Mysteries in IF

4 views
Skip to first unread message

David Michael Tuller

unread,
Apr 4, 1994, 1:50:22 PM4/4/94
to
I am currently trying to work out an IF mystery for TADS. The basic plot is
going well, but I would like to pose a topic for discussion here. What kind of
clues should be given in an IF mystery? I believe there are two basic types of
clues: tangible and intangible. The tangible clues include stuff like the murder
weapon which is generally a gun or letter opener or your generic blunt
instrument (Although, there are somtimes nice twists like the mandolin in Ellery
Queen's _The Tragedy of Y_). You also have anything left at the scene of the
crime where you can find fingerprints, etc. The intangible include how the other
characters felt, their alibis and their motives to name a few. The problem is
with deduction. I posted on this before and people seemed to agree that the
player should be forced to make deductions like those in Ellery Queen's novels.
The questions are: how do you make sure that the player has made the right
deductions before the game gives away the ending? how do you make these
deductions fair to the player? I don't want the mystery to be as simple as
_Deadline_, yet I don't it to be so complicated that no one can figure it out.
Now let the thread begin.

David M. Tuller
tul...@rpi.edu

Gerry Kevin Wilson

unread,
Apr 4, 1994, 11:16:12 PM4/4/94
to
In article <2npk0u$7...@usenet.rpi.edu>,
David Michael Tuller <tul...@rpi.edu> wrote:
[Some stuff deleted. Margins reworked a bit.]

>What kind of clues should be given in an IF mystery? I believe there are
>two basic types of clues: tangible and intangible. The tangible clues

>crime where you can find fingerprints, etc. The intangible include how the
>other characters felt, their alibis and their motives to name a few. The
>problem is with deduction.

A sticky problem, and one that deserves considerable deliberation. I
believe, and others may well agree, that deduction is what a mystery is
all about. What would Holmes be without "the game"? Thus, we need to
put considerable effort into it when writing an interactive mystery.
Now, how could we handle this in a text adventure? A graphic adventure?
Well, one thing that suggests itself to me is a new verb, "deduce". Not
very workable though. Remember in one of the Police Quests there was a
pattern to some crimes, and it formed a pentagram when connected on a
map. Once you marked the spots, I think either you then had to mark the
middle of the star or the game did it for you. That's one example of how
to handle one specific case of deduction. But think about the Holmes
story in which Holmes has an ordinary hat brought to him, and makes some
quite extraordinary deductions before explaining himself. How on earth
would we do something like that without a huge database of facts and
things the player could deduce? Well, we could take the LucasArts
conversation method and apply it to this. Less than ideal, but more
workable than a completely open slate. Still, there is probably a better
idea lurking out there, so let's hear it...

>The questions are: how do you make sure that the player has made the right
>deductions before the game gives away the ending? how do you make these
>deductions fair to the player? I don't want the mystery to be as simple as
>_Deadline_, yet I don't it to be so complicated that no one can figure it out.
>Now let the thread begin.

1. Use the obstruction method. The deduction is necessary to the
continuation of the game, and cannot be guessed. (Well, using my method
it certainly could, hence the less than ideal.) For instance, make the
deduction point to a location, or require the player to 'have a hunch'
about something before doing it. Artificial, but I can't think of
anything else.

2. More difficult. Define fair. Do you mean fair to Sherlock Holmes or
fair to Dr. Watson? Fair to the average Joe Schmoe? What is fair?
Doable? Simple? Discoverable? Again, a limited set of choices would
allow anyone to beat the game at the expense of challenge. How much
challenge it would take out I'm not sure. Probably quite a bit.
--
<~~~~~E~~~G~~~~~~~~~~~HEINLEIN~~~~~~~~~~~DOYLE~~~~~~~~~~~~~SPAM~~|~~~~~~~>
< V R I O Software. We bring words to life! | ~~\ >
< T | /~\ | >
<_WATCH for Avalon in early MAY!____wh...@uclink.berkeley.edu_|_\__/__>

Jamieson Norrish

unread,
Apr 6, 1994, 1:43:50 PM4/6/94
to
In article <2npk0u$7...@usenet.rpi.edu>, David Michael Tuller
<tul...@rpi.edu> wrote:

> What kind of clues should be given in an IF mystery? I believe there
> are two basic types of clues: tangible and intangible. The tangible
> clues crime where you can find fingerprints, etc. The intangible
> include how the other characters felt, their alibis and their
> motives to name a few. The problem is with deduction.

Well, I would say that leaving it entirely open to the player is best.
That is, present the information, and leave all the thinking up to the
player. If the player then "deduces" that so-and-so must be the
murderer, then that's all well and good. The problem then becomes -
how do you avoid the player simply guessing? In other words, the
player must be forced to "prove" the case. How this might work, I have
no idea.

Jamie

David Whitten

unread,
Apr 6, 1994, 2:09:47 PM4/6/94
to
Jamieson Norrish (ja...@kauri.vuw.ac.nz) wro te:
: In article <2npk0u$7...@usenet.rpi.edu>, David Michael Tuller
: <tul...@rpi.edu> wrote:

: Jamie


I think you have the germ of a good idea there.
but rather than deducing the name of the murderer, you deduce the
SIGNIFICANCE of the hints. IE: you found cigar ashes in the murdered
person's apartment. So you must 'Deduce A CigarSmoker was at the scene'

Now of course, none of the poossible murderers were initially described as
Cigar Smokers, so you must do more detecting to find out if anyone smokes.
When you find out that Fred the Artist used Cigarrette Ash to add 'just
the right color to the oil paints' that were used when he painted the
Victim's portrait, You must 'Deduce Fred Used Cigar Ashes'.

In other words, You can't just 'Deduce Sam is Killer' you must have already
told the computer your Deductions that point to Sam as the Killer.

I don't know how complex this would be to program, but I just like the idea
that just because the program has told the user some information, there is no
guarantee that the user know the significance of it. In a murder mystery
game, we could just require the player to 'register her deductions' so we
know that there is a deductive chain being built up.

Unfortunately, for other types of IF, we are still stick with assuming
someone knows the implications of messages we print because they can solve
the puzzle that depends on the information...

Dave Whi...@netcom.com (214) 437-5255

Tommy Nordgren

unread,
Apr 7, 1994, 11:14:10 AM4/7/94
to

--
The best way I've seen is that used in the Infocom mystery games.
I.E, the game is wrapped up, succesfully or unsuccessfully, when the player
decides to make an arrest.
Then the player will be considered succesful if the right persons
is arrested, and if there is sufficient evidence for a conviction.
-------------------------------------------------------------------------
Tommy Nordgren "We hold these truths to be self-evident"
Royal Institute of Technology "In this sign conquer."
Stockholm "The eagle has landed."
f85...@nada.kth.se "Please mister, could I have a cracker
to my onthatherium."
--------------------------------------------------------------------------

Ogawa / Taro Stephen (ISE)

unread,
Apr 9, 1994, 3:45:43 PM4/9/94
to
Firstly, an author recommendation: John Dickson Carr/Carter Dickson.
(THE expert in locked room murder mysteries - It's surprising just how
many different ways there are :)

Secondly - methods of solving:

1) Important person (Judge, policeman etc.) asks questions about who/when/why
etc, and you answer. All correct, and mystery solved
2) The typical way - Accuse Joe Bloggs, show x pieces of evidence, Joe
Bloggs jumps out the window, satisfying ending etc. :)
3) Catch Joe Bloggs in the process of committing another crime
4) Set up a trap.
5) Commit the crime yourself and frame somebody. This last method is
copyright ME and usable only by ME ;) (Actually this could make for
quite an interesting IF adventure - first commit the crime, then cover
up, then frame....But then I'm just a warped person :)

Taro.

Adam Justin Thornton

unread,
Apr 9, 1994, 9:27:59 PM4/9/94
to
In article <JAMIE.94A...@kauri.vuw.ac.nz>,
Jamieson Norrish <ja...@kauri.vuw.ac.nz> wrote:
>Unfortunately, a problem arises in that the computer must (I assume)
>make some judgement as to whether the deductions are sound and whether
>they in fact constitute evidence which is sufficient to have the
>murderer (or whatever) arrested.

This appears to be a job for Directed Acyclic Graphs.

Or maybe it's Petri Nets.

Anyway, you look and see if all the deductions leading up to a certain node
are satisfied, and that tells you whether or not you know what's going on.

Alas, this leads to the problem I found with _Witness_. I found the wires
and stuff, but didn't know what they were for. But I went through a couple
hours of game time fingerprinting stuff and asking questions. Then I got
lost and figured I'd arrest my best suspect so I'd see what the game told
me I didn't have, so I would know what I needed to look for next time.

Imagine my surprise when I won. Luckily, the precis gave me the correct
interpretation of all the evidence I had found but not understood.

Adam

--
Adam Thornton Ours is bigger VM Systems Programmer
(713) 630-8884 than yours. Galilean Reanimator
<not Rice's opinions> <save the choad!>
Lover. Dreamer. Geek.

Jamieson Norrish

unread,
Apr 10, 1994, 7:18:45 AM4/10/94
to
In article <whittenC...@netcom.com> whi...@netcom.com (David
Whitten) writes:

I think you have the germ of a good idea there. but rather than
deducing the name of the murderer, you deduce the SIGNIFICANCE of
the hints. IE: you found cigar ashes in the murdered person's
apartment. So you must 'Deduce A CigarSmoker was at the scene'

[...]

In other words, You can't just 'Deduce Sam is Killer' you must have
already told the computer your Deductions that point to Sam as the
Killer.

Unfortunately, a problem arises in that the computer must (I assume)


make some judgement as to whether the deductions are sound and whether
they in fact constitute evidence which is sufficient to have the

murderer (or whatever) arrested. This should have to take into account
the fact that although some deductions might be incorrect, the final
answer is, with enough correct deductions based on real evidence to
convict the criminal. Not something I would want to deal with.

Jamie

David Whitten

unread,
Apr 10, 1994, 4:44:50 PM4/10/94
to
Jamieson Norrish (ja...@kauri.vuw.ac.nz) wrote:
: In article <whittenC...@netcom.com> whi...@netcom.com (David
: Whitten) writes:

: [...]

: Jamie

I am of the opinion that if the player tries to convict based upon unsound
deduction, the player doesn't have a case, and thus hasn't won the game.

Yes, I assume the program would need to make deductions, thus would have a
protodeduction system built in. but this is easy. it jsut involves the use
of IF THEN statements and some simple looping operations. (would this be a
useful addition to hte archives..?)

It is part of the designer/writer's job to say waht are valid deductions and
what are not. If I as writer put cigar ashes in a room, I should be ready
for some one to 'Deduce CigarSmoker was here'.

My two pennies...

David (whi...@netcom.com) (214) 437-5255

Lon C. Thomas, J.r

unread,
Apr 13, 1994, 12:39:20 AM4/13/94
to
Here are my two bits about the "proper" method to win a mystery i-f, based
on a wild idea I got reading this newsgroup: Arrest & Trial. During the
Arrest phase, the crime is discovered, the clues are ferreted out ... all of
this in the normal adventure genre of scurrying about in various rooms. An
arrest is made, and then ...
>go to trial
"All Rise!"
In the Courtroom
The people in the courtroom rise as one as Judge Orris B. Goode enters and
takes his seat above them all.
( so on and so forth)

Trials are somewhat orderly events. A special set of verbs (SUBMIT into
evidence, CALL a witness, OBJECT(?)) could be developed. The case must be
developed by the player, getting the evidence out in the right order,
calling the witnesses. Then the jury decides (or the judge can make some
statement about how good the case was presented). I can see Mike Roberts'
directed dialogue working for this, or perhaps.

Lon Thomas

Reply all
Reply to author
Forward
0 new messages