[Theory] A Typology of Interactive Fiction

12 views
Skip to first unread message

Nick Montfort

unread,
Dec 6, 2001, 12:33:14 AM12/6/01
to
To add to the theory discussion I've posted "A Typology of Interactive
Fiction" at < http://nickm.com/if/typology.txt >. This is a 2400-word
first draft.

Summary:

Looking to the typology of text machines in Espen Aarseth's Cybertext,
and starting with the variables of his that are relevant to IF
classification while considering other factors from recent theory
discussion on raif, I offer seven independent variables for IF which
each may have either two or three values:

Determinability (indeterminate, determinate)
Transiency (intransient, world-transient, text-transient)
User Functions (explorative, configurative, textonic)
End State (one, many)
Player Characters (one, many)
Non-Player Characters (no, yes)
World Graph (regular, complex)

This typology provides 288 distinct categories of IF and I believe
distinguishes types of IF based on salient rather than arbitrary
features. When numerous works are classified a statistical analysis of
IF overall can be conducted of the sort Aarseth does of ergodic texts
in Cybertext. Clusters can be observed and works compared based on
where they are. All this will be done, in all likelihood. Before I
begin on this, I'm interested to see whether this typology seems sound
and useful.

For starters, I classified Adventure, Zork, Mindwheel, Photopia,
PUTPBAD, Ad Verbum, Dark Mage, Border Zone, Exhibition, and Rematch,
leaving fellow theorists to add their own IF and other works and to
see how reasonable these variables that I've selected actually are.

-nm

Gabe McKean

unread,
Dec 6, 2001, 1:09:03 AM12/6/01
to
I have a question regarding the "Determinability" variable: Several games
use daemons to produce random 'background' text from time to time. However,
this kind of behavior doesn't usually affect the world state. Since you
seem to be concentrating on text-generation in your definitions, would you
say that these games are indeterminate?

I think your typology is interesting, but I personally don't think an IF
typology will be complete unless it takes the world state of a game into
account on at least an equal footing to the text state.


Adam Cadre

unread,
Dec 6, 2001, 3:57:38 AM12/6/01
to
Very slight Photopia spoilers follow--

Nick Montfort wrote:
> For starters, I classified Adventure, Zork, Mindwheel, Photopia,
> PUTPBAD, Ad Verbum, Dark Mage, Border Zone, Exhibition, and Rematch,
> leaving fellow theorists to add their own IF and other works and to
> see how reasonable these variables that I've selected actually are.

It appears that you've listed Photopia as having no NPCs, when there are
lots -- chief among them Alison Dawson, but also Rob, Joyce, Sherrill,
the weather salesman, Gabriel, et al.

There's another category I'd like to discuss, but I'll wait till after
my next game is out since it's sorta the test case.

-----
Adam Cadre, Brooklyn, NY
http://adamcadre.ac

Alexandre Owen Muniz

unread,
Dec 6, 2001, 5:48:01 AM12/6/01
to
Adam Cadre wrote:
>
> Very slight Photopia spoilers follow--
>
> Nick Montfort wrote:
> > For starters, I classified Adventure, Zork, Mindwheel, Photopia,
> > PUTPBAD, Ad Verbum, Dark Mage, Border Zone, Exhibition, and Rematch,
> > leaving fellow theorists to add their own IF and other works and to
> > see how reasonable these variables that I've selected actually are.
>
> It appears that you've listed Photopia as having no NPCs, when there are
> lots -- chief among them Alison Dawson, but also Rob, Joyce, Sherrill,
> the weather salesman, Gabriel, et al.

And none of them may count, according to the non-standard definition of NPC given. Which,
I think, illustrates the uselessness of this variable as it is defined. The existence, (or
non-,) of NPCs with which one can have meaningful interaction is more important than the
presence of objects which run around the map, chewing the scenery and stealing items from
the PC.

Likewise, for the world graph variable, I think a better way to divide types is with the
question, "Is it always possible for the pc to return to a location once it has left it?"
The existence of teleporters and one-way passages can be a merely cosmetic device if there
is a "long way" of getting from any place to another, and doesn't necessarily say anything
meaningful (to a player,) about the work.

This gets back to the point about access. I think access may be a more meaningful variable
if one expands it to be "all text may be accessed from any point within a finite number of
turns," rather than within one turn. In general, almost all IF will still not meet this
definition, but we can speak meaningfully about 'strongly connected domains' within a
work, that is, sets of actions that can be taken without changing the gamestate so that it
will no longer be possible to repeat the result of an action. Replace 'text' with
'locations' and you get my formulation for world graph types above.

**Owen

Nick Montfort

unread,
Dec 6, 2001, 12:34:52 PM12/6/01
to
Alexandre Owen Muniz <mun...@xprt.net> wrote in message news:<3C0F4CE1...@xprt.net>...

> Adam Cadre wrote:
> >
> > It appears that you've listed Photopia as having no NPCs, when there are
> > lots
>
> And none of them may count, according to the non-standard definition of NPC
> given. Which, I think, illustrates the uselessness of this variable as it is
> defined. The existence, (or non-,) of NPCs with which one can have
> meaningful interaction is more important than the presence of objects which
> run around the map, chewing the scenery and stealing items from the PC.

What I've tried to write about are "actors" in the sense discussed in
the Zork IEEE paper. Adventure has none, although it has a dwarf that
can kill the PC and a pirate that can steal treasure. I don't think
the concept (which seems like it can be made formal) is useless,
although it may not be the best one. I do think it's important to
distinguish an NPC from the mention of a character in text, whether by
reference to actors or not, and I don't think the criterion of
"meaningul interaction" is rigorous enough to do this unless that can
be a formal term somehow.

Perhaps you or others would provide a formal definition of an NPC that
you find useful, and explain if things like the dragon, snake, and
bird in Adventure; the hollow voice in Zork; the tiny figure in Shade;
the Wizard of Wordplay in Ad Verbum and the parrot in Christminster
are NPCs by this definition? These definitions could also be used to
determine which of the people mentioned in Photopia are NPCs and which
aren't. Presumably "the girls" mentioned as waiting for the two
partygoers in the first segment are not NPCs by anyone's thinking, for
instance, although they motivate the situation.

> Likewise, for the world graph variable, I think a better way to divide types
> is with the question, "Is it always possible for the pc to return to a
> location once it has left it?" The existence of teleporters and one-way
> passages can be a merely cosmetic device if there is a "long way" of getting
> from any place to another, and doesn't necessarily say anything
> meaningful (to a player,) about the work.

This might be better. Even here, a single door that locked behind the
player or a room that caved in and disappeared would give a "no"
answer to that question and put it in the same category as
Christminster, HHGTTG, All Roads, and Photopia. But I think it makes
for a more interesting division of IF than "regular" and "complex."

> This gets back to the point about access. I think access may be a more
> meaningful variable if one expands it to be "all text may be accessed from
> any point within a finite number of turns," rather than within one turn.

So if any irreversible actions can be taken, for instance, the answer
is no; if you were able to destroy the widget that would mean that the
text "Widget: Dropped." (not to mention the description of the widget)
could not be accessed. Of course, this excluses "irreversible" actions
that only expand one's options for what texts can be generated. I
think this is meaningful and suppose Rematch, Lost in New York, Pick
Up the Phone Booth and Die, and Exhibition are some examples that give
"yes" answers to this question. I'll mention that if a word is
randomly selected at the beginning of the game to be displayed at any
point, the answer is also "no," since you have to start the game over
for that word to be selected once again.

On a side note, I'm surprised that no one has taken me to task for
excluding Aisle from IF and including Rematch. I think the two have to
be included or excluded together, no?

> we can speak
> meaningfully about 'strongly connected domains' within a work, that is, sets
> of actions that can be taken without changing the gamestate so that it
> will no longer be possible to repeat the result of an action.

Yes, definitely. This is a very useful formalization and I think it or
similar approaches could be useful in describing what Emily is talking
about with reference to "benchmarks," if this is a formal concept.

-nm

Lucian P. Smith

unread,
Dec 6, 2001, 12:44:01 PM12/6/01
to
Nick Montfort <ni...@nickm.com> wrote in <ae605ba7.01120...@posting.google.com>:

: I offer seven independent variables for IF which


: each may have either two or three values:

: Determinability (indeterminate, determinate)
: Transiency (intransient, world-transient, text-transient)
: User Functions (explorative, configurative, textonic)
: End State (one, many)
: Player Characters (one, many)
: Non-Player Characters (no, yes)
: World Graph (regular, complex)

: This typology provides 288 distinct categories of IF ...

Hmm.

In order to be something other than an excercise in pedantry, I believe
any attempt at IF classification must accomplish two things: distinguish
works of IF one from another, and tell you something useful about those
works.

I think some of these categories, while they may distinguish one work from
another, do not do so in a helpful manner.

I realize I'm being somewhat vague. Before I go on, let me try to
formalize what I mean by 'helpful'. I want scales (and I probably would
prefer scales to binary states) that will let me, as the
player/interactor, know something about how I will be interacting with the
work. I want to know where a work falls in terms of its gameplay. I want
to know what kind of thinking I'll have to do while engaged with the work.

Was that Espen's goal? I can see how many of these scales might actually
well distinguish between various interactive works out there. For just
IF, some of these become less useful.


OK, specifics:

Determinability: Text changing randomly is one thing, the gamestate
changing randomly is another, and the gamestate changing randomly forcing
the player towards different end states is yet another. I'm not exactly
sure how to formalize this; perhaps I'll just give some examples and we
can try to formalize them:

-Games where nothing is random
-Games where 'flavor text' changes (I think it still could be useful to
distinguish this case from the above, since it can affect mimesis, esp.
for replays of the game)
-Games with varying 'flavor text' that require the player to change their
input to match. In 'Edifice', you can slide off a horse to the right or
to the left, and getting back on requires taking that into account.
-Games that randomly change the entire game state, requiring different
swaths of input from the player. The bat carrying you off to a random
location in Zork Zero, maybe, or the randomizer in 'Guess the Verb!'. One
can imagine Grip sending you to Fit 2 randomly instead of
deterministically, too (and, from the player's perspective, it *is*
random though there's no random-number-generator at work. Hmmm.)
-Games that randomly change the game state that take control of the game
away from the player so much that you reach an end state because of it.
If the troll kills you in Zork I, the fight in 'A Crimson Spring', the
bees dying in 'Adventureland'.


Transiency: In r*if parlance, this would be called 'having a real-time
component'. Again, how this component affects the player's response is
important, as well as how much of the game is real-time vs. turn-based.
Your 'world-transient' vs. 'text-transient' gets at the reader-response, I
think, but I'm starting to think the response should have it's *own*
formality. Hmm.

User Functions: First off, I'd argue that some IF can be 'interpretive'
in parts--this is what people are complaining about when they talk about
'text dumps'. 'Ramses' also veers off in an interpretive direction.
Explorative vs. configurable are indeed useful terms--they're what we mean
when we talk about exploration games (or parts of games) or puzzle-based
games. 'Textonic' is a nigh-useless scale, but I suppose it has its
place. The only game I can think of that's usefully classifiable as such
would be 'The Other Side'. Oh, and maybe 'Lists'. Can we exclude those
games from 'formal IF', like you've done with SUTW, and forget about
'textonic'?

Another way of looking at 'textonic' is that *all* IF is textonic because
there are usually many many instances of the game repeating the verb you
just used back to you. JUMP / You jump on the spot, fruitlessly.

(Also, I don't think you can say a game is explorative *or* configurable;
rather, those are two scales that both can be present or not in a game.)

End State: I've already started using this term above, so it must be
useful ;-) It's not always very descriptive of a game, however. Maybe
another term would be helpful: 'Parallel Internal States'. A game with
lots of parallel internal states is 'wide'; a game with *no* parallel
internal states is 'railroaded', even if at the very end it suddenly
branches out to many end states.

Player Characters: I think it might be helpful to talk about parallel
player character change vs. linear player character change. By which I
mean, basically, Suspended vs. Photopia. Common Ground is an interesting
midpoint on that scale, too.

NPC/actors: I'm afraid you're not going to be able to steal the term
'NPC' or even 'actor' to mean what you want it to mean. Perhaps you could
classify NPCs as 'active' or 'passive'? I'm not exactly sure what you
want it to mean, either, having not read the Lebling paper. Do you mean
that, say, an ideal 'active NPC' would be indistinguishable from another
PC on a multi-player platform? Maybe 'game-controlled PC' would work to
convey what you mean more clearly. That would give us natural terminology
to talk about Common Ground, too: 'Common Ground has linear player
character change, but uses game-controlled PCs that mimic earlier PC
actions to create the illusion of all the events happening during the same
time period.'

World Graph: I don't think much is to be gained distinguishing games
where some connections are north/south and others are north/west. Where
would you classify a game where you moved between certain locations with
dream/wake? Or even go right/turn around then go forward? More
important, in my mind, is how clear the game is about how to travel is to
be accomplished. And, that's only really important for classic mazes,
which can be summed up much more succinctly with 'this game has a maze'.
So, I say drop the 'World Graph' terminology altogether.

Accessibility: I think this scale is more useful than you gave it credit
for. As someone else said, whether a point is accessible *in a finite
amount of turns* is still useful information. Instead of trying to
classify an entire work of IF as 'accessible' or not, though, what is more
helpful is to talk about whether one gamestate is accessible from another
gamestate, or if that avenue is or has been closed off.

----

Based on my musings above, here's another scale, based on things that
change the gamestate. I'm going to grab Emily's concept of 'benchmarks'
to talk about them, but I'm going to water it down and call them 'nodes'.
Let's say that benchmarks are game-and-plot state altering points, while
nodes are merely game state altering points.

So: something happens. Code is executed. Maybe it was a 'determinancy'
event (random); maybe it was a 'transiency' event (real-time), maybe it
was in response to player input. The following can happen:

1. The player remains at the same node
2. The player is moved to a new node from which all previously-accessible
nodes are still accessible.
3. The player is moved to a new node from which new nodes can be reached
4. The player is moved to a new node from which previously-accessible
nodes can no longer be reached
5. The player is moved to a new node from which previously-accessible
nodes can still be reached, but the *path* to them has changed.

(Note that these are not all mutually exclusive.)

I have no idea what to term these. Maybe just using the terms 'nodes',
'benchmarks', and 'accessibility' is sufficient.

Maybe we could talk about 'accessibility level' and 'accessibility state'?
The problem is that 'level' implies one linear scale, which isn't
realistic. What happens when both 3&4 happen? 'the accessibility map has
changed'? Oh, hey, I like it:

1&2: The player remains in place on the accessibility map
3: There are new regions on the accessibility map
4: There are regions that have closed off on the accessibility map
3 plus 4: The accessibility map has shifted
5: The player is moved on the accessibility map.

(is distinguishing between 1 and 2 helpful? I'm not entirely sure even I
know what I mean ;-)

OK, that's enough fodder for now ;-)

-Lucian

Nick Montfort

unread,
Dec 6, 2001, 12:53:39 PM12/6/01
to
"Gabe McKean" <gmc...@wsu.edu> wrote in message news:<9un234$fvb8$1...@murrow.it.wsu.edu>...

> I have a question regarding the "Determinability" variable: Several
> games use daemons to produce random 'background' text from time
> to time. However, this kind of behavior doesn't usually affect the world
> state. Since you seem to be concentrating on text-generation in your
> definitions, would you say that these games are indeterminate?

I don't mean to concentrate on text-generation, except that there seem
to be formal things one can say about that fairly usefully. The
answer, though, is that in this first draft I've taken
"determinability" directly from Aarseth's cybertext. I agree that it's
better to do what I did with "transiency" and distingusish
"text-indeterminate" and "world-indeterminate" IF.

> I think your typology is interesting, but I personally don't think an IF
> typology will be complete unless it takes the world state of a game into
> account on at least an equal footing to the text state.

You're right as I see it, becuase I think the world model (as opposed
to randomly writing out one of 30 strings) is what makes IF what it
is.

-nm

Sean T Barrett

unread,
Dec 6, 2001, 2:58:38 PM12/6/01
to
Nick Montfort <ni...@nickm.com> wrote:
>I offer seven independent variables for IF which
>each may have either two or three values:

It's a good try. I'm not clear on the utility of trying
to classify things into one of two states when it's
usually a continuum. Almost all of the variables suffer
from this problem; should the fact that there is one random
message printed somewhere shift a game between categories?
I suspect that perhaps in the vast majority of works
analzyed in other typologies, if a work does something
once it is likely to do it again, whereas if you have to
program, it's always a lot of work so you may only do it
once.

>Determinability (indeterminate, determinate)

Example given above.

>Transiency (intransient, world-transient, text-transient)

Too rare to make a useful distinction?

>User Functions (explorative, configurative, textonic)

Again, all or nothing from one thing. If a game lets me
name my dog, or rename all instances of the word "west"
to some other string, is that worth putting the whole
game into another category?

Also, how do we define configurative meaningfully
different from explorative? If you look at a transcript,
people configure the text by choosing two different moves
in order. If you define configuration as "changing the
text that appears between two prompts", then typing
"GET BALL.E.DROP BALL" at a prompt plays havoc with that.
Defining what "command" means to get around that is complex.

Also, at what level does text synthesis have to occur
to get "configurative"? If a Galatea-like character will
print:

>ASK GALACTUS ABOUT SILVER SURFER
Galactus frowns. "He was always, like, so rebellious,
so I'm glad I replaced him eons ago."

but the "frowns" is generated based on the mood you
put Galactus in the *previous* turn, and the text is
based on the command from this turn, is it now configurative?
Or if you require it to be within a single sentence,
make the above read "Galactus frowns and says, ...".

>End State (one, many)

We could toss 'zero' into here and go ahead and classify
things like Aisle which may not technically be IF under
most formal definitions. (I don't believe formal definitions
are that useful; indeed it may be more meaningful to say
'the more a work falls into one of the IF-specific values
for these seven variables, the more that work is IF'.

One problem with this variable is nobody but the author
can know for sure whether a game that appears to only
have one ending actually has more.

>Player Characters (one, many)

As with the above, we could toss 'zero' in as an option.

>Non-Player Characters (no, yes)

We could say 'none', 'npcs', 'actors' as the values of this.
Having orderable-actors is pretty rare in IF I think.
Then again, having an IF game entirely free of NPCs in the
traditional sense is also exceedingly rare, and I recognize
the problem of distinguishing NPCs from plain old objects
(consider, e.g., the 'toolman' and 'songlantern' in For a Change).

>World Graph (regular, complex)

Ugh. Shouldn't a one-room game with no enterables be "none" under
this classification?

Probably the notion of global reachability is more interesting;
is a room always reachable? (You might have to solve puzzles to
do so.) If rooms become locked away, then they're not reachable.

One can simply apply this to the entire text of the game, rather
than distinguish 'rooms'; but most games feature non-reversible
actions--for example, you can never get back the 'initial' text
for an object once it's moved--so maybe not.

It seems like you've tried to focus on properties that are
entirely part of the interactive component of IF, which I think
is good, but I wonder if there aren't a few interesting qualities
of just the text itself. For example, an awful lot of games are
essentially continuous--there are no narrative leaps across time,
nor flashbacks etc. That this is true of IF more than of static
fiction is *because* the works are interactive and authors find
that good in this form; thus it would probably not be that
interesting a property for static fiction, but it might be for IF,
despite being purely a property of the text. I'm not sure how
you can define it coherently, though; how much time can pass
in response to a command for it to still be continuous? If I
type "WAIT 10 HOURS" in Deadline, it's actually still continuous,
in fact the game simulates all that intervening time, and things
can happen and I can interrupt.

SeanB

Robotboy8

unread,
Dec 6, 2001, 10:32:52 PM12/6/01
to
It'll be interesting to see if the general community comes up with a system of
our own to classify such games, a general standard which would be applied as a
graph to different games. Then these graphs could be displayed (CGI? Perl? I
have no clue) on brasslantern or textfire. Anyone care to write a script to
generate such tables from user-made data?
<sees all the shaking heads>
Let's make this more interesting. Write a utility in Inform or Tads to do this
and submit it to the NoFunComp.
I hope suggestions for un-games are allowed in this "comp".

--
Sanity is a sure sign of a lazy mind.

Nick Montfort

unread,
Dec 6, 2001, 11:53:11 PM12/6/01
to
buz...@TheWorld.com (Sean T Barrett) wrote in message news:<Gnxu5...@world.std.com>...

> It's a good try. I'm not clear on the utility of trying
> to classify things into one of two states when it's
> usually a continuum.

Well, there are plenty of reasons, and I hope to show some results
once I complete the analysis here. Until then this may just be
hand-waving. Anyway: one reason is that it helps to describe what
categories are actually better seen as discrete when you might
otherwise assume they're best seen as a spectrum. Maybe it points out
where cardinality is more important than number. So for instance it's
true that you can have n player characters, where n is any positive
integer, but actually considering the case where you have one versus
the case where you have many seems interesting.

> Almost all of the variables suffer
> from this problem; should the fact that there is one random
> message printed somewhere shift a game between categories?

Yes. If I have "sets that contain only integers" in my left hand and
"all other sets of numbers" in my right, adding ".2" to a set in my
left hand means that it has to move to the right. It doesn't matter if
the set has 5,000 integers and only one number that isn't an integer;
it's still not all integers.

> I suspect that perhaps in the vast majority of works
> analzyed in other typologies, if a work does something
> once it is likely to do it again, whereas if you have to
> program, it's always a lot of work so you may only do it
> once.

This isn't about likelihood or about how much work it is to do
something. It's a formal description of the properties of different IF
works. It may very well be that what you mention is the only reason
it's useful, but the point is to figure out what things can be said
with certainty.

> If you define configuration as "changing the
> text that appears between two prompts", then typing
> "GET BALL.E.DROP BALL" at a prompt plays havoc with that.
> Defining what "command" means to get around that is complex.

Ah, that's right! Good catch. Yes, I'll have to say that any IF that
accepts multiple commands on the same line is at least configurative
(maybe textonic). That makes explorative IF pretty rare (Dark Mage)
but it may still be worthwhile to consider -- unless there's another
sort of formal distinction to be made here with reference to this
variable. I'll ponder.

I consider the Galactus example definitely configurative, by the way.

> >End State (one, many)
> We could toss 'zero' into here and go ahead and classify
> things like Aisle which may not technically be IF under
> most formal definitions.

Aisle's problem isn't only the lack of end state (you could say each
of its replies are an end state, although maybe not ...) but perhaps
also the lack of world model, the way I see it.

> One problem with this variable is nobody but the author
> can know for sure whether a game that appears to only
> have one ending actually has more.

That just means that after careful analysis, you might still come up
with the wrong answer in classifying something. That doesn't mean that
this isn't a formal property of the computer program itself; it still
is. Whether or not there are multiple end states is encoded in the
program, not the author.

> >World Graph (regular, complex)
> Ugh. Shouldn't a one-room game with no enterables be "none" under
> this classification?

I call it regular because you can get back to that room by reversing
your direction from every one of the (zero) exits.

> Probably the notion of global reachability is more interesting;
> is a room always reachable? (You might have to solve puzzles to
> do so.) If rooms become locked away, then they're not reachable.

As you say, ugh. What you suggest is more interesting, and I think
I'll use it in the next draft.

> One can simply apply this to the entire text of the game, rather
> than distinguish 'rooms'; but most games feature non-reversible
> actions--for example, you can never get back the 'initial' text
> for an object once it's moved--so maybe not.

Actually, the problem is that we may not be able to formalize whether
something is a 'room' or not (which would mess up my proposed category
pretty badly!) so considering the entire text may be better.

> I wonder if there aren't a few interesting qualities
> of just the text itself.

Yes, certainly! (Rest of very insightful comment below:)

> For example, an awful lot of games are
> essentially continuous--there are no narrative leaps across time,
> nor flashbacks etc. That this is true of IF more than of static
> fiction is *because* the works are interactive and authors find
> that good in this form; thus it would probably not be that
> interesting a property for static fiction, but it might be for IF,
> despite being purely a property of the text. I'm not sure how
> you can define it coherently, though; how much time can pass
> in response to a command for it to still be continuous? If I
> type "WAIT 10 HOURS" in Deadline, it's actually still continuous,
> in fact the game simulates all that intervening time, and things
> can happen and I can interrupt.

Good points. But you can apply narratology directly to the possible
narratives that a work of IF generates, for instance looking to Gerard
Genette's Narrative Discourse and his discussion of ellipsis,
paralipsis, etc. Then you still have the theorize how these narratives
get produced and what this person is doing typing things in. It's the
workings of IF as a text-accepting, text-producing computer program
that aren't as easily understood with previous theories.

-nm

Nick Montfort

unread,
Dec 6, 2001, 11:54:18 PM12/6/01
to
"Lucian P. Smith" <lps...@rice.edu> wrote in message news:<9uoap1$7p3$1...@joe.rice.edu>...

> I want scales (and I probably would
> prefer scales to binary states) that will let me, as the
> player/interactor, know something about how I will be interacting with
> the work.

Personally, in this piece, I'm not trying to create a typology based
on the player's reaction, but based on formal properties of the works.
These two are realted, of course, but not the same.

> I want to know where a work falls in terms of its gameplay.

That's assuming it's a game; I don't think all IF is. How does
Exhibition rate in terms of gameplay?

> I want
> to know what kind of thinking I'll have to do while engaged with the work.

Again, this isn't something I'm attempting to cover; reading different
print novels that are formally the same type of text, for instance,
requires different types of thinking. But it's an important topic and
could be a useful standpoint from which to construct a typology: Feel
free.

> Was that Espen's goal?

To describe formal independent properties of text machines, as far as
I know, to create a framework for strong theoretical work; you'd only
have to read the first 75 pages of his book (through the end of
chapter 4, Textonomy) to hear how he puts it.

> Determinability: Text changing randomly is one thing, the gamestate
> changing randomly is another, and the gamestate changing randomly
> forcing the player towards different end states is yet another.

Yes, the second and third are both lumped together in
world-indeterministic (once that's added). However my variables would
not be independent if I added the last category as something separate
because some works only have one end state.

> I'm not exactly
> sure how to formalize this; perhaps I'll just give some examples and we
> can try to formalize them:
>
> -Games where nothing is random
> -Games where 'flavor text' changes (I think it still could be useful to
> distinguish this case from the above, since it can affect mimesis, esp.
> for replays of the game)

I agree, without using the m-word, that works with this type of text
are in an important category.

> -Games with varying 'flavor text' that require the player to change their
> input to match. In 'Edifice', you can slide off a horse to the right or
> to the left, and getting back on requires taking that into account.

Yes, becuase the state of the world has changed -- which side of the
horse the player is on.

> -Games that randomly change the entire game state, requiring different
> swaths of input from the player. The bat carrying you off to a random
> location in Zork Zero, maybe, or the randomizer in 'Guess the Verb!'.

I don't know how to formally distinguish this from the case above; I
believe it's a matter of the extent of the change to the world rather
than something in a different category.

> -Games that randomly change the game state that take control of the
> game away from the player so much that you reach an end state
> because of it. If the troll kills you in Zork I, the fight in 'A Crimson Spring',
> the bees dying in 'Adventureland'.

All sessions can reach some end state, so this must be the case where
randomness changes the end state you can reach. Again, my variables
wouldn't be independent if I included this category, since some IF has
only one end state. If you're saying "randomness can bring you to an
end state in in the next move," that's independent and I may include
that.

> Your 'world-transient' vs. 'text-transient' gets at the reader-response, I
> think, but I'm starting to think the response should have it's *own*
> formality. Hmm.

None of my taxonomy has anything to do at all with reader-response. It
is all based on formal properties of IF works; how people interpret
the resulting text is something else entirely.

In Mindwheel, text appears on the screen if you do nothing. This is
text-transient.

In Border Zone, if you type a reply after waiting one minute a
different text will be produced than if you typed the same reply after
30 minutes. This is world-transient.

Notice I have said nothing about the interpretation of any text here;
such interpretation should indeed be considered separately.

> User Functions: First off, I'd argue that some IF can be 'interpretive'
> in parts--this is what people are complaining about when they talk about

> 'text dumps'. 'Ramses...

The category as Espen has it is for works that are *only*
interpretative, in which the reader has no other role. The player has
to type in order to 'operate' or play Ramses, so it can't be
interpretive.

> Explorative vs. configurable are indeed useful terms--they're what we
> mean when we talk about exploration games (or parts of games) or
> puzzle-based games.

They're not that, but rather distinguish works where your input
configures the texts that are shown and work where your input selects
between existing fixed texts. Similarly 'textonic' is different
formally and represented in many works of IF where the player can name
something and therefore write part of a text that is displayed.

> *all* IF is textonic because there are usually many many instances
> of the game repeating the verb you just used back to you. JUMP / You
> jump on the spot, fruitlessly.

"JUMP" is an input which selects the fixed text "You jump on the spot,
fruitlessly." which was stored in the machine in that form beforehand.
An IF which only offered replied like this would be explorative. There
are other standard library replies that may make works textonic,
though, so you're right that many things I listed as out of this
category may be in it.

> Player Characters: I think it might be helpful to talk about parallel
> player character change vs. linear player character change. By which I
> mean, basically, Suspended vs. Photopia. Common Ground is an
> interesting midpoint on that scale, too.

Good point -- it's one I'll have to think on to see how to precisely
make the distinction. Of course, multiple-PC IF of any sort is fairly
rare, so if this taxonomy didn't capture that nuance it could still be
valuable. But I do believe in making whatever strong formal
distinctions I can, even if there are only a few works that happen to
represent these categories now.

> I'm not exactly sure what you
> want it to mean, either, having not read the Lebling paper.

I'll try to summarize and perhaps formalize the relevent part in the
next draft.

> Where
> would you classify a game where you moved between certain locations
> with dream/wake?

Very good point -- some 'directions' or ways of moving between places
don't have opposites. This 'World-Graph' notion need some work, and
I'm looking into it.

> More imporant ... is how clear the game is about how to travel is to
> be accomplished.

Those games that are written entirely in German are very tough for me
to get around in. I think this may be an interpretive rather than
formal matter.

> Accessibility: I think this scale is more useful than you gave it credit
> for. As someone else said, whether a point is accessible *in a finite
> amount of turns* is still useful information. Instead of trying to
> classify an entire work of IF as 'accessible' or not, though, what is more
> helpful is to talk about whether one gamestate is accessible from
> another gamestate, or if that avenue is or has been closed off.

Yes, this would be a worthwhile addition I think.

> Based on my musings above, here's another scale, based on things
> that change the gamestate.

This is a proposal for a typology of events -- what can happen when
code runs -- not for works of IF, yes?

> I'm going to grab Emily's concept of
> 'benchmarks' to talk about them, but I'm going to water it down and call
> them 'nodes'. Let's say that benchmarks are game-and-plot state
> altering points, while nodes are merely game state altering points.

I thought every turn is a 'gamestate'-altering point, usually, since
the number of moves is incremented and that is part of the world's
state? I don't see yet that 'benchmark' describes a formal property of
IF and am not sure how 'node' is meant here, so I can't comment in
detail. I do think that enumerating the types of things that can
happen in between inputs is very interesting and a good direction, and
this may be a good start to that, but I don't understand it well
enough yet to classify things and see how it works.

Thanks very much both to you and to Sean, to whom I just replied, for
the thoughtful responses.

-nm

OKB -- not okblacke

unread,
Dec 7, 2001, 12:55:36 AM12/7/01
to
ni...@nickm.com (Nick Montfort) wrote:
>> Almost all of the variables suffer
>> from this problem; should the fact that there is one random
>> message printed somewhere shift a game between categories?
>
>Yes. If I have "sets that contain only integers" in my left hand and
>"all other sets of numbers" in my right, adding ".2" to a set in my
>left hand means that it has to move to the right. It doesn't matter if
>the set has 5,000 integers and only one number that isn't an integer;
>it's still not all integers.

I don't see that this applies to IF, though. Mathematics is by its nature
a rigidly-defined system where this kind of categorizing makes sense, but I
don't think IF is. I think your system of variables is interesting and useful
in that it can help guide one's thinking about IF, but I don't see that any
extra usefulness comes from trying to cookie-cut IF to fit within the slots
you've created.

In other words, while it may be true that games with one random message
are in a different category than games with no such randomness, it may well
also be true that such a distinction is not meaningful in IF. You've said that
you're trying to just develop a descriptive formal system to categorize IF
according to certain variables, and that's fine, but I think it's important
(for everyone, including me) to remember that just because you write this stuff
down and it has a theoretical meaning doesn't mean (and I hope I don't sound
too blunt here) we should care. Some of these distinctions are bound to be
much more "useful" to actual writing/playing/thinking than others.

--OKB (Bren...@aol.com) -- no relation to okblacke

"Do not follow where the path may lead;
go, instead, where there is no path, and leave a trail."
--Author Unknown

Magnus Olsson

unread,
Dec 7, 2001, 6:29:34 AM12/7/01
to
In article <ae605ba7.01120...@posting.google.com>,

Nick Montfort <ni...@nickm.com> wrote:
>"Gabe McKean" <gmc...@wsu.edu> wrote in message
>news:<9un234$fvb8$1...@murrow.it.wsu.edu>...
>> I have a question regarding the "Determinability" variable: Several
>> games use daemons to produce random 'background' text from time
>> to time. However, this kind of behavior doesn't usually affect the world
>> state. Since you seem to be concentrating on text-generation in your
>> definitions, would you say that these games are indeterminate?
>
>I don't mean to concentrate on text-generation, except that there seem
>to be formal things one can say about that fairly usefully.

(...)

>> I think your typology is interesting, but I personally don't think an IF
>> typology will be complete unless it takes the world state of a game into
>> account on at least an equal footing to the text state.
>
>You're right as I see it, becuase I think the world model (as opposed
>to randomly writing out one of 30 strings) is what makes IF what it
>is.

My spontaneous reaction - which I'm not sure is entirely relevant
since I haven't had the time to read your article yet - is that the
literary theorists who concentrate on IF as text generation are
missing something crucial.

Perhaps this is natural if you come from the hypertext direction,
since in hypertext, there's much more of a one-to-one relation between
text and narrative than in Infocom-style IF, where it is possible
to generate huge amounts of text with essentially no bearing on the
narrative (parser error messages, repeated descriptions, etc).

Of course, viewed in the context of IF as a construct by the player,
that is, the player's subjective experience, text is extremely
important; a trivial example would be that a changed parser message
may change the perception of what's going on.

But to me, IF is more of a story generator than a text generator.

--
Magnus Olsson (m...@df.lth.se, m...@pobox.com)
------ http://www.pobox.com/~mol ------

Magnus Olsson

unread,
Dec 7, 2001, 6:35:15 AM12/7/01
to
In article <ae605ba7.01120...@posting.google.com>,
Nick Montfort <ni...@nickm.com> wrote:
>I do think it's important to
>distinguish an NPC from the mention of a character in text, whether by
>reference to actors or not, and I don't think the criterion of
>"meaningul interaction" is rigorous enough to do this unless that can
>be a formal term somehow.

A thought (if it is kosher to discuss one's own games): Is Uncle Zebulon
in _Uncle Zebulon's Will_ an NPC? I'd say not, since he never appears
in the game, yet reviewers have commented on how he influences
the whole game in his absence. And that's of course what I intended.

So I think it's meaningful to distingusih between NPCs - which are
actors - and characters which are part of the narrative, like
characters in static fiction - but not part of the simulation. It
strikes me that almost all discussion of Infocom-type IF focuses
exlusively on NPCs in the actor sense, but perhaps that's natural,
because few works seem to have characters that are not NPCs. The only
one I can think of right now is the Master in _Metamorphoses_, who
also has a strong presence in the game despite not being an NPC.

Magnus Olsson

unread,
Dec 7, 2001, 7:11:35 AM12/7/01
to
In article <20011207005536...@mb-fi.aol.com>,

OKB -- not okblacke <bren...@aol.comRemove> wrote:
>ni...@nickm.com (Nick Montfort) wrote:
>>> Almost all of the variables suffer
>>> from this problem; should the fact that there is one random
>>> message printed somewhere shift a game between categories?
>>
>>Yes. If I have "sets that contain only integers" in my left hand and
>>"all other sets of numbers" in my right, adding ".2" to a set in my
>>left hand means that it has to move to the right. It doesn't matter if
>>the set has 5,000 integers and only one number that isn't an integer;
>>it's still not all integers.
>
> I don't see that this applies to IF, though. Mathematics is by its nature
>a rigidly-defined system where this kind of categorizing makes sense, but I
>don't think IF is. I think your system of variables is interesting and useful
>in that it can help guide one's thinking about IF, but I don't see that any
>extra usefulness comes from trying to cookie-cut IF to fit within the slots
>you've created.
>
> In other words, while it may be true that games with one random message
>are in a different category than games with no such randomness, it may well
>also be true that such a distinction is not meaningful in IF.

The number one question for any taxonomist must be: is the distinction
meaningful? Does anybody care?

Does a game change in any important way if it contains one random
message? Obviously, the answer depends on what that message is.

Shouldn't the distinction be made between games where random messages
are in some way important, and those that either have no random
messages or only unimportant ones?

Alexandre Owen Muniz

unread,
Dec 7, 2001, 10:20:19 AM12/7/01
to
Mild spoilers for Rematch below

Nick Montfort wrote:
>

> What I've tried to write about are "actors" in the sense discussed in
> the Zork IEEE paper. Adventure has none, although it has a dwarf that
> can kill the PC and a pirate that can steal treasure. I don't think
> the concept (which seems like it can be made formal) is useless,
> although it may not be the best one. I do think it's important to
> distinguish an NPC from the mention of a character in text, whether by
> reference to actors or not, and I don't think the criterion of
> "meaningul interaction" is rigorous enough to do this unless that can
> be a formal term somehow.
>
> Perhaps you or others would provide a formal definition of an NPC that
> you find useful, and explain if things like the dragon, snake, and
> bird in Adventure; the hollow voice in Zork; the tiny figure in Shade;
> the Wizard of Wordplay in Ad Verbum and the parrot in Christminster
> are NPCs by this definition? These definitions could also be used to
> determine which of the people mentioned in Photopia are NPCs and which
> aren't. Presumably "the girls" mentioned as waiting for the two
> partygoers in the first segment are not NPCs by anyone's thinking, for
> instance, although they motivate the situation.

I see this as an area where only informal definitions with fuzzy boundaries are useful. As
a starting point for an informal definition: "An NPC is a fictional sentient being that
may interact with the PC during a game session." The result of applying this definition is
generally 'sort of' for the examples you gave, which seems right.


> On a side note, I'm surprised that no one has taken me to task for
> excluding Aisle from IF and including Rematch. I think the two have to
> be included or excluded together, no?
>

No, and that touches on the matter of player knowledge, external to a single session of a
game. A player's experience of a game divides pretty cleanly into two entities. One is the
game as experienced by the pc(s), in a single sequence of steps from beginning to end,
(where it may take multiple sessions to construct such a sequence.) The other is the
metagame as experienced by the player over multiple sessions, where turn sequences in
different sessions may repeat or diverge, and player knowlege from previous sessions
becomes a part of the metagame state.

In Rematch all actions are possible at all points, but they are not necessarily
*conceivable*. Rematch maps the entire structure of the game into the domain of external
knowledge, giving a metagame with a structure very like a traditional (non-meta-)game with
areas that are inaccessable at the beginning that become accessable and are passed through
to arrive at a single end. (And since the pc supposedly experiences the multiple sessions
in succession, the meta-game is not meta- at all in a sense, but I digress.) Aisle[1] does
not use external knowledge to give structure to its metagame.

The balance in importance in a player's mind between 'game as canonical pc experience' and
'game as metagame' varies from game to game and even player to player; I can't formalise
the concept, so it's probably useless to you.
Rematch is an extreme case, but when I think of games like I-0[2] with frequent death and
suboptimal endings, I remember the catalogue of failures much better than I remember the
sequence of successful moves. So Far[3], by contrast is a game where the suboptimal
endings felt like aberrations to me, and what "really happened" is the sequence of turns
where the pc did the right things. I guess the difference has as much to do with the care
the authors devote to the writing in the suboptimal endings, relative to other game text,
as much or more than the frequency of suboptimal endings.

Oh, but I see in another post that you are entirely uninterested in the player's reaction
to the game. Never mind then.

**Owen

[1] Haven't played, but know well enough from reputation.
[2] Haven't finished. Will probably get back to some day.
[3] Haven't finished. Yada.

M. D. Krauss

unread,
Dec 7, 2001, 11:27:54 AM12/7/01
to
On 7 Dec 2001 11:35:15 GMT
m...@df.lth.se (Magnus Olsson) wrote:

I'm not to comfortable with that terminology. Maybe I'm just nit-picking,
but NPC stands for "Non-Player Character". Characters which are only in
narrative or referred to, but not coded or interactive, are still
characters and they are still not player characters, so semantically
speaking, they should be considered NPCs. Maybe coin new terms -- "INPC"
for "Interactive Non-Player Character" and "NNPC" for "Narrative
Non-Player Character"?

Not that I'm fond of acronyming everyone to death. Personally, I'd rather
say "character" and "actor" or some such.

-M
(Acting out of Character? Characteristic Actions?

You are in an open area at the center of town. Some
people call it the "Town Square" but to you it's the
TS. There is an innocent Everyone grazing on some
wild diction shrubs across the way.

> i

You are carrying an ACME AB (Acronym Blaster) in a
holster around your waist.

> draw ab

You pull the AB out of it's holster.

> aim ab at Everyone

As you take aim, the Everyone looks up at you with
sad eyes. It seems to be able to sense something is
wrong.

> fire ab

IFRTFMTTFNCYAIMHOWTFNPCGNUFSF! The poor Everyone lies twitching
helpless on the floor.

Now that's out of character.)

>
> --
> Magnus Olsson (m...@df.lth.se, m...@pobox.com)
> ------ http://www.pobox.com/~mol ------


--
To email me convert my address to something resembling reason

Paul O'Brian

unread,
Dec 7, 2001, 12:52:19 PM12/7/01
to
On 5 Dec 2001, Nick Montfort wrote:

> Player Characters (one, many)

Where would LASH fall in this categorization? In a sense, it has no PC. In
another sense, the player is the PC. In another sense, Linda is the MULE's
PC, though the relationship is analogous, not identical, to the player's
relationship with the MULE. In another sense, Linda and the MULE are the
PCs.

Hmmm. Between all those senses, I'm not sure I'm making sense anymore.

--
Paul O'Brian obr...@colorado.edu http://ucsu.colorado.edu/~obrian

Lucian P. Smith

unread,
Dec 7, 2001, 2:17:23 PM12/7/01
to
: "Lucian P. Smith" <lps...@rice.edu> wrote in message news:<9uoap1$7p3$1...@joe.rice.edu>...

:> I want scales (and I probably would
:> prefer scales to binary states) that will let me, as the
:> player/interactor, know something about how I will be interacting with
:> the work.

: Personally, in this piece, I'm not trying to create a typology based
: on the player's reaction, but based on formal properties of the works.
: These two are realted, of course, but not the same.

Hmm. I guess I was unclear.

I didn't mean that typologizing IF should take into account the internal
state of the reader. I meant that I'm going to judge the *usefulness* of
any of these scales based on how a reader might internalize them and
obtain insights into the work in question.

I can dream up a million different useless ways to categorize IF that
would satisfy all of your listed criteria. IF that has more A's than O's,
for example, or IF that prints some text in a proportional font vs. fixed
width. Who cares? You might be able to pick up 8 games and classify 6 as
A and 2 as B, but if it doesn't affect the interactivity, nobody is going
to ever use the scales for anything.

This is also why I see greater usefulness in formalizing scales than
binary states. In an actual conversation or write-up about a game, nobody
is going to say 'This game is Determined' or 'This game is Indetermined'.
They're going to use qualifiers, and say, "This game is fairly Determined"
or "This game is somewhat Indetermined".

We're obviously approaching this from two different directions; maybe in
the academic world more use is made of these binary states. Could you
give us examples, maybe?

:> I want to know where a work falls in terms of its gameplay.

: That's assuming it's a game; I don't think all IF is. How does
: Exhibition rate in terms of gameplay?

I was using 'gameplay' in a more formalized sense, and I'm not sure
exactly how to define it. Perhaps, "Gameplay is the banter between the
reader and the work; the rhythm of give-and-take between player input and
game output; the rate at which information is conveyed to the player,
internalized, and shows up again in player output."

In that sense, I could indeed talk to you about Exhibition's 'gameplay'.

:> Determinability: Text changing randomly is one thing, the gamestate

:> changing randomly is another, and the gamestate changing randomly
:> forcing the player towards different end states is yet another.

: Yes, the second and third are both lumped together in
: world-indeterministic (once that's added). However my variables would
: not be independent if I added the last category as something separate
: because some works only have one end state.

Our biases are showing again--you want to classify works of IF as a whole,
and I want some terminology that enables me to talk about IF more clearly,
which may include talking about the whole work, or may include just
talking about one section of the work.

: I don't know how to formally distinguish this from the case above; I


: believe it's a matter of the extent of the change to the world rather
: than something in a different category.

Yeah, I think these changes are better formalized in their own category,
since different things can cause them.

:> Your 'world-transient' vs. 'text-transient' gets at the reader-response, I

:> think, but I'm starting to think the response should have it's *own*
:> formality. Hmm.

: None of my taxonomy has anything to do at all with reader-response. It
: is all based on formal properties of IF works; how people interpret
: the resulting text is something else entirely.

Again, I was judging their usefulness.

:> User Functions: First off, I'd argue that some IF can be 'interpretive'

:> in parts--this is what people are complaining about when they talk about
:> 'text dumps'. 'Ramses...

: The category as Espen has it is for works that are *only*
: interpretative, in which the reader has no other role. The player has
: to type in order to 'operate' or play Ramses, so it can't be
: interpretive.

If this were a scale, it would be more useful, because you could then
indeed make claims that, say, 'Ramses' was 'virtually interpretive' and
people would know what you meant. Or that Ramses worked best for you on
an interpretive level, while the explorative aspects failed miserably (not
that I've actually played Ramses; this just strikes me as something
someone might say based on what I've read).

:> *all* IF is textonic because there are usually many many instances

:> of the game repeating the verb you just used back to you. JUMP / You
:> jump on the spot, fruitlessly.

: "JUMP" is an input which selects the fixed text "You jump on the spot,
: fruitlessly." which was stored in the machine in that form beforehand.
: An IF which only offered replied like this would be explorative.

This is getting to be absurdly mechanistic, to my thinking. Does it
really matter what is going on in the innards of the program? If I code
up the library response to look at the input buffer, copy that input to a
string ("jump"), then print the pre-stored string "You", the user-inputted
"jump", then the pre-stored, "on the spot, fruitlessly", this is
completely invisible to anyone interacting with the work. More important
is that the game accepts player input, then responds to that input using
the concepts encoded in it. That might include using the exact word
typed. Maybe it capitalizes it and uses it at the beginning of the
sentence. Maybe it uses a synonym (hit/attack). Maybe it doesn't use the
word at all but lets you see the aftereffects of the concept taking place
in the game world (>THROW BALL "The window shatters!") This is the heart
of what IF does, and trying to classify bits in terms of the inner
workings of the program and the raw letters and strings used is rather...
unhelpful.


:> Based on my musings above, here's another scale, based on things
:> that change the gamestate.

: This is a proposal for a typology of events -- what can happen when
: code runs -- not for works of IF, yes?

Right.

:> I'm going to grab Emily's concept of


:> 'benchmarks' to talk about them, but I'm going to water it down and call
:> them 'nodes'. Let's say that benchmarks are game-and-plot state
:> altering points, while nodes are merely game state altering points.

: I thought every turn is a 'gamestate'-altering point, usually, since
: the number of moves is incremented and that is part of the world's
: state? I don't see yet that 'benchmark' describes a formal property of
: IF and am not sure how 'node' is meant here, so I can't comment in
: detail.

Would it help to know I'm thinking of DAG's (Directed Acyclic Graphs) in
my head when I'm talking about these things? They're talked about in
http://www.xyzzynews.com//xyzzy.6g.html, for example. The concepts are
still fuzzy, but I think they're important. We'll get them nailed down
eventually.


-Lucian

Dennis G. Jerz

unread,
Dec 7, 2001, 3:59:23 PM12/7/01
to
"M. D. Krauss" <MDKraus...@home-nospam.com-nospam> wrote in message
news:20011207112754.02140d5

> I'm not to comfortable with that terminology. Maybe I'm just nit-picking,
> but NPC stands for "Non-Player Character". Characters which are only in
> narrative or referred to, but not coded or interactive, are still
> characters and they are still not player characters, so semantically
> speaking, they should be considered NPCs. Maybe coin new terms -- "INPC"
> for "Interactive Non-Player Character" and "NNPC" for "Narrative
> Non-Player Character"?

I don't think it's nit-picking... yet. PC vs. NPC is certainly not
sufficient to address the issue we're discussing here.

M.D. Kraus correctly points out that being an NPC is only one way to be "a
character who is not the player." Consider Photopia, in which Ally is never
the player; yet in some chapters she is the narrator (when she tells Wendy
the interactive bedtime stories). That's a different way of being "not a
PC", and it's definitely narrative, yet it's not what I think M.D. meant by
an NNPC.

Rather than NNPC, why not just NC (for "narrative character"). This could
describe a character who exists in the game world because somehow, the game
world has told you about it. This might be said to include a bellhop who
appears, delivers a telegram, and disappears in one turn -- the character
seems to have affected the game world (the telegram's there, after all), but
it's only the narration of the event that created the existence of the
bellhop in the player's mind. This brief reference to the bellhop might
plant in the player's mind the idea that would be a good idea to find a
bellhop later. The bellhop's phantom presence, at an early stage of the
text, before the player has learned whether it's possible to meet the
bellhop and talk to him.

A minor example would be the girl who seems to have lost her sketchbook in
the Jigsaw pre-game; a major example would be the master in Metamporphoses.

One might still write functions or routines that describe the bellhop's
behavior, and have that behavior influenced by the PC's actions, of course,
so "is there code attached to the character" is probably not the crucial
question.

There's a tradition in drama of using a phantom actor, George Spelvin, on
the cast list, to avoid spoilers for an audience member who reads his or her
program too closely. For example, it might be established that the *issing
Bandit is one of the people in the cast, but the bandit is masked whenever
his/her nasty deeds occur onstage. (I suppose that if you did stage such an
incident, the audience would have a pretty good idea of the gender of the
culprit, but I digress.)

So.. is the key thing here that an NPC is somebody with whom the PC can
interact; thus, a character that does not exist as an object to be examined,
touched, listened to, etc., is a NIC (non-interative chararacter).

Anybody want to pick up the baton from here?

--
Dennis G. Jerz, Ph.D.; (715)836-2431
Dept. of English; U Wisc.-Eau Claire
419 Hibbard, Eau Claire, WI 54702
------------------------------------
Literacy Weblog: www.uwec.edu/jerzdg

Sean T Barrett

unread,
Dec 7, 2001, 9:14:26 PM12/7/01
to
Nick Montfort <ni...@nickm.com> wrote:
>buz...@TheWorld.com (Sean T Barrett) wrote in message
>> One can simply apply this to the entire text of the game, rather
>> than distinguish 'rooms'; but most games feature non-reversible
>> actions--for example, you can never get back the 'initial' text
>> for an object once it's moved--so maybe not.
>
>Actually, the problem is that we may not be able to formalize whether
>something is a 'room' or not (which would mess up my proposed category
>pretty badly!) so considering the entire text may be better.

Yeah, this was one of the reasons I mentioned it (and put rooms in
quotes), but consider this:

If you're not allowed to type RESTART, the prologue is almost
certainly unreachable in 99.99% of games. If you are allowed
to type RESTART, all text is always reachable (except in the
very rare games that intercept it, two that I know of).

SeanB

Nick Montfort

unread,
Dec 7, 2001, 9:30:31 PM12/7/01
to
Thanks very much to everyone who has posted to this thread so far. I
very much appreciate the replies and the insights in them. I intend to
address the relevant points -- all of them I can possibly address --
as I revise and complete the essay on IF taxonomy that I posted.

That's where my real and best reply will come, but I'd like to mention
a few matters related to the thread before I turn to work on that. As
I've pointed out in email already, what I have offered here in draft
form is a taxonomy, not -the- taxonomy. Categorizing IF is not the
only useful theoretical activity. Certainly, looking at other things
besides formal properties of the text-producing program -- including
the interactor's reaction and qualities of narrative -- is important
for understanding why we like IF and what makes great works of IF
great. When I said "I'm not interested in that here" on the thread
it's just that I'm trying to limit my own scope as I work on this
specific theory. I do care about other perspectives.

I can't represent the academic world and its position on IF, and I
don't write hypertext theory. Said theory bears no relation to my
typology. Except for myself and Dennis, I don't see that the academic
world cares much about IF; net yet, anyway. Perhaps it's seen as an
amusing old novelty or as one specific example of new media or games,
but I don't think we have to be worried that our theory must meet some
existing expectation about what IF theory should be.

Finally, as far as I can tell the replies here have involved rigorous
new discussion of what a PC is, what an NPC is, whether what the
computer program is doing internally matters, what the nature of the
map and rooms are, how the space of possible texts compares to the
world, and whether there is a formal difference between two "one turn"
IF works -- among other things. (Some of these topics I may still
reply to on this thread, to discuss that issue that's been raised
rather than to talk specifically about revising the taxonomy. That's
if I can think of anything worthwhile to say.) Such discussion --
rather than a new indexing scheme for the IF archive -- is what this
theoretical taxonomy was for; that's why I proposed it. So, whether or
not my own results are worthwhile or indeed are ever forthcoming, I am
sorry that I cannot agree with those who think these categories
useless.

Thanks again for the discussion so far.

-nm

Sean T Barrett

unread,
Dec 8, 2001, 1:20:29 AM12/8/01
to
Alexandre Owen Muniz <mun...@xprt.net> wrote:
>In Rematch all actions are possible at all points, but they are not necessarily
>*conceivable*. Rematch maps the entire structure of the game into the
>domain of external
>knowledge, giving a metagame with a structure very like a traditional
>(non-meta-)game with
>areas that are inaccessable at the beginning that become accessable and
>are passed through
>to arrive at a single end. (And since the pc supposedly experiences the
>multiple sessions
>in succession, the meta-game is not meta- at all in a sense, but I
>digress.) Aisle[1] does
>not use external knowledge to give structure to its metagame.

There are topics that you can think about in Aisle which are
not revealed to you during the opening text. Hence Aisle
already has this behavior.

One obvious concrete difference is that Rematch actually implements
actions that affect the world state, then brings the game to a close,
whereas Aisle simply brings the game to a close; Rematch synthesizes
text in response to commands whereas Aisle simply reproduces previously
chosen texts; or, Aisle is exploratory and Rematch is
configurawhatevery.

Another diference is that Rematch drops you to an 'end of the game'
metagame prompt, allows you to UNDO, and allows you to automate the
UNDO so the metagame is never visible; Aisle dispenses with metagame
entirely and automatically returns to the original scene--metafictionally
but not metagamey?

SeanB

Sean T Barrett

unread,
Dec 8, 2001, 1:26:28 AM12/8/01
to
M. D. Krauss <MDKraus...@home-nospam.com-nospam> wrote:
>I'm not to comfortable with that terminology. Maybe I'm just nit-picking,
>but NPC stands for "Non-Player Character". Characters which are only in
>narrative or referred to, but not coded or interactive, are still
>characters and they are still not player characters, so semantically
>speaking, they should be considered NPCs.

As far as I know, "NPC" is a jargon term from pencil-and-paper
RPGs, where it meant "a character portrayed by the gamemaster"--
that is, a character you could interact directly with (talk to,
rather than fight, generally), and would never have meant a
character operating behind the scenes but referred to frequently.

Given that that is what the term already meant before IF borrowed
it and given that that is how the term is used in the IF community
now, I don't see it as particularly useful to change the meaning
of the term.

Trying to read out the meaning of a jargon term by examining the
semantics of the word making it up does not generally work, and
defeats the purpose of having jargon--concise, expressive terms
that mean more than they say. We need to talk about NPCs a lot
more than we need to talk about non-interactive non-NPC characters.

SeanB

Magnus Olsson

unread,
Dec 10, 2001, 5:19:59 AM12/10/01
to
In article <ae605ba7.01120...@posting.google.com>,
Nick Montfort <ni...@nickm.com> wrote:
>Finally, as far as I can tell the replies here have involved rigorous
>new discussion of what a PC is,

(...)

>Such discussion --
>rather than a new indexing scheme for the IF archive -- is what this
>theoretical taxonomy was for; that's why I proposed it. So, whether or
>not my own results are worthwhile or indeed are ever forthcoming, I am
>sorry that I cannot agree with those who think these categories
>useless.

Just a general comment here, not necessarily directed at your proposed
taxonomy:

Should the purpose of a taxonomy be just to provoke discussion?

I think that a taxonomy should have two purposes:

One is to aid in cataloguing things. Linnaeus's _Systema Naturae_
brought order in the chaos of the living world by simply providing
a system to classify animals and plants. But you've stated that this
is not your primary purpose, and there isn't so much IF around that
we need a very advanced categroization just to keep track of it.

The other, more theoretical, purpose is to reveal patterns,
conenctions; to say something about the things being categorized. The
taxonomy of living organisms showed the relations between them, which
had earlier been hidden in the chaos, and paved the way for the theory
of evolution, which explains the structure behind the
taxonomy. Similarly, the periodic table showed patterns in the
properties of the elements, patterns which together with quantum
mechanics led to the modern atomic theory.

So I think a taxonomy of IF should be useful in the sense that it
should let us discover new things about IF, but for this the
distinctions it makes should be in some way essential, telling us
something about the works, not just arbitrary distinctions to divide
the games into convenient subsets. (Yes, I know, the analogy is not
perfect; Linnaeus used some arbitrary distinctions, but you
hav eto start somewhere).

This is not to be taken as a criticism of your system; I'm just
pointing out that the fact that a taxonomic distinction doesn't seem
meaningful to someone is in fact a valid criticism, and that I think
that a taxonomy should be useful for more than generating discussion.

(To return to the biological analogy: if I divide all birds into
classes depending on colour, I will probably generate a lot of
interesting discussion about the pigmentation and behaviour of birds,
but I'll be missing the deeper connections shown in a genetic
classification).

Jon Ingold

unread,
Dec 10, 2001, 7:01:53 AM12/10/01
to
> (To return to the biological analogy: if I divide all birds into
> classes depending on colour, I will probably generate a lot of
> interesting discussion about the pigmentation and behaviour of birds,
> but I'll be missing the deeper connections shown in a genetic
> classification).

A nice parallel with statistics - if you divide your samples into an
incorrect sampling set, it's entirely possible you'll generate totally
arbitrary results, which suggest a causality that's not there.

Jon


Nick Montfort

unread,
Dec 10, 2001, 9:04:52 PM12/10/01
to
I do agree that just provoking discussion shouldn't be the main point
of any typology -- categorization should also be done based on
relevant features so as to help future work.

m...@df.lth.se (Magnus Olsson) wrote in message news:<9v228f$j96$1...@news.lth.se>...

> I think that a taxonomy should have two purposes:
>

> One is to aid in cataloguing things. ... But you've stated that this


> is not your primary purpose, and there isn't so much IF around that
> we need a very advanced categroization just to keep track of it.

I do think some categorization can help IF here, but it needn't be
formal or even consistent. Looking at IF by genre (even though no one
agrees completely as to what's in what genre) can still help someone
find some IF work they like. There are other factors that might help
someone seeking to play IF: speed-if vs. comp-length vs. longer works,
for instance. These may or may not be of much use to a theorist, and I
was just seeking to build a typology useful in theoretical discussion.
Implications of it might help to categorize IF for other purposes.

> The other, more theoretical, purpose is to reveal patterns,
> conenctions; to say something about the things being categorized.

Definitely. If you are able to make strong statements about the nature
of something, dividing that into categories based on those statements
should (if you're right) help you learn more about what you're
studying. You can see other relationships and dependencies more easily
after making such a division, for instance.

In the case of the typology I drafted, I had no demonstration of its
usefulness in my draft, and yet I did ask if it seemed useful. I feel
like I'd have to figure out some implications before I actually knew
whether it was useful or not, but I do think that if it allows
discussion of new definitions and strong principles of theory that
it's helping about as much as it can right now, prior to any more
widespread categorization and statistical analysis.

I think I didn't pick all the best or most salient variables, but may
have been close enough so that something of more use than this can be
figured out, based on this thread's discussion and further thinking.

-nm

Magnus Olsson

unread,
Dec 11, 2001, 10:10:54 AM12/11/01
to
In article <ae605ba7.01121...@posting.google.com>,

Nick Montfort <ni...@nickm.com> wrote:
>I do agree that just provoking discussion shouldn't be the main point
>of any typology -- categorization should also be done based on
>relevant features so as to help future work.

Yes, and as I wrote, I wasn't really criticizing your typology but
just making some observations on taxonomy in general.

>m...@df.lth.se (Magnus Olsson) wrote in message
>news:<9v228f$j96$1...@news.lth.se>...

>> One is to aid in cataloguing things. ... But you've stated that this
>> is not your primary purpose, and there isn't so much IF around that
>> we need a very advanced categroization just to keep track of it.
>
>I do think some categorization can help IF here, but it needn't be
>formal or even consistent. Looking at IF by genre (even though no one
>agrees completely as to what's in what genre) can still help someone
>find some IF work they like. There are other factors that might help
>someone seeking to play IF: speed-if vs. comp-length vs. longer works,
>for instance. These may or may not be of much use to a theorist, and I
>was just seeking to build a typology useful in theoretical discussion.

I think that some theoretically useful factors may also be interesting
for people looking for games to play: if we could classify the
interactivity of the piece, for example, or the linearity, I think it
would be helpful, since many people have strong opinions of how these
aspects affect playability. One theoretical issue would of course be
to find a workable definition of "interactivity" and "linearity".

Reply all
Reply to author
Forward
0 new messages