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 Historical content remains viewable.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

NPC Responses: One vs. Many

12 views
Skip to first unread message

Jim Aikin

unread,
Jul 30, 2007, 6:32:54 PM7/30/07
to
One of my testers has voiced a preference for NPCs' ask-tell and
give-show responses to "know" their own history:

>ask gwendolyn about artillery
"The cannon is fully loaded and ready to fire," Lieutenant Gwendolyn
informs you.

>g
"I already told you," Lieutenant Gwendolyn says. "The cannon is fully
loaded and ready to fire."

This is fairly easy to do in TADS 3. And with such a short, simple
example there's little reason not to do it.

Some NPC responses, however, are fairly long and contribute more to
characterization than to informing the player of useful details. In such
cases, I can't help feeling it's rude to the reader to print a fully
fleshed out conversation the first time a question is asked, and then
refuse to print it out again using a "You already asked the emaciated
prisoner about the dungeon" message. What if the player simply wants to
re-read the passage in question? Yet if you DO print it out the same way
a second time, it's _clearly_ a violation of mimesis.

My line of thinking, then, was this: If a number of long responses
should be printed in their entirety each time a topic is brought up,
does it make sense to vary the short, easy-to-vary responses? If you do
that, then the player is going to get in the habit of asking every
question two or three times just to see what will happen. And that's
just as much of an annoyance for the player, because it would be
foolhardy indeed of the game designer to bury an _essential_ piece of
information in the 3rd response to 'ask gwendolyn about artillery'.

So I've designed my WIP in such a way that the NPC almost always gives
the same response each time (unless the game state has changed, of
course). That way, players will quickly acclimate to the fact that
asking more than once is not useful, and will stop doing it.

There may not be any perfect answer to this problem, but I'm curious how
others feel about it, or what has been tried and found wanting (or
non-wanting).

--Jim Aikin

George Oliver

unread,
Jul 30, 2007, 6:52:27 PM7/30/07
to
What about something like a conversation thread where you use a
StopEventList though? Maybe you're doing this already, but it seems
like you would want to arrange things so that if there is one entry or
multiple entries in the list, it's still obvious which is the last
one.

Jim Aikin

unread,
Jul 30, 2007, 7:08:46 PM7/30/07
to
George Oliver wrote:
> What about something like a conversation thread where you use a
> StopEventList though? Maybe you're doing this already, but it seems
> like you would want to arrange things so that if there is one entry or
> multiple entries in the list, it's still obvious which is the last
> one.

Using a StopEventList is a good solution, unless you've hand-coded
topicResponse (which I sometimes do in order to allow for changes in the
scenario). In that case, adding a bit more code to produce the same
result is not difficult.

My question is not technical, however. My question is philosophical:
SHOULD the responses of an NPC always be multiple, or is it acceptable
to have them always be single. If they should be multiple, isn't it rude
to the player to refuse to reprint the first response a second time if
it contains actual written prose (as opposed to Processed Imitation
Prose Substance, which is by far the more common output in IF)?

--JA

George Oliver

unread,
Jul 30, 2007, 7:23:30 PM7/30/07
to
Given that I think it's not pleasing to reprint an entire
StopEventList if a player tries again on a topic that's been
exhausted, and that I'd rather have responses be consistent whether
it's a multi-entry list response or a single entry response, I prefer
that the last entry on a topic is a summary -- not a repeat of the
previous entry.

>a blarg

blarg 1

>a

blarg 2

>a

blarg 3

>a

>She's already told you that blarg 1 is blah, 2 is bleh, and 3.is bluh

>a foo

foo 1

>a

>She's already told you that foo 1 is fee.


Eric Eve usually does something like this.

George Oliver

unread,
Jul 30, 2007, 8:01:01 PM7/30/07
to
Replying to my own post, there are some obvious command/formatting
errors there, apologies.

Poster

unread,
Jul 30, 2007, 10:56:10 PM7/30/07
to

I've thought about this quite a bit while working on my WIP. I think the
problem with conversations in IF is that we've all become accustomed to
some formula that doesn't match up with real-life conversations.

For instance, have you noticed that rarely do NPCs or players greet one
another or bid farewell? Players esp. are bad about this. They just
start quizzing NPCs the instant they find one. Yet in the real world,
you'd be hard-pressed to find people who often communicate this way.

All of that is to say that modeling the conversation with a touch more
realism may be the way out. One way to do this is to have the NPC's
responses grow less inviting with each time the question is asked,
culminating with a refusal to speak with the player at all. Of course
different NPCs may handle this process differently. A blind six-year-old
girl might simply clap her hands over her ears after the second
question, whereas a loquacious NPC might decide to speak at length about
his life story so that the player can't get a word in edgewise.

Also, you might allow the NPC to give the informative, initial response
if the player adds the adverb "again". That way you're not breaking
mimesis, and the player can still get what he wants.

Still another way of handling this issue is to supplant "ask/tell" with
"talk" a la the Earth and Sky games.

My WIP uses a combination of the first and the third techniques.

-- Poster

www.intaligo.com Building, INFORM, Seasons (upcoming!)

Ron Newcomb

unread,
Jul 30, 2007, 11:25:13 PM7/30/07
to
>Also, you might allow the NPC to give the informative, initial response
>if the player adds the adverb "again". That way you're not breaking
>mimesis, and the player can still get what he wants.

That's a great idea. Mind if I steal it? :)

Also, some works use "You remember that..." and then the screen gets
all swimmy from the flashback.

"Remember" or "recall" is frequently used to call up old information,
from conversations, examines, room descriptions, etc.

I'm reminded of re-narration as internal dialogue, like in those old
detective movies when the detective thinks to himself as a way of
reminding the viewer of important facts,

'When I'd asked the dame about why her parlor door was unlocked the
night of the murder, she had replied in a funny way that she just
wanted some air, even though it was winter, and she had just gotta
over the pneumonia. She was hiding something all right, but I didn't
want to rehash the subject with her and make her suspicious.'

--R.

Emily Boegheim

unread,
Jul 31, 2007, 11:24:39 AM7/31/07
to
Jim Aikin <midig...@sbcglobal.net> wrote in
news:f8louo$8gq$1...@aioe.org:

> One of my testers has voiced a preference for NPCs' ask-tell and
> give-show responses to "know" their own history:
>

>[snip]


>
> Some NPC responses, however, are fairly long and contribute more to
> characterization than to informing the player of useful details. In
> such cases, I can't help feeling it's rude to the reader to print a
> fully fleshed out conversation the first time a question is asked, and
> then refuse to print it out again using a "You already asked the
> emaciated prisoner about the dungeon" message. What if the player
> simply wants to re-read the passage in question? Yet if you DO print
> it out the same way a second time, it's _clearly_ a violation of
> mimesis.
>

> [snip]


>
> There may not be any perfect answer to this problem, but I'm curious
> how others feel about it, or what has been tried and found wanting (or
> non-wanting).

Personally, I lean towards something like this:

>ask gwendolyn about artillery
"The cannon is fully loaded and ready to fire," Lieutenant Gwendolyn
informs you.

>g
You've already asked her about the artillery, and she told you it's


loaded and ready to fire.

It saves the weirdness of having the character repeating him/herself
verbatim every time the player asks about the artillery, but still
allows the player to access the necessary information if he/she forgets
what the character said. If you're consistent, it also makes it pretty
clear that the player doesn't have to ask any question more than once.

But just having one response for each question and repeating it if the
question is repeated works fine too. As long as you're consistent.

I really don't like having more than one (or one-and-an-auxiliary)
response, though. TADS 3 (the documentation, at least) seems to
encourage multiple responses, but typing "ask gwendolyn about artillery.
g. g. g. g. g" is pretty boring. Yes, it may be realistic to have more
than one question about the artillery, but in real life I'd ask the full
question, not just walk up to Gwendolyn and say, "Artillery?" hoping she
understands what my query is.

Emily

Emily Short

unread,
Jul 31, 2007, 11:47:19 AM7/31/07
to
On Jul 31, 4:24 pm, Emily Boegheim <emily.boegh...@gmail.com> wrote:

> I really don't like having more than one (or one-and-an-auxiliary)
> response, though. TADS 3 (the documentation, at least) seems to
> encourage multiple responses, but typing "ask gwendolyn about artillery.
> g. g. g. g. g" is pretty boring. Yes, it may be realistic to have more
> than one question about the artillery, but in real life I'd ask the full
> question, not just walk up to Gwendolyn and say, "Artillery?" hoping she
> understands what my query is.

This has caught me out more than once in recent TADS 3 games -- I ask
a question and then assume that's the only thing to learn about that
keyword, but it turns out that I can't make the game progress unless I
ask the character about the same keyword another three times. This is
partly just a sign of my own training as an IF player, and I suspect
my habits will change as more such games are released, but for right
now I tend to stumble on this unless the author gives some hint that
there's more to ask on the same topic.

Jim Aikin

unread,
Jul 31, 2007, 3:50:08 PM7/31/07
to
Emily Boegheim wrote:
>
> Personally, I lean towards something like this:
>
>> ask gwendolyn about artillery
> "The cannon is fully loaded and ready to fire," Lieutenant Gwendolyn
> informs you.
>
> >g
> You've already asked her about the artillery, and she told you it's
> loaded and ready to fire.
>
> It saves the weirdness of having the character repeating him/herself
> verbatim every time the player asks about the artillery, but still
> allows the player to access the necessary information if he/she forgets
> what the character said.

That's a fine solution if you're mainly concerned about functionality --
that is, NPC answers that provide concrete information. However, if
you're also concerned with literary quality, then refusing to reprint a
text that the reader may want to re-read is sort of rude, it seems to
me. Especially since days may have passed since the reader first asked
the NPC that particular question.

There's an additional issue, though, which is -- what if the information
content in the NPC's answer to a question is oblique or ambiguous? What
if it _may_ contain information, or may simply be local color? I have at
least half a dozen responses of that type in my WIP. They're in-game
clues, but not (I hope) obvious ones. So printing a summary of the
useful information in response to 'g' would materially detract from the
quality of the game.

--JA

George Oliver

unread,
Jul 31, 2007, 4:27:15 PM7/31/07
to
On Jul 31, 12:50 pm, Jim Aikin <midigur...@sbcglobal.net> wrote:
>
> if you're also concerned with literary quality, then refusing to reprint a
> text that the reader may want to re-read is sort of rude, it seems to
> me. Especially since days may have passed since the reader first asked
> the NPC that particular question.

There are a lot of situations where the player could lose important
scrollback from one session to the next. Maybe you can address the
larger issue by providing a clean transcript of the whole game to date
that the player could check back in.

Khelwood

unread,
Jul 31, 2007, 4:37:28 PM7/31/07
to

I'm not sure how repeating large chunks of text improves literary
quality.

If the information is not essential, then players don't need to read
it twice. Repeating conversations exactly makes NPCs look dumb and
badly coded, but if you think it's necessary, try and find an in-game
justification for it.
If the clues they are giving are essential, and you want to player to
be able to re-read them, then you might also (1) put them somewhere
else, (2) summarise, (3) program some elaborate system whereby the PC
takes notes or records all his/her conversations, or (4) put a warning
in the game to strongly encourage players to take detailed notes.

I suspect that many IF writers wouldn't hide essential clues in long
passages of conversation, so that they wouldn't have to compromise on
this sort of thing.

Jim Aikin

unread,
Jul 31, 2007, 5:36:14 PM7/31/07
to
Khelwood wrote:
>
> I'm not sure how repeating large chunks of text improves literary
> quality.

Yeah, there's really no point in reading Hamlet more than once. I mean,
once you know he's going to die in the end, what's the point?

> If the information is not essential, then players don't need to read
> it twice. Repeating conversations exactly makes NPCs look dumb and
> badly coded, but if you think it's necessary, try and find an in-game
> justification for it.

Difficult, since real people seldom repeat long slugs of text verbatim.

> If the clues they are giving are essential, and you want to player to
> be able to re-read them, then you might also (1) put them somewhere
> else, (2) summarise, (3) program some elaborate system whereby the PC
> takes notes or records all his/her conversations, or (4) put a warning
> in the game to strongly encourage players to take detailed notes.

No, the clues aren't essential -- they're subtle. That's kind of the
point, actually. Thinking that a clue is either essential or useless is
much too digital for my taste. And summarizing is a bad idea if you want
the clue to be oblique.

Of course the game has a hint system, so the in-game clues aren't
strictly needed. But the alert player is rewarded for asking the right
questions and thinking about the responses by needing to use the hints less.

--JA

Khelwood

unread,
Jul 31, 2007, 6:05:08 PM7/31/07
to
On 31 Jul, 22:36, Jim Aikin <midigur...@sbcglobal.net> wrote:
> Khelwood wrote:
>
> > I'm not sure how repeating large chunks of text improves literary
> > quality.
>
> Yeah, there's really no point in reading Hamlet more than once. I mean,
> once you know he's going to die in the end, what's the point?

Your logic does not resemble earth logic.
I haven't seen Hamlet for a few years, but I'm fairly sure its acclaim
isn't due to characters repeating verbatim conversations from earlier
scenes so that the audience doesn't miss their significance.

David Fisher

unread,
Jul 31, 2007, 7:12:53 PM7/31/07
to
"Emily Short" <ems...@mindspring.com> wrote in message
news:1185896839.1...@d55g2000hsg.googlegroups.com...

Something I tried out in an unreleased game is to print an unobtrusive "."
on the right side of the status bar whenever the player performs an action
that is worth repeating (eg. a conversation topic that isn't exhausted yet).
This looked like it might work OK, as long as the player knows what to look
out for.

I don't much like the idea of having to try every conversation topic several
times to make sure I don't miss something important, but it does sometimes
make sense for a game to have several different responses to the same topic,
so this seemed like a reasonable compromise. Printing some text like "(more
available)" or "(type g)" seemed a bit too blatant, thus the "." idea.

David Fisher


The Wanderer

unread,
Jul 31, 2007, 7:40:13 PM7/31/07
to
Poster wrote:

> For instance, have you noticed that rarely do NPCs or players greet
> one another or bid farewell? Players esp. are bad about this. They
> just start quizzing NPCs the instant they find one. Yet in the real
> world, you'd be hard-pressed to find people who often communicate
> this way.

I find myself immediately thinking of the prospector in "Gateway". I
replayed the entire game twice just to get that last point...

(no details because it's technically a spoiler and doesn't seem
important enough for me to bother with phrasing my way around spoiler
space)

--
The Wanderer

Warning: Simply because I argue an issue does not mean I agree with any
side of it.

Secrecy is the beginning of tyranny.

Message has been deleted

Zylon

unread,
Aug 1, 2007, 6:40:44 AM8/1/07
to
"Jim Aikin" <midig...@sbcglobal.net> wrote in message
news:f8omqa$h0n$1...@aioe.org...

> I was talking about giving the player the opportunity to re-read. I feel
> it's important that the player be able to re-read conversations if
> desired. Only letting them be viewed once strikes me as rude. One might
> want to re-read Hamlet (not that most IF resembles Shakespeare in any
> particular) for much the same reason -- simply to enjoy re-reading it.

It's an interesting question. Basically you're talking about two different
emphases: one is on making something "literary" enough that someone might
want to reread it just for the sake of doing it. The other is on adhering to
some sort of "mimesis" such that repeated verbatim conversations are seen as
something that doesn't "really" happen.

These are two things at odds. If someone wants to re-read just to re-read,
they should probably make a transcript or play again. But if they want to
re-read because they think they might have missed a vital clue that's a
different issue. This may differ for various players. Some players may like
the idea of being able to call up specific information at any time even if
they have already seen it. Others may find that a break in reality since, in
the "real world", you couldn't do this. The trick seems to be coming up with
a mechanism that makes it possible to "recall" a converastion or at least
the basics of it. Then the player can try asking the NPC about it again to
determine if there's more to learn. Example:

> ASK NPC ABOUT RING OF POWER
{Here the NPC says a lot of stuff about the Ring or Power}

> WHAT DO I KNOW ABOUT THE RING OF POWER?
{Can you do this in Inform or TADS?}
From your last conversation with NPC, you know that the Ring is ....

> ASK NPC ABOUT RING OF POWER
{NPC says something bland}

> WHAT DO I KNOW ABOUT THE RING OF POWER?
You asked NPC a couple of times but it seems that all he know is that the
Ring is ....


Blank

unread,
Aug 1, 2007, 7:36:08 AM8/1/07
to
Jim Aikin wrote:
> Emily Boegheim wrote:
>>
>> Personally, I lean towards something like this:
>>
>>> ask gwendolyn about artillery
>> "The cannon is fully loaded and ready to fire," Lieutenant Gwendolyn
>> informs you.
>>
>> >g
>> You've already asked her about the artillery, and she told you it's
>> loaded and ready to fire.
>> It saves the weirdness of having the character repeating him/herself
>> verbatim every time the player asks about the artillery, but still
>> allows the player to access the necessary information if he/she forgets
>> what the character said.
>
> That's a fine solution if you're mainly concerned about functionality --
> that is, NPC answers that provide concrete information. However, if
> you're also concerned with literary quality, then refusing to reprint a
> text that the reader may want to re-read is sort of rude, it seems to
> me. Especially since days may have passed since the reader first asked
> the NPC that particular question.
>

I think the problem here is that reader != player.

The player just wants the information: a reminder, or a nudge. Here, a
summary of a long conversation chain is *better* than a complete replay.
The author selects out the important info precisely so that the player
can glance over it, smack forehead ("from *Patagonia* not *Paraguy*")
and carry on with the game.

The reader wants to re-experience a particular piece of writing because
it's enjoyable *as writing*. Consequently, this action is really
happening outside the game. (Okay it might be a graceful series of
exchanges, so the reader wants to re-experience that good bit of game
design, but my distinction is still valid, I think: the "player" wants
to progress the game, the "reader" does not.)

I can't see any way that you can gracefully make allowance for those
different desires without requiring the human to give the game that
extra information.

Something blunt like:

> re-enact (yeah, I know, horrible command name)
Re-enaction mode is now on: all characters will react as though they
have not encountered the player before.

R>> ask Mabel about canary

"I like your canary," you say.
"Isn't he pretty?" she trills. "My great aunt Jemima was hugely fond of
canaries...." etc.

Though storing and re-setting the entire game state seems like a lot of
work: better just to go back to a previous save.

jz

Emily Short

unread,
Aug 1, 2007, 9:05:28 AM8/1/07
to
On Aug 1, 11:40 am, "Zylon" <zylo...@hotmail.com> wrote:
> If someone wants to re-read just to re-read,
> they should probably make a transcript or play again.

I agree. To look at it another way: there are other passages of text
in an IF game that could have literary quality and be of interest to
reread: room descriptions, responses to commands, and one-time events
could all produce text I want to see again. But there is no player
expectation that all of these will remain accessible to be re-read
after they're first encountered -- and trying to construct a game so
would be severely limiting and unrealistic.

Jim Aikin

unread,
Aug 1, 2007, 12:58:49 PM8/1/07
to
Emily Short wrote:
>
> I agree. To look at it another way: there are other passages of text
> in an IF game that could have literary quality and be of interest to
> reread: room descriptions, responses to commands, and one-time events
> could all produce text I want to see again. But there is no player
> expectation that all of these will remain accessible to be re-read
> after they're first encountered -- and trying to construct a game so
> would be severely limiting and unrealistic.

Good point. I was aware as I was writing an earlier message that the
same argument could be made about cut-scenes such as the death of a
major character, which certainly couldn't be played again (other than by
undoing and rereading it immediately, which is beside the point).

--JA

Jim Aikin

unread,
Aug 1, 2007, 1:06:17 PM8/1/07
to
Zylon wrote:

>> WHAT DO I KNOW ABOUT THE RING OF POWER?
> {Can you do this in Inform or TADS?}
> From your last conversation with NPC, you know that the Ring is ....

Oh, yeah, that's not difficult at all. It's just a lot of work. Let's
suppose there are fifty topics in a game, and half a dozen NPCs, each of
whom will have responses to some (but perhaps not all) of the topics. So
then you have this:

>think about ring of power
When you asked Gollum about the Ring of Power, he said, "Precious! Me
wants it!"

When you asked Gandalf about the Ring of power, he said, "There is no
such thing. Silly hobbit!"

...and so on. You can put verbatim quotes in the 'think about' code and
set flags when a particular NPC discusses a particular topic. But of
course, if the NPC's response can change depending on where you are in
the story, the code will start to get messy.

So it _is_ a lot of work, and what's the gain? Mimesis is not God. We
all understand that in the real world you don't have a prompt where you
can type commands. We all understand that in the real world, 'tell
gollum that he's never going to get his filthy hands on it' would be a
meaningful command, but it's not going to be understood by the parser.
We all understand that "You see no trees here" is just a trivial bug,
not an indication that the player character has suddenly gone blind.

So why should it be such an awful crime against mimesis to have an NPC
provide the same output twice in response to the same input?

Maybe I'm just being lazy. Dunno.

--JA

Lodestone

unread,
Aug 1, 2007, 1:19:09 PM8/1/07
to
Hallo Jim,

I must confess to getting a little thrill for being the "one of my
testers" mentioned in your original post. There's something
deliciously clandestine about it, mixed with something else. Or
something.

Anyway:

> That's a fine solution if you're mainly concerned about functionality --
> that is, NPC answers that provide concrete information. However, if
> you're also concerned with literary quality, then refusing to reprint a
> text that the reader may want to re-read is sort of rude, it seems to
> me. Especially since days may have passed since the reader first asked
> the NPC that particular question.
>
> There's an additional issue, though, which is -- what if the information
> content in the NPC's answer to a question is oblique or ambiguous? What
> if it _may_ contain information, or may simply be local color? I have at
> least half a dozen responses of that type in my WIP. They're in-game
> clues, but not (I hope) obvious ones. So printing a summary of the
> useful information in response to 'g' would materially detract from the
> quality of the game.

I'm not sure all of the responders have quite grasped the crux of the
problem you feel you're facing, though everyone's contributed
something, so I'm going to put in two cents again. Or, rather, try and
pull together a lot of what's been said in this thread, maybe adding a
couple of things.

SOMETHING ONE MIGHT WANT: Mimesis
- If the player's in the habit of asking the same question a couple of
times, due to previous game experience (more likely for TADS game),
then printing exactly the same response is going to seriously break
the mimesis. For some reason, this is more disturbing to the player's
suspension of disbelief than other repeats. (I have some ideas about
why, but they're better talked about elsewhere.)
POSSIBLE SOLUTION 1: Give each character a set of stock "I already
told you about that" responses.

BUT HERE'S A PROBLEM: Repeating essential information
- If all asking the same question gets is "I already told you", then
unless the player has been keeping notes they're going to get
frustrated about having lost information, especially several sessions
later.
POSSIBLE SOLUTION 2a: Give each character a second response in which
they recap the essential information (Con: maybe also not particularly
mimetic)
POSSIBLE SOLUTION 2b: Develop a recall function which recaps essential
information

BUT HERE'S A PROBLEM: Wanting to preserve subtlety of hints
- If all you ever get is recap of essential information, then subtle
hints within the text will be lost. Moreover, that information is
recapped shouts it to the hills as being a hint.

AND HERE'S ANOTHER: Wanting to allow the player to reread the
literature
- What if the player wants to read a particularly good passage again?

- Now I'll argue for a bit. I think IF is essentially a temporal art
form. So, I would say, is drama. There is as big a difference between
watching Hamlet and reading Hamlet as there is between playing IF and
reading an IF transcript. In order for Hamlet (and an IF piece) to
work as an artform, that temporality has to be preferred. But Hamlet
has some beautiful poetry in it, and it's a pain to have to go and
watch Hamlet every time I want to hear it. So I keep a copy of Hamlet
around, so's to be able to read that poetry again. Similarly, if I'm
to critique Hamlet I'm going to watch it several times, perhaps
keeping a copy with me so's to refer back to it during a watching
session in order to tie threads together. So:

POSSIBLE SOLUTION 3a: On top of the above solution, develop a meta-
command which allows the player to repeat a particular scene or
conversation. (Con: lots of coding!)
POSSIBLE SOLUTION 3b: If you're really worried about the player not
being able to reread things, simply encourage them very strongly in
the notes to keep a transcript.

I think a combination of the above solutions should cover all
bases :-)

Hx

Zylon

unread,
Aug 1, 2007, 1:22:25 PM8/1/07
to
"Jim Aikin" <midig...@sbcglobal.net> wrote in message
news:f8qei9$3pk$1...@aioe.org...

> Zylon wrote:
>
>>> WHAT DO I KNOW ABOUT THE RING OF POWER?
>> {Can you do this in Inform or TADS?}
>> From your last conversation with NPC, you know that the Ring is ....
>
> Oh, yeah, that's not difficult at all. It's just a lot of work.

It's not that difficult; just a lot of work. :) To me, something that's a
lot of work _is_ difficult. It may not be challenging in one sense, but it's
challenging in another.

> ...and so on. You can put verbatim quotes in the 'think about' code and
> set flags when a particular NPC discusses a particular topic. But of
> course, if the NPC's response can change depending on where you are in the
> story, the code will start to get messy.

Well maybe that's where some extension can come in to help modularize. With
an object-oriented language, this shouldn't have to get too messy. Of course
I'm still fresh on the scene so that's easy for me to say. I'm just thinking
of other programming languages. But I also see this as really one aspect
that text adventures can provide that other games can't, at least not as
easily. So if it's a distinguishing aspect, I would say it's worth exploring
it and making it even more distinguishing.

> So it _is_ a lot of work, and what's the gain? Mimesis is not God.

I agree but like you started out asking, there's at least some balance
you're trying to achieve between doing something that's not totally
unbelievable and apparently something that's not too much work. As far as
what's the gain, I guess I could see a lot of gain if it made the whole game
world seem a bit more realistic in relation to itself or made the characters
seem a little "smarter." It's not so much that you're going for "mimesis" as
you are trying to provide a robust set of characters for the player to
interact with. If your game deals with monsters, ghosts, and wizards anyway
you've already thrown out "mimesis" so who cares about it at that point?
What you're looking at is how much interaction is possible between player
and characters.

> So why should it be such an awful crime against mimesis to have an NPC
> provide the same output twice in response to the same input?

I'm not sure what a crime against mimesis is. Do you mean violating some
notion of "real life"? If so, I don't think it's a "crime" per se. I'm just
saying that even in the context of game worlds it's more just annoying
rather than reality-destroying because it exposes the game mechanics so
obviously. It's kind of like when you watch a movie and one part pisses you
off because you just know that they could have covered up that gaping plot
crater with just a line or two of dialogue. Same for me with game mechanics.
If you could have hidden them, why didn't you? Why couldn't repeated
responses have been more natural or more in context? Why couldn't there have
been a way for me to remember or figure out what was already said without
having to rely on a transcript or having to rely on repeated chunks of text?


Jim Aikin

unread,
Aug 1, 2007, 2:57:16 PM8/1/07
to
Blank wrote:
>
> Something blunt like:
>
> > re-enact (yeah, I know, horrible command name)
> Re-enaction mode is now on: all characters will react as though they
> have not encountered the player before.
>
> R>> ask Mabel about canary
>
> "I like your canary," you say.
> "Isn't he pretty?" she trills. "My great aunt Jemima was hugely fond of
> canaries...." etc.
>
> Though storing and re-setting the entire game state seems like a lot of
> work: better just to go back to a previous save.

In TADS 3 it's not actually a lot of work. Multiple responses to
ask/tell are stored in a StopEventList. You can reset the index of this
list to 0 (or, I think, to 1 ... most things in T3 are 1-indexed) for
all of the items in a single tidy block of code.

That may be the optimum solution, in fact. Thanks for helping thrash
this out!

OTOH, if the author has inconsiderately buried some vital information in
the 2nd or 3rd response to a given topic, switching on reenact mode
would prevent the information from ever appearing....

--JA

Lodestone

unread,
Aug 1, 2007, 3:21:17 PM8/1/07
to
On 1 Aug, 18:06, Jim Aikin <midigur...@sbcglobal.net> wrote:
> Mimesis is not God. We
> all understand that in the real world you don't have a prompt where you
> can type commands. We all understand that in the real world, 'tell
> gollum that he's never going to get his filthy hands on it' would be a
> meaningful command, but it's not going to be understood by the parser.
> We all understand that "You see no trees here" is just a trivial bug,
> not an indication that the player character has suddenly gone blind.
>
> So why should it be such an awful crime against mimesis to have an NPC
> provide the same output twice in response to the same input?

Alright, since you've asked the specific question, I think I'm going
to have a bash at expanding what I said in my post down the bottom:

> - If the player's in the habit of asking the same question a couple of
> times, due to previous game experience (more likely for TADS game),
> then printing exactly the same response is going to seriously break
> the mimesis. For some reason, this is more disturbing to the player's
> suspension of disbelief than other repeats. (I have some ideas about
> why, but they're better talked about elsewhere.)

Everything which is mimetic -- which is to say, all of art (and much
more) -- involves, if we are to take it as a facsimile of reality,
some form of suspension of disbelief. When we watch Hamlet, we must
ignore the fact that it's happening on a stage, that people don't talk
in pentameter, and so on. It is a sylisation of reality -- and how it
stylises reality is largely a matter of convention. These conventions
can be exploited to dramatic effect (or other kinds of effect) -- in
postmodern art, mimesis is often totally undermined by the artwork
explicitly recognising that it is an artwork (quite a few IF games do
this as well, often for comic effect).

But the sum of it is: there are some things we are happy to accept,
even though they're unreal, within the context of the artwork. In IF
we're so used to the prompt we barely notice it until it's messed with
(see "Shrapnel" or "Bad Machine"); we're used to not worrying about
repeated room descriptions, and so on. But sometimes things really
interfere with our experience. For example, if a room description
involves a personal reaction ("You are shocked to see that ...") which
doesn't make sense when it's repeated, it's a really bad idea to print
it every time you go to a room.

Character speech is a similart thing. Because we imagine what's going
on to be an interaction between the PC and the NPC, we take part in
that interaction, place ourselves in it, experience it. It's not the
same as a prompt (or any of the computing paraphenalia), which we can
happily ignore, or a repeated room description, which we don't invest
in psychologically. Dialogue, like cutscenes and other big plot
points, is highly experiential, and so it's particularly prone to
upsetting us when the game mechanics are so balatantly exposed as when
characters say the same thing twice -- especialy when there's now, I
think it's fair to say, a convention of this not happening. We're used
to our NPCs feeling like people, now -- not objects which happen to
give out text between quotation marks.

So that's why I think non-mimetic NPCs are a particular problem :-)

Lx.

0 new messages