Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Third Person IF

5 views
Skip to first unread message

Paul Oliveira

unread,
Jun 6, 1996, 3:00:00 AM6/6/96
to

I have a theory about a third person digital narrative, and although
perhaps it's been already discussed here, I wonder if anyone could give me
opinions or examples that would shed more light on it.

The current paradigm for text I'll call 2nd person imperative, in which the
narration is written in 2nd person and the commands are written as
imperatives:

You are walking along a river which runs north to south.
>GO SOUTH.

The problem with this is that it's a schizophrenic grammar. If *I'm* the
one all this is happening to, who am I telling to go south? It seems I'm
talking to the computer, since there's nobody in the actual story I could
be talking to besides myself, and I don't believe most games want to
position the main character as so eccentric. But why am I talking to the
computer? Is it part of the story? Usually not. I believe it's harder to
immerse yourself in the "message" if you're constantly typing commands at
the "medium".

Somebody here suggested that the computer be positioned as an audience, to
whom the player reveals the next step. I'll call this 3rd person
descriptive. I don't remember the example used, or the name of the person,
so with apologies, here is the way I remember it:

Fred walked along a river which runs north to south.
What did he do next?
>HE WENT SOUTH.

This is a bit more grammatically consistent. But why is the audience (the
computer) also doing the majority of the storytelling? Besides, strictly
speaking, isn't the player really also the audience? So the computer is the
audience, but so is the player, and the player is the storyteller, but also
the computer. In my opinion, this creates an even more schizoid feeling
than the current 2nd person imperative standard. And thus draws even more
unnecessary attention to the medium.

I think it's too complicated, and all for the sake of giving the player a
larger role of "interactivity". But the fact is, it's the nonlinearity of
the storyline and freedom of action, rather than the grammar, that gives a
work a more interactive feel. So why not 3rd person imperative, of a kind
that has been sort of used in more RPG-type games, like so:

Fred walks along a river which runs north to south.
>GO SOUTH.
Fred decides to follow the river south...

In this case, the player is "directing" the actions of the main characters,
rather than directly assuming their roles. The grammar makes sense. When
you type, you're talking to Fred, and he's part of the story. I think I can
anticipate that most objections to this will be that the player feels less
part of the story because he or she is not playing an actual person in it.
But as I said before, I believe that the amount of input the player has
into how things go is more important than whether the player is
characterised as the main person.

Actually, I think there are many offsetting advantages. The player's input
is addressed directly to the main character who is part of the story, so
"interaction" is actually with the story, rather than with the computer.
The main character can have a well defined personality, and can exhibit all
sorts of thoughts about what he or she encounters. The main character can -
*very* occasionally - not listen to the player, or make smart-ass
responses, which can make for more psychologically interesting puzzles.
Multiple characters can be controlled without stressing the central
metaphor. Room descriptions could change based on whom you're currently
"whispering" instructions to.

The only caveat is that it's sort of a mystery *why* the player can speak
to the main characters, and why the main characters listen to this strange
voice in their heads. I believe that this can safely be ignored in the same
way an audience watching a film doesn't question how it is able to see
what's on the screen, or why the main characters in the film don't notice
that the audience is watching them. But the question can also be
camouflaged by giving the player a nominal but purely communicative
identity. Like an artificial intelligence advising a futuristic swat team:

Dave is standing by the inner bay door.
>OPEN THE INNER BAY DOOR.
"I read you, HAL" says Dave, pushing the red button...

I realise that this is basically the interface for SUSPENDED, but has it
been tried with real fully functioning human beings as characters? I think
the beauty of this 3rd person imperative is that with every command, the
player is actually participating in the next action, rather than just
describing it to the computer. The player doesn't have to describe what HAL
itself does, because HAL does nothing. HAL only gives commands and asks
questions. You could call this the "ghost-player". I'm taking this tack in
the game I'm currently scripting, only the "ghost-player" is not an
artificial intelligence, but rather a disembodied being whose identity is
at the heart of the mystery...

Any comments?

Paul Oliveira

P.S. 3rd person imperative also provides a much more logical way to
restrict movement in open spaces. Examples:

You are by a lake. It's very cold.
>ENTER LAKE.
It's too cold.

versus:

Fred is by a lake. "I'm freezing," he mutters.
>ENTER LAKE.
"No bloody way," says Fred.

Richard G Clegg

unread,
Jun 7, 1996, 3:00:00 AM6/7/96
to

Paul Oliveira (73023...@compuserve.com) wrote:

> {Various stuff about person changes and tense changes in IF}

: Any comments?

Only that it reminds me of the rather desperate attempts in Victorian
novels to logicalize the "Who is the narrator?" problem. This lead
(IMHO anyway) to a whole crop of novels plagued by ugly framing devices
(diaries and letters within diaries and diaries within letters within
diaries) in an attempt to make the story-telling more "convincing".

For me, the past-tense would be not only clumsy but also irritating to
work with.

She was in a maze of twisty passages all alike.
What did she do next?
> S
She didn't understand that. What did she do next?
> SOUTH
She didn't understand that. What did she do next?
> SHE WENT SOUTH
She was in a twisty maze of passages all alike.

There is:
a) A huge overhead on typing
and
b) An unnatural sentence structure

also

c) It's really, rather too remeniscent of having to type "Simon Says" at the
front of every command.

Also, you have to remember that the command structure in IF is natural
if you're placing yourself in the position of the adventurer. If I'm
thinking about the problem then I'm thinking "What do I do next" not
"what did I do next" and if I have a flash of insight I think "aha... tie
the sheets to the bed" rather than "aha... he tied the sheets to the bed".
In this sense the current construct is natural and involving since the
player more directly types what he/she thinks into the parser.

The idea of having the "player" as a puppet guided by the player is
perhaps slightly better - and indeed has been tried in various forms.

--
Richard G. Clegg There ain't no getting round getting round
Dept. of Mathematics (Network Control group) Uni. of York.
email: ric...@manor.york.ac.uk
www: http://manor.york.ac.uk/top.html


Andrew C. Plotkin

unread,
Jun 7, 1996, 3:00:00 AM6/7/96
to

73023...@compuserve.com (Paul Oliveira) writes:
> The current paradigm for text I'll call 2nd person imperative, in which the
> narration is written in 2nd person and the commands are written as
> imperatives:
>
> You are walking along a river which runs north to south.
> >GO SOUTH.
>
> The problem with this is that it's a schizophrenic grammar.

The problem with your problem is that nobody considers it a problem.
We're all used to it. Since there will always be *something* to get
used to when you play your first skap, this is not a big deal.

Scott Adams used first-person imperative ("I'm in a forest." "CLIMB
TREE"), but I don't like that as much. Apparently a majority of
skap authors agree with me.

Footnote: If you haven't, go find the TADS game "Past Tense". It has
an option to do, well, past tense instead of present. This turns out
to be easy to get used to.

> I believe it's harder to
> immerse yourself in the "message" if you're constantly typing commands at
> the "medium".

Harder than what? Build something which works differently and we'll
see how we like it. You are, of course, at a disadvantage because
there *is* a format that we're all used to.

I'm not yelling at you, by the way. I want to see this
director-actor paradigm; I want to see all sorts of alternatives.
Very little has been done with this aspect of presentation, which
means that lots *can* be done. I think I'm just reacting to your claim
that the way it's usually done -- second person -- is causing problems
in and of itself.

> Fred walks along a river which runs north to south.
> >GO SOUTH.
> Fred decides to follow the river south...

> I think I can
> anticipate that most objections to this will be that the player feels less
> part of the story because he or she is not playing an actual person in it.
> But as I said before, I believe that the amount of input the player has
> into how things go is more important than whether the player is
> characterised as the main person.

But, in fact, when I compare first-person to second-person skaps I
have played, I feel more part of the story in second-person.

> Any comments?

Write something! (This is my reply to *any* experimental IF
suggestion, whether I bother to post it or not.)

> P.S. 3rd person imperative also provides a much more logical way to
> restrict movement in open spaces. Examples:
>
> You are by a lake. It's very cold.
> >ENTER LAKE.
> It's too cold.
>
> versus:
>
> Fred is by a lake. "I'm freezing," he mutters.
> >ENTER LAKE.
> "No bloody way," says Fred.

I sometimes use the format

You are by a lake. It's very cold.
>ENTER LAKE.

No bloody way.

That is, idiomatic speech, or something with a little personality.
This is of course helpful towards the problem of defining the "you"
character, just as your suggestion is helpful towards defining Fred's
character.

--Z

"And Aholibamah bare Jeush, and Jaalam, and Korah: these were the borogoves..."

Gord Jeoffroy

unread,
Jun 7, 1996, 3:00:00 AM6/7/96
to

73023...@compuserve.com (Paul Oliveira) wrote:

> You are walking along a river which runs north to south.
> >GO SOUTH.

>The problem with this is that it's a schizophrenic grammar. If *I'm* the
>one all this is happening to, who am I telling to go south?

I've always reconciled this problem with what I've seen in FAQs and
the help menus of individual games. Where those have tried to explain
how parser commands work, they've often suggested that the player
mentally preface the commands with "I want to...." So, when I'm
playing, I'm mentally doing this:

> You are walking along a river which runs north to south.

> >I WANT TO GO SOUTH.

Works for me.

--Gord


Magnus Olsson

unread,
Jun 8, 1996, 3:00:00 AM6/8/96
to

In article <8li6ql200...@andrew.cmu.edu>,
Andrew C. Plotkin <erky...@CMU.EDU> writes:

>73023...@compuserve.com (Paul Oliveira) writes:
>> The current paradigm for text I'll call 2nd person imperative, in which the
>> narration is written in 2nd person and the commands are written as
>> imperatives:
>>
>> You are walking along a river which runs north to south.
>> >GO SOUTH.
>> > The problem with this is that it's a schizophrenic grammar.
>
>The problem with your problem is that nobody considers it a problem.
>We're all used to it. Since there will always be *something* to get
>used to when you play your first skap, this is not a big deal.

and

In article <4p9aap$a...@netty.york.ac.uk>,


Richard G Clegg <ric...@manor.york.ac.uk> wrote:
>Paul Oliveira (73023...@compuserve.com) wrote:
>
>> {Various stuff about person changes and tense changes in IF}
>
>: Any comments?
>
> Only that it reminds me of the rather desperate attempts in Victorian
>novels to logicalize the "Who is the narrator?" problem. This lead
>(IMHO anyway) to a whole crop of novels plagued by ugly framing devices
>(diaries and letters within diaries and diaries within letters within
>diaries) in an attempt to make the story-telling more "convincing".


I agree with both Andrew and Richard here.

The combination of second-person narrative with imperative commands
seemingly directed at the computer is a well-established convention,
and very few IF players seem to be bothered by it. Personally, I think
the second-person narrative is very good at providing that feeling
of "actually being there" that is one of the hallmarks of good IF.
The third-person POV is much more detached.

Of course, if you, as an author, find that you have trouble writing a
narrative in the second person, or if you feel that the imperatives
somehow detract from your work, by all means try to adopt another
convention. But be prepared that this mabe irritating to your
audience, the vast majority of which will be used to the "second
person imperative" convention.

Perhaps those well-known words of wisdom are in place: "If it ain't
broke, don't fixe it". Most people don't seem to regard the "second
person imperative" convention as "broke". But, of course, the only way
to find out is to see how people react to IF written with other
conventions. I'm afraid I would find it rather contrived and awkward.

--
Magnus Olsson (m...@df.lth.se)
Speaking as a private citizen & taxpayer - no more, no less.

Greg Ewing

unread,
Jun 11, 1996, 3:00:00 AM6/11/96
to

>GO SOUTH.
> The problem with this is that it's a schizophrenic grammar.

Actually, I don't ever type "go south", just "s":-)

I don't really think of the input language as English,
but just another computer command language. It bears
some resemblance to a subset of English, but so does
Pascal, and nobody ever argues about whether whether
a Pascal program should be written in first or second
person. The concept just doesn't apply.

If it bothers you, just think of it as answering the
question "What do you want to do next?" that is
implicitly asked at each prompt. Or if you're
writing a game (or should I say poif?), make that the
actual prompt!

(I'd provide an option to turn it off, though,
otherwise many players will get annoyed with it.)

Also, I don't think the player's commands should
be regarded as part of the narrative, and any
distortion of the command syntax to make them so
would seem contrived to me.

Greg

Andrew C. Plotkin

unread,
Jun 11, 1996, 3:00:00 AM6/11/96
to

Greg Ewing <gr...@cosc.canterbury.ac.nz> writes:
> I don't really think of the input language as English,
> but just another computer command language. It bears
> some resemblance to a subset of English, but so does
> Pascal, and nobody ever argues about whether whether
> a Pascal program should be written in first or second
> person. The concept just doesn't apply.

I wonder if anyone ever worries about the "grammar" of graphical
games. You look at an image of a room, move a hand-cursor around, and
click on the image of an object to take it. Now is that *your* hand,
or the hand of a player-puppet; and are you performing the action, or
ordering someone to perform it, or just specifying an action in a
non-verbal command language...? Really all the same considerations
apply as in textual input. But since the science of "iconic language"
is only a couple of decades old, we breeze right past all the issues.
(Which, I still contend, most people do for text game input as well.)

Now, if our native language was pictographic rather than alphabetic,
would we get just as pissy about the grammar of Myst's input language?
:-)

> Also, I don't think the player's commands should
> be regarded as part of the narrative, and any
> distortion of the command syntax to make them so
> would seem contrived to me.

Oop! They're definitely part of the narrative, in the sense that you
can't read the transcript without the commands. (I mean, there isn't
enough information to tell what's going on, unless you read both the
commands and responses.)

If you just mean that the commands aren't in the same syntax as the
responses, then I agree.

The classic adventure command language is shaped by practical
concerns, not theoretical ones. It's gotta be parsable; and it's gotta
be as simple as possible, because people hate to type. Parsers will
continue to improve, but nobody will ever voluntarily type "I WANT THE
PROTAGONIST TO EAT THE HAM" when "EAT HAM" will work.

In fact, the balance is more complicated than that. It would be lovely
to have a parser that could accept "GRAB SOME LUNCH" as a synonym for
"EAT HAM". What, you say, Inform can be hacked to do that? Well, of
course. The reason nobody bothers -- I'm not sure this has been said
in so many words, so I'd better -- is because there are dozens of
idioms which spring to mind as readily as "GRAB SOME LUNCH". If the
player expects to be able to use that level of idiom, the parser must
know all of them. We don't have the technology for that. But in
classic adventure English, there is exactly one way to eat the damn
ham. The restricted syntax benefits the player just as much as it
benefits the parser-writer.

Except that then Deadline came along, and the model turned out to not
handle NPC conversation very well at all. (I believe Deadline was the
first game that tried to let you talk with people.) And we've been
stuck on that front ever since.

I think I've gotten *way* off-topic. What was the topic again?

Greg Ewing

unread,
Jun 13, 1996, 3:00:00 AM6/13/96
to

Andrew C. Plotkin wrote:
>
> Oop! They're definitely part of the narrative, in the sense that you
> can't read the transcript without the commands.

I think what I meant was that I don't feel the need for the
transcript to read like a novel. I suppose that when I enter
"s" the system could write "You go south." into the transcript,
but I would be highly annoyed if I had to *type* something
like that as input.

Also, if you're going to be pedantic enough to require a
totally literiform (a word I've just made up to mean
"having the form of a piece of literature") transcript, then it ought
to be written in the *past* tense, since by the time you
come to re-read it, the events it describes are in the past!

> The restricted syntax benefits the player just as much as it
> benefits the parser-writer.

Quite so. I also believe that the command syntax and all the
verbs should be documented and made available to the player
from the outset. Guess-the-verb puzzles, intended or otherwise,
suck in a big way!

> Except that then Deadline came along, and the model turned out to not

> handle NPC conversation very well at all. ... And we've been


> stuck on that front ever since.

Maybe we're being too pig-headed about trying to barge through
that barrier head-on, when there might be other ways around it?

We seem to be quite comfortable (most of us, anyway :-) giving
game commands in a language which is not English. Do we really
have to insist so dogmatically that the language for communicating
with NPCs has to be English (or whatever native natural language
the player speaks)?

When talking to the game, there are established conventions
concerning what sort of concepts the player ought to be able
to talk about (moving, picking things up and putting them
down, eating food, etc.) and what words and phrases to use
to refer to them.

If the same were true of NPCs, would we mind very much if the
language we talked to them in was restricted and stylised?
Might we actually prefer it that way, in the same way that
we like knowing that if "eat ham" doesn't work, the ham is
not edible?

What I'm suggesting is that we *invent* a small language
for talking to NPCs, and use that language to define
a "domain of discourse" that all NPCs are expected to
be able to hold a conversation about. The benefits of
this are:

(a) The restricted language covers a restricted domain,
so it's easier to program the NPC to respond in a
reasonable way over the whole domain.

(b) The player knows the limits of the language and
therefore the NPC, and can avoid wasting time saying
things that the NPC hasn't a hope of replying to in
any useful way.

Am I making any sense?

Greg

Andrew C. Plotkin

unread,
Jun 13, 1996, 3:00:00 AM6/13/96
to

I was thinking about this the other day, and I decided a nice simple
model would be the following classes of syntax:

PERSON, X is Y.

Fred, Jane is the victim.
Jane, I am the Wizard Floobits.

X and Y would be objects taken from a special class, the subjects of
discourse. (Objects of discourse?) Some of these would have the same
name as regular game objects, of course, but some wouldn't. ("is",
"are" and "am" are synonymous. Add a special hack so that "you" always
evaluates to the discourse-object associated with PERSON.) Note that
all such statements are symmetrical, so you'd want to set up the
library to test (Y, X) is there is no routine for (X, Y).

PERSON, tell me about X.
ask PERSON about X.

X is a subject of discourse. This works the same as in the Infocom
tradition. Not all people know about all subjects of discourse.

PERSON, X is STATE.

X is a subject of discourse; STATE is pretty much anything. This is
the loose-and-floppy part of the idea. "Fred, the army is
approaching." "Jane, the dragon is green." "Floyd, you are going to
die." This may be too hard to work in -- not too hard to code (it's
easy to make an object called "going to die"); but it may give the
player too much room.

I'd leave out "Tell PERSON about X." It's too vague; to use it, you
have to restrict the situation so that there's only one thing the
player might ever want to say about X.

Keep the obvious structure of ordering people to do things. "Fred, go
north."

Of course, I don't have time to implement this. But I think that it's
strong enough to support a scenario containing NPCs. (Even ignoring
the "X is STATE" command, which I like less and less the more I think
about it.) If you get it working, please post the source code.

Footnote: Yeah, forget the STATE syntax. Instead, have a limited
number of new connectives, replacing "is". "was", "will be", "is
near", "likes", "hates". A general "X CONNECTIVE Y" syntax.

It is *important* to have a *familiar* set of connectives. It's just
like regular command verbs. The player should know most of them right
at the beginning of the game, and there should only be a couple of new
ones in any game. Otherwise people will complain about "guess the
connective" puzzles, which will be just as bad as "guess the verb"
puzzles.

Damien Neil

unread,
Jun 14, 1996, 3:00:00 AM6/14/96
to

On Thu, 13 Jun 1996 14:17:43 +1200, Greg Ewing <gr...@cosc.canterbury.ac.nz> wrote:
>If the same were true of NPCs, would we mind very much if the
>language we talked to them in was restricted and stylised?
>Might we actually prefer it that way, in the same way that
>we like knowing that if "eat ham" doesn't work, the ham is
>not edible?

I'd like such a system. I have had an idea for a system of
interacting with NPCs rattling around in the back of my head
for some time, although I don't know if I'll ever do anything with it.

Given current technical restrictions, it is not possible to
communicate with NPCs using anything even remotely similar
to natural language. I feel, therefore, that trying to give
the impression that an NPC can, in fact, understand a natural
language is a pointless excercise. Fooling the player is not
possible.

The current traditional style of NPC interaction acknowledges this
to a degree. One may generally use statements of the form
> Ask Ahab about the whale.
> Ahab, tell me about the whale.
and receive a reasonable response.

Rarely, interactions in the opposite direction are possible:
> Tell Ahab about Queequeg.

I would say that this style of interaction (perhaps combined with
the ability to give commands to characters) is sufficient to create
far more complex situations and puzzles than have been made to date.
I would love to see a game with very few useful physical objects, but
very complex informational objects. Most of the player's actions
would involve learning information (either through exploration, or
interrogation of characters), and imparting it to other characters.

A nice convenience feature would be a command similar to `inventory'
for knowledge objects.

> topics
You know about:
Fred's drinking problem.

> Tell John about Fred's drinking problem.
``Terrible isn't it? After Jenny died, he just went to pieces.''

> Ask John about Jenny.
``She was Fred's wife for fifteen years. She died two years back,
giving birth to their daughter.''

> Ask John about Fred's son.
``With Jenny gone, Fred didn't feel he could raise the poor kid
properly. He put her up for adoption.''

> topics
You know about:
Fred's drinking problem.
Fred's daughter.

Perhaps someday I'll actually find the inspiration to put my
money where my mouth is and write something using this style
of interaction. :>

- Damien
--
The earth is flat.
All opinions expressed in the above are mine, not necessarily JPL's.


Christopher E. Forman

unread,
Jun 17, 1996, 3:00:00 AM6/17/96
to

Damien Neil <ne...@godzilla.jpl.nasa.gov> wrote:
: The current traditional style of NPC interaction acknowledges this

: to a degree. One may generally use statements of the form
: > Ask Ahab about the whale.
: > Ahab, tell me about the whale.
: and receive a reasonable response.
:
: Rarely, interactions in the opposite direction are possible:
: > Tell Ahab about Queequeg.

This is an interesting observation. Implementing it in a useful and
realistic manner, however, would require that each NPC have a "memory"
of some sort to permit responses like, "You already told me that.", or
plot points could be repeated.

: > topics


: You know about:
: Fred's drinking problem.

This is also interesting, but unfortunately less feasible. Consider this,
from the first region of Jigsaw:

> topics
You know about:
The party at Century park.
The sparkler.
The space behind the beer tent.
The wooden packing crate.
The tagged key.
The curious device.
The canvas rucksack.
The mysterious stranger.
The corner jigsaw piece.
The night-jar.
The chapel.
The abstract sculpture.
The piano stool.
The pencil.
Emiletch-book.
The crowd.
The monument.
The lightning conductor.

All this from half a dozen rooms! Obviously, we could shorten the list to
plot points, but that would simply give away what's important.

File this away with the "NPC language" idea.

--
C.E. Forman cef...@rs6000.cmp.ilstu.edu
Read the I-F e-zine XYZZYnews, at ftp.gmd.de:/if-archive/magazines/xyzzynews,
or on the Web at http://www.interport.net/~eileen/design/xyzzynews.html
Vote I-F in 1996! Visit http://www.xs4all.nl/~jojo/pcgames.html for info!

Greg Ewing

unread,
Jun 17, 1996, 3:00:00 AM6/17/96
to

Damien Neil wrote:
>
> > topics
> You know about:
> Fred's drinking problem.

Urk... Keeping track of what each NPC knows about is a
reasonable idea, but I'm not so sure about trying to
tell the *player* what he does and doesn't know!

Greg

Fredrik Ekman

unread,
Jun 19, 1996, 3:00:00 AM6/19/96
to

In article <31BF7A...@cosc.canterbury.ac.nz> Greg Ewing <gr...@cosc.canterbury.ac.nz> writes:
I think what I meant was that I don't feel the need for the
transcript to read like a novel. I suppose that when I enter
"s" the system could write "You go south." into the transcript,
but I would be highly annoyed if I had to *type* something
like that as input.

Ah, but there is another side of the coin. Me, I am interested in
using IF for language education. Right now, I am trying to improve
my Spanish by playing some Spanish text adventures. This has made
me come to the conclusion that the Spanish imperative isn's such an
easy thing as one would wish for. However, if I make a mistake when
typing the verb, the parser usually will not tell me, becuase it
only reads the first few letters of each word. If it read the
entire words I would be forced to really learn the correct form.

Similar things could be said about letting the player drop the
definite article of the noun. Not only is this usually
grammatically questionable at best, but it also does not help the
player to learn when to use what article in languages such as
Spanish, French and German.

Forcing the player to type whole words should also be an excellent
way to teach spelling.

Then again, you may only be interested in providing some mindless
entertainment, but in that case I would advise you to try to make
talk shows on TV instead.

/F


0 new messages