How come there are no good detective games?

[MattH]

I think it's a good question. Computers and graphics are sophisticated enough today that one-case detective games (ex. CSI, Law & Order) are useless. How hard is it to semi-randomly compile a list of related topics, relationships, and clues? There are a good number of easily-identifiable factors:

- Type
- Crime (robbery, murder)
-- Weapon
-- Single / Serial
- Motivations (money, passion, etc.)
- Characters
-- Game-world relationship to victim(s)
- Clues
-- Based on weapon
--- Markings, etc.
-- Environment (hair, fibers)
-- Documents
- Difficulty (dumb criminal or Moritary)

The in-game world could be a section of a city, with houses, floor plans, offices, furniture, etc. generated from set parameters.

I might try making something like this, even if it is only text-based. But using a genetics system like the sims would make this an interesting retail title.

[vovan]

Great idea, MattH. I myself have been toying around with the idea of coming up with an algorythm to generate a random, but solvable, crime and then have the player solve it. I was thinking in a bit smaller scale, though. My idea was that the player would find himself in a city, like you say, but only the houses of interest would be accessible - with floor plans, and all. But then that would kind of limit the amount of choices the player can make, thus simplifying the solution for the crime.

Now, you are saying you would use a genetic algorythm kind of thing to generate a full crime? I am not sure really that a genetic algorythm would apply here... I mean, you have a crime after one generation of the algorythm, but then how would you determine the fitness of the genotypes to bring to the next generation? I imaging it would be hard to analyze the "difficulty" of the crime programmatically... And then, what would be the other criteria?

[MattH]

Originally posted by vovan
Great idea, MattH. I myself have been toying around with the idea of coming up with an algorythm to generate a random, but solvable, crime and then have the player solve it.

Hmmm-- So I'm not totally crazy then.

I was thinking in a bit smaller scale, though. My idea was that the player would find himself in a city, like you say, but only the houses of interest would be accessible - with floor plans, and all. But then that would kind of limit the amount of choices the player can make, thus simplifying the solution for the crime.

I was thinking in the same limiting way. Most people wouldn't let you in without a search warrant, which you would have to have X amount of 'Strong evidence', or Y amount of 'Inconclusive evidence' etc. If there was another house you wanted to go into (and the owner let you), the game could randomly generate it on the spot and it would serve as a waste of time, losing you time or letting the evidence grow old (or some time-limiting factor that would make it possible to loose).

Now, you are saying you would use a genetic algorythm kind of thing to generate a full crime? I am not sure really that a genetic algorythm would apply here... I mean, you have a crime after one generation of the algorythm, but then how would you determine the fitness of the genotypes to bring to the next generation? I imaging it would be hard to analyze the "difficulty" of the crime programmatically... And then, what would be the other criteria?

Well, you could pick a difficulty (or the game would ramp up). Then the game would decide on a type of crime- easier involving basic robbery, small crimes, all the way up to hit men and bank robbery. Based on the crime, the game would generate most of the remaining factors:
-A location (seedy section of town, docks, country club).
-The characters involved (ex. Business man killed: partners, someone with a grudge.
- Random names, personality (up front, liar, shy)
-The motive
-The M.O. (strangulation, smash and grab) and further details (
-A weapon that fits with the 'nature' of the crime.
-Evidence that fits with the nature of the crime. For example, DNA under fingernails for a fight.

This is not in order since I haven't thought it out fully. As you can see, the probability tree would be huge but logical. It would, however, need enough changes that you can't solve the case by knowing the algorithm ("Aha- if the guy was drowned, that means it's Mr. Plum with the candlestick in the bedroom)

[MattH]

Oh, and evidence would be logical too. 'Papers' would go on or in 'Desk' 'Cabinet' 'Safe', or in care of 'Lawyer', 'Banker', or 'Wife'. Bullet traces could be analyzed CSI-Laser style.

The really hard part will be generating lifelike conversations.

[DrSpike]

Not sure if you would be interested but out of games that exist you might want to look at the excellent Discworld Noir , which has a unique 'clue' system. Also there is the solid Sherlock Holmes series.

[MattH]

Discworld... Any relation to the terry pratchett discworld?

[DrSpike]

Yeah but it's nothing like the cartoony Discworld adventure games. It's the story of a hard-boiled Bogardesque PI. Try it.

[MattH]

I'll have to 'look' (cough) for it right now. I browsed for the discworld game a long time ago to see if it would invalidate my thought for a discworld Morrowind mod, but all I found were those terrible cartoons. I though it would just be a sequel.

[Sava]

How come there are no good detective games?

because they suck?

seriously... being a detective might be fun to watch on TV, like Law and Order , but it would be boring and monotanous for a game.

[vovan]

Originally posted by Sava
seriously... being a detective might be fun to watch on TV, like Law and Order , but it would be boring and monotanous for a game.

Well, no necessarily... Take the board game Matt was citing for example... I think it's called Clue? It's a good board game. Therefore, one could make a good computer game based on the concept. I guess the incentive to solve the crime should not only be for the sake of solving the crime, though, but let's say, to beat another detective to it. You could even build in some role-playing elements into it.

Say you could choose certain skills for your detective:

1. Speech - can pull the facts out of the suspects and witnesses well.
2. Attentiveness to details - can spot clues that evade a less careful observer.
3. etc.

As you solve more crimes, your skills go up. As well, one could build in things like the career ladder: you start out as a constable, and work up the ladder to become the chief of police.

[MattH]

I was thinking of the game as more a challenge to the player, not the in-game character. Skills etc. are limiters or assists for games that go too fast (in this case) or are impossible (RPGs).

[vovan]

Originally posted by MattH
I was thinking of the game as more a challenge to the player, not the in-game character.

So more of a puzzle kind of thing?

[pg]

i thought there were some games based on csi or law and order. what are those exactly?

[DinoDoc]

Beats me. I kind of liked Dick Tracey on the NES.

[vovan]

Well, there's "CSI: Crime Scene Investigation" and "Law & Order". I haven't played either myself, but from what I gather, the replayability isn't that great. I think it would really be an interesting challenege to come up with a system for dynamically generating crimes to solve...

[MattH]

As far as I can tell, they're one-shot, one-crime, one-solution games. I. E. NO replayability. What I'm looking for is a game styled after those but created from the algorithms that I've listed above.

[okblacke]

Originally posted by vovan
Well, no necessarily... Take the board game Matt was citing for example... I think it's called Clue?

"Clue" has been the basis for a number of retail and shareware games, though generally using a more complex scheme than the 6x6x9 format of the board game. (The board game has six potential playres, six potential weapons, and nine locations.)

There have been a number of text and graphic adventure mysteries. "Blade Runner" is, essentially, a mystery. Mysteries are incorporated into many RPG games. (Many mysteries, usually.)

One perennially popular board game "221B Baker Street" has a good format that could be adapted.

The problem with random crime generation is that you need a lot of dialogue and atmosphere. Tough to do in a text environment. Tougher (maybe impossible) if you want decent voice acting.

"221B" did it by having packs of adventures you could buy. That could work well in a computer format.

[MattH]

I thought about the voice part. While speech synth is getting very flexable (Mac OSX), it's not up to the standards that a game like this demands. That's why I'm thinking about the dialouge in a Morrowind or NeverWinter Nights style.

[Jamski]

It could be done in NWN. In fact I could do it. The trouble is - to make it replayable you need soooo much redundant material (i.e. dialogues, items, loctations that are never seen by the player) that I'm completely daunted by the whole idea. It would be an enormous project.

-Jam

[DrSpike]

Exactly, which is why the one-shot but rich adventure game system predominates. The Sherlock Holmes games are great if you like murder mysteries. You'll only play them once though.

[MattH]

Jam: can you create locations using scripts?

[Sava]

I remember the older detective sims... there were a few good ones for NES... Maniac Mansion... umm... crap there's more. Those games were fun, but later games of that genre just sucked. The old Monkey Island games were great... the new ones were too cartoony for my tastes. I liked the games where you moved around with the mouse and searched for items and your commands were "use, push, pull" , etc.

I don't think a detective game like you are describing would be easy to make. It's hard enough to script an entire game, but writing some sort of random case type game would just be too hard. Maybe you can make your own

[vovan]

So, I've been thinking about the idea a little more, and it seems to me that there are two directions you could take this in:

1. Adventure game. This is what MattH is talking about. Here, the solution of one crime is the whole game. You start at the beginning, the enormous storyline is generated, and you go through it. After you are done, it's game over, and it's time to fire up that generator again. I think this would be really hard to implement, just because of the sheer amount of stuff than needs to be generated randomly. If you want a game to last at least 40 minutes, that's already a lot of story and dialog to create. To achieve any kind of realism and coherence in the story, you would already have to work hard - to make the stories vary enough to not become boring after a couple trials will make this even harder.

2. An RPG kind of thing. This is what I was more leaning towards. Here you have some kind of big plot that is recurring from game to game. Say, we were making a Sherlock Holmes RPG, then the grand goal would be to catch Moriarty. No matter how many games you start, that's your ultimate goal. However, along the way to that goal, you need to perform a number of "side quests", sort of like in Fallout 2 - you don't go and kill ol' Frank right away - you first roam about the wastelands and solve numerous other quests. Similarly here: you don't go after your ultimate goal right away - you must first solve a number of smaller cases - starting with maybe thefts, going on to burglaries, rapes, murders, and finally serial murders. Now, these cases would be randomly generated each game. I'm thinking this RPG system would be much easier to implement, simply because the game would not revolve around the one single case, but rather, each game would be comprised around solving a maybe 20-30 smaller cases. Then, the even the amount of information generated for every given case is going to be much much smaller. That way, it would be much easier to achieve diversity in the cases. Also, it would be easier to make them more coherent, simply because there are fewer pieces of data that need to be stringed together.

[Space05us]

Theres an old DOS detective game. Your Dad is jailed for the murder of your mom iirc, and you have a limited number of days to prove him innocent. The setting is somewhere in England I think, I remember the bar was called a pub. If you played your cards right you could get a date with a nurse from the hospital, and if you werent careful you could end up getting killed before the limit was up.

I'll go look fer it.

[Standup]

Bladerunner from what i remember had several of the detective elements. I always much prefered those parts to the 'arcade' elements. It also had some limited replayability.

[duke o' york]

Originally posted by vovan

2. An RPG kind of thing. This is what I was more leaning towards. Here you have some kind of big plot that is recurring from game to game. Say, we were making a Sherlock Holmes RPG, then the grand goal would be to catch Moriarty. No matter how many games you start, that's your ultimate goal. However, along the way to that goal, you need to perform a number of "side quests", sort of like in Fallout 2 - you don't go and kill ol' Frank right away - you first roam about the wastelands and solve numerous other quests. Similarly here: you don't go after your ultimate goal right away - you must first solve a number of smaller cases - starting with maybe thefts, going on to burglaries, rapes, murders, and finally serial murders. Now, these cases would be randomly generated each game. I'm thinking this RPG system would be much easier to implement, simply because the game would not revolve around the one single case, but rather, each game would be comprised around solving a maybe 20-30 smaller cases. Then, the even the amount of information generated for every given case is going to be much much smaller. That way, it would be much easier to achieve diversity in the cases. Also, it would be easier to make them more coherent, simply because there are fewer pieces of data that need to be stringed together.

I think that you could go with an RPG style. Basically you need a certain number of experience points to get promotion to Sergeant, etc, and you have to collect them by solving crimes. For each level there are about ten crimes, increasing in difficulty as the levels go up, and you would need to solve maybe four of these per level to be promoted. Basing your game entirely on side-quests would prolong its playing life, and you could have many many different ways through to the final crime, which would probably be the same each time round. If you were going to use something like Jamski's NWN suggestion, then you could have different crimes taking place at different times of day, and the extras would move about following a daily routine. Various shops would open and close according to time of day, and different people would be in different locations. With each level you are promoted, a new area of the city is opened up (like in GTA), but you still have to go back to your old favourite grass, etc for the more advanced investigations.

[Barinthus]

Where in the World's Carmen Santigo was a great game back then.

While it's not your traditional detective story, a lot of skills related are used in the King's Quest series.

[vovan]

So, in NWN, you could essentially randomly generate events and quests at every replay of the game?

I guess what bothers me most about this is that you cannot really at present implement a truly flexible storytelling engine, without which such a game would suffer (and the adventure variety would be entirely impossible). What one would have to do it seems is essentially provide a bunch of pieces of story to the game to string together randomly into a whole. But to achieve a good replay value, like Jamski said, you'd have to have a LOT of these little pieces.

[bipolarbear]

There was actually a really great game for the NES called Nightshade, where you were a dectective, and you had to solve all of these mysteries. It was quite advanced as far as interactivity is concerned for a game of its time. It was really, fun, and it was really long too. The big downside though, was that there was no way to save your progress at all. I think that's why it never took off.

[bipolarbear]

I've tried te demos for the law and order games, and to be honest, they really suck. I think that if there were dectective games that were sort of like that one part of KOTOR where you have to play detective on Dantooine, only waaaaaay more heavy on the detective end of things, then that would sell really well. It would have to include a little bit of gunplay too, just to keep things from getting too dull.