanc you’re doing!

Getting the player to do what you want in parser is much harder, which is why (in my personal opinion) parser is better for exploratory games “after an event” (like gathering evidence) rather than creating any plot momentum - any good parser player is going to throughly explore before advancing the plot so you have scenes where the detective mercilessly ransacks a desk rather then questioning the Mayor who is sitting right there like they’re “supposed” to… :smiley:

And many new players who didn’t grow up playing Zork unfortunately!

One thing that may have been mentioned is going hybrid. There are a couple extensions that allow choice menus in a parser game. I used the extension “Hybrid Choices” in Fair to automate conversations and interactions that would cause major “guess the verb” situations. One of the activities in that game is trying to sell books which could have been a nightmare of nouns and verbs and mixed in conversation and “what do I do next?” (should I make verbs CONVINCE, PEDDLE, HAGGLE the player can use mid-conversation?) so I automated it with menus.

That could be a good balance for a mystery game which benefits from detailed parser exploration at a crime scene, then switches to choice menus so the author has some control of conversations and major plot scenes that require momentum that would killed by the player trying to examine the wallpaper or ransacking a desk while a gun is being waved at them.


If you want a modern, robust parser mystery, I’d suggest the train dimension in Mathbrush’s ‘‘Never Gives Up Her Dead’’. You talk to four people aboard a train to figure out who murdered a passenger. Clues get added to a list, and you can ask anyone about them. Once you find a contradiction, you can link two clues to get another one which is more pressing. You also get to play through each passenger’s alibis and witness them firsthand. It runs on an extension he published publicly, but my mind is blanking on the name.


That triggered a thought I believed long forgotten. I recall teaching a group of physics students back in the 70’s. At one point after various attempts to answer my question, a student, despairing as much as me about the lack of an answer, piped up: “Oh Emma, you’d know how many hairs were on the bum of a mouse and not know how many ears it had”.

Yes, I know, a bit random… apologies.


I don’t know if this has already been said or not, but games like Photopia do the conversations well (you should see to understand).

I shall try Photopia.

In the last few years the world has changed massively in respect to machine understanding of speech. I imagine the future of conversation-like games and games where questions can be asked now lies in the various implementations of large language models. If people can now ask online chatbots, or indeed Alexa, about this or that or pose questions via ChatGPT, the experience of anything less responsive will have an increasingly hard time.

An interesting future though.


1 Like

Co-creator of Erstwhile here (the other author is not in the community) and I find these responses very odd, because the game is very, very explicit about who killed whom on the surface at least once you get to the end of the game, and is centered around finding clues (albeit non-standard detective game evidence – I admit it’s a non-standard detective game). Also…the player character was killed, which is clear by the end of the intro.

I realize that maybe this is on us for designing it so that you can possess someone and make them confess to the murder straight off the bat, after the intro. But I feel like that is fairly well-discouraged by the game, and if you possess someone as soon as you get to the police station, you have missed 90% of the game.

(Thank you for playing though, it’s nice to see people’s take on it even after all this time :slight_smile: )


There is a very tricky balance between parsers being too open and difficult to navigate (at least for newbie players) and choice-based games “leading players by the nose” too much, ie not feeling open enough. It depends on what you want to lean toward, a more “open world” set up like Make it Good or Toby’s Nose (parsers) or choice-based ramps like Erstwhile.