[ANNOUNCE] MOOCE (alpha) and Troll Bridge
Magnus Olsson
working on an Inform library for menu-driven conversation, called
MOOCE ("Magnus' Object-Oriented Conversation Engine", no relation to
Space Moose).
I'm planning to release the source of MOOCE as a library contribution,
but first I'd like to see how it works in a reasonably practical
context. For that reason, I've written a small demo game, "The Troll
Bridge", and I'd be very grateful if some of you could have a look at
it and tell me what you think.
The game is available from my web page as
http://www.df.lth.se/~mol/troll.zip
What I'd like to know is
a) Do the conversations feel "right" to the player, and does the
interaction work well?
b) Can you force it to misbehave or even crash by doing things I
haven't thought of trying myself?
MOOCE, and hence The Troll Bridge, is currently Glulx-only, so you'll
need some version of glulxe to run it (any version should be OK; I've
tested WinGlulxe and Linux glulxe linked with Glkterm, XGlk and
cheapglk). I'm planning to add support for the Z machine as well.
The game is nothing remarkable (though it is a complete, winnable,
game) and as for the conversations - well, I'm not aspiring to be the
new Emily Short. Not in this game, anyway.
Enjoy! Dissect! Criticize!
--
Magnus Olsson (m...@df.lth.se, m...@pobox.com)
------ http://www.pobox.com/~mol ------
OKB -- not okblacke
>a) Do the conversations feel "right" to the player, and does the
>interaction work well?
First of all, the system seems fundamentally good. Having developed a
system like this myself (of which more anon), I have a few nitpicks. To wit:
Personally, I prefer a menu with a proportional font. For one thing, it
lets you use different text sizes and use center justification, so you can have
the NPC's name centered above the menu options and displayed in a larger font
size. No offense, but the current display -- "Talking to NPC" displayed
left-aligned in a different background color -- looks a bit tacky to me.
As far as the interaction, it might be worthwhile putting in some code
that automatically enters conversation with the NPC if there is only one NPC in
the room. Also, I think the doubled prompt is misleading: it implies that
there is some different mode of interaction, when really all the normal
commands (TAKE, LOOK, etc.) are still valid. More generally (although this may
be a conscious design decision on your part), I don't think you need a separate
conversation mode at all; if there are menu options available, they're valid
commands, and if not, then regular commands are still valid.
>b) Can you force it to misbehave or even crash by doing things I
>haven't thought of trying myself?
Yes. If I am in a no-menu state (such as the beginning of the game), and
I then begin conversation and the menu opens, and I then UNDO or restore to a
no-menu state, the menu remains on the screen.
Now I will take the opportunity to briefly plug my own system, GxScript,
which solves all of the nitpicks I mentioned above (except that it requires
"SAY 1" instead of simply "1", just because I considered that a low priority
feature and haven't implemented it yet).
As an author, I am particularly interested in how your system looks when
used in a game's source code. I have found it relatively easy to design the
conversation system in theory, but relatively difficult to find a practical
form for it (i.e., one that doesn't require writing endless amount of code for
relatively simple things). I am especially interested in how you did the
conversation tree, what kind of control MOOCE gives the author over
activating/deactivating topics, shifting the subject, etc.
You are welcome to take a look at GxScript -- the newest version is
available from my page at
http://members.aol.com/brenbarn/IF-download.html -- and see the approach I
took. I'd be glad to hear what you think. (I'd be glad to hear what anybody
thinks, although I got little response when I first announced GxScript on
here.)
--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
OKB -- not okblacke <bren...@aol.comRemove> wrote:
>m...@df.lth.se (Magnus Olsson) wrote:
> Personally, I prefer a menu with a proportional font. For one thing, it
>lets you use different text sizes and use center justification, so you can have
I think I prefer that as well, actually; but since I'm doing my
development using glkterm which uses one text size and one font, I
haven't thought much about it. I think I'll have to move to Windows
(urk!) to see how things look in WinGlulxe (I've tested the game in
WinGlulxe, but only to see that it was playable).
>the NPC's name centered above the menu options and displayed in a larger font
>size. No offense, but the current display -- "Talking to NPC" displayed
>left-aligned in a different background color -- looks a bit tacky to me.
Different background colour? You're using strange default settings :-9
(or perhaps it's your interpreter that does). Time to bring out the
old stylehints....
> As far as the interaction, it might be worthwhile putting in some code
>that automatically enters conversation with the NPC if there is only one NPC in
>the room.
Or at least code that gives the game designer the possibility to do
that. Because I'm not certain I want my games to behave that way.
I think it's rather a deep issue, actually, that depends on how you
think about PC/NPC interaction. Does the PC always have something to
say to the NPC?
>Also, I think the doubled prompt is misleading: it implies that there
>is some different mode of interaction, when really all the normal
>commands (TAKE, LOOK, etc.) are still valid. More generally
>(although this may be a conscious design decision on your part), I
>don't think you need a separate conversation mode at all; if there
>are menu options available, they're valid commands, and if not, then
>regular commands are still valid.
This is an interesting point. There is some modality, since either the
game accepts single-digit commands or it doesn't, but perhaps the
double prompt overemphasizes this. And with the menu at the bottom,
the player will see it. I think I'm still influenced by my experience
with Pytho's Mask, where I kept mentally shutting out the menu, since
I don't normally keep looking at the status line.
>>b) Can you force it to misbehave or even crash by doing things I
>>haven't thought of trying myself?
>
> Yes. If I am in a no-menu state (such as the beginning of the
>game), and I then begin conversation and the menu opens, and I then
>UNDO or restore to a no-menu state, the menu remains on the screen.
Strange - I thought I had got that to work. Must be some last-minute
change where I undid my own bug fixes.
> Now I will take the opportunity to briefly plug my own system, GxScript,
>which solves all of the nitpicks I mentioned above (except that it requires
>"SAY 1" instead of simply "1", just because I considered that a low priority
>feature and haven't implemented it yet).
Let's see, you wrote SITTM, right? I think that apart from the
"SAY 1" thing, which I found irritating, that game had a very good
conversation system.
I solved the "SAY 1" vs. "1" problem by having BeforeParsing() look
for an input line consisting entirely of digits, and in case it
detects such input, it just inserts "say " at the beginning of
the buffer and re-tokenizes it.
> As an author, I am particularly interested in how your system looks when
>used in a game's source code.
That's the interesting thing, really. The reason I'm writing my own
system rather than using somebody else's is that I want full control
over the API; I find coding conversation difficult enough without
having to think like some other designer :-).
I will release source as soon as the bugs are stamped out.
> I have found it relatively easy to design the
>conversation system in theory, but relatively difficult to find a practical
>form for it (i.e., one that doesn't require writing endless amount of code for
>relatively simple things).
Tell me about it... I think the problem is that we're looking for
the solution to a very general and very difficult problem: how to
represent the evolving state of human-to-human interaction.
> I am especially interested in how you did the
>conversation tree, what kind of control MOOCE gives the author over
>activating/deactivating topics, shifting the subject, etc.
Currently, it's in a rather embryonic state, since the conversation
tree in "Troll Bridge" is quite a simple one.
My problem is that I don't really know yet what the kind of conversations
I want to write require from MOOCE. (It's a chicken-and-egg problem -
I can't start implementing my WIP until I have conversation menus, but
until I get well udnerway I don't know what my characters are going
to say. They tend to develop minds of their own...)
This is in many ways a virgin field, with Emily as a trailblazer in
one direction, but I'm not really after her kind of NPC interaction
right now.
> You are welcome to take a look at GxScript -- the newest version is
>available from my page at
>http://members.aol.com/brenbarn/IF-download.html -- and see the approach I
>took. I'd be glad to hear what you think. (I'd be glad to hear what anybody
>thinks, although I got little response when I first announced GxScript on
>here.)
Will do. I just wanted to get going on the design of MOOCE before I
was too influenced by other designs (which would be hubris if there
were mature systems out there, but with the current state of the
field I think setting off in one's own direction can prove fruitful).
>"Do not follow where the path may lead;
>go, instead, where there is no path, and leave a trail."
> --Author Unknown
Very appropriate in this context. :-)
OKB -- not okblacke
>>the NPC's name centered above the menu options and displayed in a larger
>font
>>size. No offense, but the current display -- "Talking to NPC" displayed
>>left-aligned in a different background color -- looks a bit tacky to me.
>
>Different background colour? You're using strange default settings :-9
>(or perhaps it's your interpreter that does). Time to bring out the
>old stylehints....
Oh, stylehints are essential, I'd think. You definitely have to give the
game author some control over the color of the window, so that he can pick
colors that go well with whatever colors he may have picked for the main and/or
status windows. My system, by default, explicitly sets the menu window color
scheme to the same settings as in the main window.
This reminds me: another thing I did in GxScript was let the game author
specify a color and background color for each NPC, so that the menus would be
"color-coded" differently for different NPCs. I think this is a useful thing
to have if there are lots of NPCs, although it's obviously fairly meaningless
if there's only 1.
>> As far as the interaction, it might be worthwhile putting in some code
>>that automatically enters conversation with the NPC if there is only one NPC
>in
>>the room.
>
>Or at least code that gives the game designer the possibility to do
>that. Because I'm not certain I want my games to behave that way.
>
>I think it's rather a deep issue, actually, that depends on how you
>think about PC/NPC interaction. Does the PC always have something to
>say to the NPC?
Well, the way I did it was to always display a menu if there was even a
single NPC in the room. The menu might be empty, but it would be there. If
the menu wasn't present at all, that meant there were no NPCs to be seen. If
there is just 1 NPC, I find it annoying to have to type "TALK TO NPC" just to
see the menu (especially if it's obvious that this one NPC is the focus of the
scene).
>>Also, I think the doubled prompt is misleading: it implies that there
>>is some different mode of interaction, when really all the normal
>>commands (TAKE, LOOK, etc.) are still valid. More generally
>>(although this may be a conscious design decision on your part), I
>>don't think you need a separate conversation mode at all; if there
>>are menu options available, they're valid commands, and if not, then
>>regular commands are still valid.
>
>This is an interesting point. There is some modality, since either the
>game accepts single-digit commands or it doesn't, but perhaps the
>double prompt overemphasizes this. And with the menu at the bottom,
>the player will see it.
Right. The very presence of the menu can itself be the clue that digits
are acceptable input.
>> Now I will take the opportunity to briefly plug my own system,
>GxScript,
>>which solves all of the nitpicks I mentioned above (except that it requires
>>"SAY 1" instead of simply "1", just because I considered that a low priority
>>feature and haven't implemented it yet).
>
>Let's see, you wrote SITTM, right? I think that apart from the
>"SAY 1" thing, which I found irritating, that game had a very good
>conversation system.
Yes, I did, and thanks. :-)
>> I am especially interested in how you did the
>>conversation tree, what kind of control MOOCE gives the author over
>>activating/deactivating topics, shifting the subject, etc.
>
>Currently, it's in a rather embryonic state, since the conversation
>tree in "Troll Bridge" is quite a simple one.
>
>My problem is that I don't really know yet what the kind of conversations
>I want to write require from MOOCE. (It's a chicken-and-egg problem -
>I can't start implementing my WIP until I have conversation menus, but
>until I get well udnerway I don't know what my characters are going
>to say. They tend to develop minds of their own...)
Don't I know it :-). I tried to implement as many features as I could up
front, but I was still adding and tweaking things in the early stages of
writing the NPCs for SITTM.
>This is in many ways a virgin field, with Emily as a trailblazer in
>one direction, but I'm not really after her kind of NPC interaction
>right now.
I've been thinking about this. Emily has mentioned that part of her
conversation system is a knowledge-modelling type of thing, and in this respect
her focus differs from my own (and, apparently, yours). I've been more
concerned with making a system that will let me lay out, as easily as possible,
a conversation tree with fixed (albeit numerous) paths through it. I'm laying
out the whole conversation, along with the specific instances of "saying this
opens up this topic", rather than laying out a series of topics and a general
ruleset that determines how it can be traversed.
--OKB (Bren...@aol.com) -- no relation to okblacke
"Do not follow where the path may lead;
Gregg V. Carroll
> m...@df.lth.se (Magnus Olsson) wrote:
> Personally, I prefer a menu with a proportional font. For one thing, it
> lets you use different text sizes and use center justification, so you can have
> the NPC's name centered above the menu options and displayed in a larger font
> size. No offense, but the current display -- "Talking to NPC" displayed
> left-aligned in a different background color -- looks a bit tacky to me.
I actually didn't notice the "Talking to NPC" line. I'm a status line
ignorer myself, and not seeing that "Talking to" line stems from that.
My eyes go right for the conversation text. In fact, I don't even note
the number I need until after I've picked the line I want. I say trash
the "Talking to" part altogether. If you don't remember who you're
talking to, you probably shouldn't be playing interactive fiction. If
people insist on having it, a more graceful option might be to put it
in the actual status line. Ex. "Chasm (talking to the Troll)" or
"Chasm, talking to the troll." Keep all the status stuff together. If
there's a situation where you're in a bed, say, talking to someone, it
should say "Bedroom, in bed, talking to Mom." or something similar.
Maybe "Bedroom (in bed), talking to Mom" or "Bedroom, in bed (talking
to Mom)."
> As far as the interaction, it might be worthwhile putting in some code
> that automatically enters conversation with the NPC if there is only one NPC in
> the room.
I disagree. I think that would be a mistake, *unless* the game has the
NPC start the conversation (prompting it, basically), otherwise, use
the TALK TO command first. I don't mind conversation menus, but I
personally dislike the idea of a game railroading me into a
conversation, even if it's under a special case and is the obvious
thing to do.
But then, Galatea has been, and remains, my favorite NPC interface and
NPC period, so that says a lot right there.
> >b) Can you force it to misbehave or even crash by doing things I
> >haven't thought of trying myself?
I didn't try breaking the game, but I did note that this odd sequence
of conversation took place:
>>1
"Guarding the bridge? Against what? Theft?"
The troll chuckles. "Hardly. I'm guarding it against cheapskate
adventurers who think they can pass without paying."
>>2
"Do I really have to pay to cross the bridge?"
The troll chuckles. "Of course. Everything has its price."
>>1
"Doesn't that get boring?"
"It does, but nowadays I can always post to Usenet."
The flow of conversation sort of implies that "Doesn't that get
boring" is refering to the fact that "Everything has its price,"
rather than the guarding of the bridge. I *know* it doesn't but it
still is a bit awkward, which is part of the problem I have with this
kind of system to begin with.
Besides that, the menu at the bottom is a good thing. The command
input is right there, so my line of sight doesn't have to wander to
see the menu.
The double-prompt thing... eh. Can't decide. Since the menu's there, I
don't think you need it. And you can perform other commands, so it's
not as if it's prompting you of a "locked out" state from normal
commands. Guess it should go.
I think you should keep "Say 1" optional. As you can see above, I
don't bother with the "Say" myself. If you have a menu, it seems
rather pointless. "Say" is only useful if there's no menu and you have
to prompt the parser to "say" a manually-typed input. Definitely don't
make it mandatory, keeping it optional is probably a good move though,
since different players might be used to using menus different ways.
That's all I can think of. Felt natural overall, as conversation
interfaces go.
What if the menu system said, instead of literal quoted options,
something like:
1 A troll interested in Usnet? What's that all about?
2 Give an ugly, vile troll money? Yeah right.
3 [End conversation.]
> 2
"Stand aside foul creature...!"
Blah blah blah...
>1
"Usnet? Really? What newsgroups..."
Blah blah blah.
I only bring it up because it seems somewhat repetative to re-display
the quote of the conversation after it's chosen. Just a thought.
Late night thoughts from a long time
player and newbie programmer,
Gregg
Magnus Olsson
has much more general interest than in conjunction with my feeble
attempts at modelling conversation).
In article <20011201190943...@mb-fo.aol.com>,
OKB -- not okblacke <bren...@aol.comRemove> wrote:
>m...@df.lth.se (Magnus Olsson) wrote:
>>> As far as the interaction, it might be worthwhile putting in some code
>>>that automatically enters conversation with the NPC if there is only one NPC
>>in
>>>the room.
>>
>>Or at least code that gives the game designer the possibility to do
>>that. Because I'm not certain I want my games to behave that way.
>
> Well, the way I did it was to always display a menu if there was even a
>single NPC in the room. The menu might be empty, but it would be there. If
>the menu wasn't present at all, that meant there were no NPCs to be seen. If
>there is just 1 NPC, I find it annoying to have to type "TALK TO NPC" just to
>see the menu (especially if it's obvious that this one NPC is the focus of the
>scene).
I think this is a game design issue, rather than a library design
issue. In one game, the behaviour you're sugesting may be
appropriate; in another, it may not.
The major problem I have with it would be in a situation like this:
the PC is out doing something with his sidekick NPC. Most of the time,
both are quite busy, so there's no time for conversation, and the
menu is empty. Then there suddenly is a real reason for the player
to ask the sidekick something ("Suddenly it struck me that Gabrielle
may have heard something about the feeding habits of chimerae").
If there suddenly appears a line in the previously empty conversation
menu, the game is essentially prompting me "Look, now would be
a good time to ask Gabrielle something". But what if I want the
player to think of this herself? I may even want to make this an
actual puzzle, and then SITTM-style conversation would give the solution
away.
Similarly, there's a convention in IF that if the player finds
some berries, it's considered a bit pushy to give them a description
like "The berries really look yummy. It might be a good idea to eat
them."
>>This is an interesting point. There is some modality, since either the
>>game accepts single-digit commands or it doesn't, but perhaps the
>>double prompt overemphasizes this. And with the menu at the bottom,
>>the player will see it.
>
> Right. The very presence of the menu can itself be the clue that digits
>are acceptable input.
Agreed, and the ">>" prompt is gone in release 2 of "Troll Bridge".
(about the modelling of conversations, conversation trees, and
their care and pruning)
>>This is in many ways a virgin field, with Emily as a trailblazer in
>>one direction, but I'm not really after her kind of NPC interaction
>>right now.
>
> I've been thinking about this. Emily has mentioned that part of her
>conversation system is a knowledge-modelling type of thing, and in this respect
>her focus differs from my own (and, apparently, yours). I've been more
>concerned with making a system that will let me lay out, as easily as possible,
>a conversation tree with fixed (albeit numerous) paths through it. I'm laying
>out the whole conversation, along with the specific instances of "saying this
>opens up this topic", rather than laying out a series of topics and a general
>ruleset that determines how it can be traversed.
My goals are the same as yours, but only because I don't think my
WIP needs very deep conversations - it will be plot-driven, albeit
with more NPC interaction than I've attempted before.
But in a more general context, I think conversation trees aren't
a satisfying solution.
The thing is that the problem is one of modelling knowledge. Real-life
conversations don't exhibit much tree-like structure. My feeling is
that a more realistic NPC would have a pool of subjects and associated
knowledge and interests. Which topics are available at any time
depends on the history of the conversation, and on the current
knowledge and interests of the NPC, and on her knowledge of the
conversation partner's knowledge and interests (I think there's a lot
to the idea that modelling the internal state of other sentient beings
- empathy, if you like - is at the core of intelligent behaviour). But
there's no tree structure.
But conversation trees are still useful, as one kind of model of the
internal state of the NPC. A very crude model, but it will probably be
enough for simple IF conversations.
The problem is that it's a pain to maintain the tree and keep it
updated according to the game state. And I *think* this is inevitable,
giving the crudeness of the model - if you're using a crude model that
doesn't fit the phenomenon you're modelling very well, you'll have to
be prepared to do a lot of manual tweaking to keep it from diverging
from "reality". (I could make some interesting analogies to
theoretical physics - which is all about finding appropriate models
for physical systems - here, but I'll desist).
Magnus Olsson
Gregg V. Carroll <gr...@midcoast.com> wrote:
>bren...@aol.comRemove (OKB -- not okblacke) wrote in message
>news:<20011201165819...@mb-fc.aol.com>...
>> m...@df.lth.se (Magnus Olsson) wrote:
>I say trash
>the "Talking to" part altogether. If you don't remember who you're
>talking to, you probably shouldn't be playing interactive fiction.
As long as there's only one NPC around, it shouldn't be a problem,
but what if there are several, and the player has just restored
from an old save file?
>If
>people insist on having it, a more graceful option might be to put it
>in the actual status line. Ex. "Chasm (talking to the Troll)" or
>"Chasm, talking to the troll."
That's an interesting suggestion. Does anybody else have any thoughts
on this?
>>>1
>"Guarding the bridge? Against what? Theft?"
>The troll chuckles. "Hardly. I'm guarding it against cheapskate
>adventurers who think they can pass without paying."
>
>>>2
>"Do I really have to pay to cross the bridge?"
>The troll chuckles. "Of course. Everything has its price."
>
>>>1
>"Doesn't that get boring?"
>"It does, but nowadays I can always post to Usenet."
>
>The flow of conversation sort of implies that "Doesn't that get
>boring" is refering to the fact that "Everything has its price,"
>rather than the guarding of the bridge. I *know* it doesn't but it
>still is a bit awkward, which is part of the problem I have with this
>kind of system to begin with.
This is simply a matter of bad phrasing on my part, and it's of
course a symptom of the usual old trap that designers fall into:
forgetting that people may do things in a different order from
that in which they were designed.
I've changed the wording in release 2 of "Troll Bridge", to
"Doesn't guard duty get boring?". A more sophisticated solution
would be have several alternative formulations of the same
question depending on what has been said before.
>I think you should keep "Say 1" optional.
I will *definitely* keep it optional. Having to type "say" all
the time in SITTM was a source of irritation.
Magnus Olsson
>something like:
>
>1 A troll interested in Usnet? What's that all about?
>2 Give an ugly, vile troll money? Yeah right.
>3 [End conversation.]
>
>> 2
>"Stand aside foul creature...!"
>Blah blah blah...
I'm a bit in two minds about this.
On one hand, it may make the player feel less constrained, since the
alternatives can be made more general (in the extreme case, something
like "Tell Baldrick that his plan sounds OK"/"His plan sounds like
utter rubbish"); my own experience with conversation-menu games is that
sometimes you feel like "My character would never say *any* of those
horrible lines".
On the other hand, the game would literally be putting words in the
PC's mouth. "Hey, I just wanted to tell him to keep a low profile,
I didn't mean to cast doubts on his ancestry!" This can be a *big*
mimesis-breaker. It can also be used to great effect in some cases;
comic effects, if nothing else.
And there are of course games where the onyl conversation command
is "talk to NPC" which sets off a whole dialogue.
A further advantage of non-specific menu entries is that they can
refer to long speeches, rather than just to one-line quips.
Another way of doing that is to have abbreviated menu entries -
"But what about Plato's thought on this?" would lead to the
PC actually saying "But what about Plato's thoughts on this?
As you're no doubt aware, in his _Defence of Socrates_, he lets
Socrates suggest that..." and so on.
What do people think about this? Which is preferable?
And what about mixing actual lines and short summaries, like
this:
0: End conversation
1: Thank Theron for his help
2: "Theron, I'll never forget what you've done for me."
3: Ask Theron about his views on dialectic materialism.
Would such a menu be confusing?
>I only bring it up because it seems somewhat repetative to re-display
>the quote of the conversation after it's chosen. Just a thought.
Oh, but there's a reason for that: if the quote isn't repeated in the
main window, a transcript of the game will be essentially unreadable,
an the player will not be able to scroll back to find out what he
actually said five moves earlier to make Theron so upset.
M. D. Krauss
m...@df.lth.se (Magnus Olsson) wrote:
> In article <20011201165819...@mb-fc.aol.com>,
> OKB -- not okblacke <bren...@aol.comRemove> wrote:
> >m...@df.lth.se (Magnus Olsson) wrote:
[snip]
> >>b) Can you force it to misbehave or even crash by doing things I
> >>haven't thought of trying myself?
> >
> > Yes. If I am in a no-menu state (such as the beginning of the
> >game), and I then begin conversation and the menu opens, and I then
> >UNDO or restore to a no-menu state, the menu remains on the screen.
>
> Strange - I thought I had got that to work. Must be some last-minute
> change where I undid my own bug fixes.
[snip]
Interesting. I just tried it and this is not a problem. Have you fixed it?
Or is this possibly a terp difference? I'm using um... Whatever terp is
embedded in Simon Baldwin's IFPE on Linux, I think Glulxe 0.3.5.
-M
(Oh by the way... looks pretty good :)
OKB -- not okblacke
>>I say trash
>>the "Talking to" part altogether. If you don't remember who you're
>>talking to, you probably shouldn't be playing interactive fiction.
>
>As long as there's only one NPC around, it shouldn't be a problem,
>but what if there are several, and the player has just restored
>from an old save file?
Even disregarding meta-game considerations, it could be possible for
another NPC to break into the conversation. This would probably be cued in the
main text, but it could be quite disconcerting to see an entirely different set
of choices on the menu and not know why until you read all the main text. (I,
personally, tend to notice the changes in the menu BEFORE reading the main
text.)
>>If
>>people insist on having it, a more graceful option might be to put it
>>in the actual status line. Ex. "Chasm (talking to the Troll)" or
>>"Chasm, talking to the troll."
>
>That's an interesting suggestion. Does anybody else have any thoughts
>on this?
I think it's better to have it in the menu window, centered and boldfaced
over the menu options. If the point is to keep the player apprised of who he's
talking to, there's little point in sticking it up in the status line, since
there seems to be a consensus that fewer people pay attention to that.
>>I think you should keep "Say 1" optional.
>
>I will *definitely* keep it optional. Having to type "say" all
>the time in SITTM was a source of irritation.
As a side note, Magnus was kind enough to show me his BeforeParsing code,
and I've incorporated into GxScript, so when the post-comp release of SITTM
comes out, it will allow digit-only commands.
OKB -- not okblacke
>In article <a5560027.01120...@posting.google.com>,
>Gregg V. Carroll <gr...@midcoast.com> wrote:
>>What if the menu system said, instead of literal quoted options,
>>something like:
>>
>>1 A troll interested in Usnet? What's that all about?
>>2 Give an ugly, vile troll money? Yeah right.
>>3 [End conversation.]
>>
>>> 2
>>"Stand aside foul creature...!"
>>Blah blah blah...
>
>I'm a bit in two minds about this.
I tend to prefer a menu that shows the actual words I'm going to say,
because I think that, in conversation, HOW something is said can be just as
important as WHAT is said. There is the option of giving different choices
like:
1: Gently ask the troll about Usenet.
2: Harshly interrogate the troll about Usenet.
3: Beat the information out of the troll.
. . . but I think if we're going to go the route of different tones, it's
more interesting to see the actual dialogue:
1: "Dear troll, please tell me about Usenet."
2: "Tell me about Usenet RIGHT NOW!"
3: "Tell me everything you know about Usenet before I break your nose."
>A further advantage of non-specific menu entries is that they can
>refer to long speeches, rather than just to one-line quips.
>Another way of doing that is to have abbreviated menu entries -
>"But what about Plato's thought on this?" would lead to the
>PC actually saying "But what about Plato's thoughts on this?
>As you're no doubt aware, in his _Defence of Socrates_, he lets
>Socrates suggest that..." and so on.
SITTM did this to some extent, although it was generally more the type of
thing where choosing "But what about Plato's thoughts on this?" would lead to:
"But what about Plato's thoughts on this?"
"What do you mean?"
"Well, in his _Defense of Socrates_, he says blah blah blah."
"Oh yeah? Well yadda yadda yadda?"
"Your mother was a hamster, etc."
. . . sort of like eliding several menu-choice breakpoints.
>What do people think about this? Which is preferable?
I think it's really an authorial decision that's going to be specific to
the game and its state. If the player chooses an option which refers to
something so crucial that you don't want to let him abandon that line of
questioning and start talking about rutabagas, you can spit out a more lenghty
dialogue. Similarly if the option he chooses is something that, in the context
of the game, is obviously going to lead to a culturally pre-scripted sequence
("Hi." "Hello." "How are you?" "Fine, you?" "Good. And the missus?") -- and
you don't want the player to disrupt this.
>And what about mixing actual lines and short summaries, like
>this:
>
>0: End conversation
>1: Thank Theron for his help
>2: "Theron, I'll never forget what you've done for me."
>3: Ask Theron about his views on dialectic materialism.
>
>Would such a menu be confusing?
I don't think so. It could be interesting, although I think in most cases
I'd prefer seeing the actual dialogue. In this example, however, #3 seems just
the kind of thing that you might want to summarize rather than spell out.
OKB -- not okblacke
always display a menu if there was even a
>>single NPC in the room. The menu might be empty, but it would be there. If
>>the menu wasn't present at all, that meant there were no NPCs to be seen.
>If
>>there is just 1 NPC, I find it annoying to have to type "TALK TO NPC" just
>to
>>see the menu (especially if it's obvious that this one NPC is the focus of
>the
>>scene).
>
>I think this is a game design issue, rather than a library design
>issue. In one game, the behaviour you're sugesting may be
>appropriate; in another, it may not.
>
>The major problem I have with it would be in a situation like this:
>the PC is out doing something with his sidekick NPC. Most of the time,
>both are quite busy, so there's no time for conversation, and the
>menu is empty. Then there suddenly is a real reason for the player
>to ask the sidekick something ("Suddenly it struck me that Gabrielle
>may have heard something about the feeding habits of chimerae").
>
>If there suddenly appears a line in the previously empty conversation
>menu, the game is essentially prompting me "Look, now would be
>a good time to ask Gabrielle something". But what if I want the
>player to think of this herself? I may even want to make this an
>actual puzzle, and then SITTM-style conversation would give the solution
>away.
Hmm, an interesting point. Personally, my approach would be to not leave
the menu empty, so that that contrast doesn't arise. Even at the busy times
when the two pals are pulling the ears off a Gondark, I'd tend to throw in
buddy-buddy lines or something ("All in a day's work, eh?" "This reminds of
that time in Ecuador.").
More generally, I think this kind of thing runs perilously close to guess
the verb (the verb in this case being TALK TO). I think it would be preferable
to show the menu, but not make it obvious that that is what should be done --
either as I described above, or by hinting at other possible actions apart from
conversation ("Out of the corner of my eye, I saw the sun gleam on something in
the underbrush.") so that the player is left to choose between several
apparently equally useful actions.
I should note that GxScript DOES allow the game author a choice in this
matter. You can choose to have the menu window always open unless there are no
NPCs in the room (which is what SITTM did); you can choose to have it always
open no matter what; or you can choose to have it always resize itself to be
exactly as big as it needs to be to hold the current menu options (so that if
there's nothing to say to that particular NPC, the menu window will disappear).
>The thing is that the problem is one of modelling knowledge. Real-life
>conversations don't exhibit much tree-like structure. My feeling is
>that a more realistic NPC would have a pool of subjects and associated
>knowledge and interests. Which topics are available at any time
>depends on the history of the conversation, and on the current
>knowledge and interests of the NPC, and on her knowledge of the
>conversation partner's knowledge and interests (I think there's a lot
>to the idea that modelling the internal state of other sentient beings
>- empathy, if you like - is at the core of intelligent behaviour). But
>there's no tree structure.
I don't think that topic availability of the type you refer to necessarily
excludes a tree structure or requires knowledge modelling. To be sure, it
excludes a STRICT tree structure, but this is why GxScript provides things like
adopting and sharing nodes.
We may be using different definitions here. When I say that I don't think
knowledge modelling is necessary, I mean I don't think there needs to be an
abstract sense of the NPC's "knowing kung fu" or "being able to speak Italian".
Certainly, keeping of track of what has and hasn't been said, and when and
where it was or wasn't said, could be considered a primitive type of knowledge
modelling, but that's not what I use the term to mean.
Now, I'm not saying we should throw "real" knowledge-modelling out the
window. I'm just saying that, fundamentally, I'm not sure whether trying to
have the NPC keep track of abstract ideas like "do I know whodunnit?" or "how
did I feel when Eric dumped me?" is going to be more fruitful than simply
keeping track of who said what when. At the current stage of development, I
think this question is largely moot, since the majority of attempts at
knowledge modelling are still being used in such limited scenarios that one
could hardly tell whether it was "real" or "fake" knowledge modelling. (Emily
seems to be the exception here, since from what I've heard she's actually
modelling NPC knowledge/emotional state separate from the conversation state.)
A few months ago, we were discussing the usefulness of uselessness of
making artificially-intelligent NPCs. Basically, I think that trying to make
intelligent NPCs that "know" and "think" is not going to be a better solution
than carefully crafting "dumb" NPCs. I could be wrong, and maybe in 6 months
I'll be contradicting myself in a similar thread. We'll see.
Marnie Parker
>and Troll Bridge)
>From: bren...@aol.comRemove (OKB -- not okblacke)
>Date: 12/2/2001 9:57 AM Pacific Standard Time
>1: Gently ask the troll about Usenet.
>2: Harshly interrogate the troll about Usenet.
>3: Beat the information out of the troll.
>1: "Dear troll, please tell me about Usenet."
>2: "Tell me about Usenet RIGHT NOW!"
>3: "Tell me everything you know about Usenet before I break your nose."
Either set is fine with me. I almost prefer the first set, because then there
is a bit of surprise when I read theactual words that I, the PC, say. Otherwise
it's just a boring repeat.
However, I agree that tone is important. Options should always allow for
varying emotional responses from the PC.
Doe :-)
doea...@aol.com
IF http://members.aol.com/doepage/intfict.htm
(An Iffy Theory | Glulx/Glk for Duncies | unglklib | Inform Primer)
IF Art Gallery http://members.aol.com/iffyart/
IF Review Conspiracy http://www.plover.net/~textfire/conspiracy/
Eytan Zweig
"Marnie Parker" <doea...@aol.com> wrote in message
news:20011202135309...@mb-cj.aol.com...
> >Subject: Re: Putting words in the PC's mouth (was: [ANNOUNCE] MOOCE
(alpha)
> >and Troll Bridge)
> >From: bren...@aol.comRemove (OKB -- not okblacke)
> >Date: 12/2/2001 9:57 AM Pacific Standard Time
>
> >1: Gently ask the troll about Usenet.
> >2: Harshly interrogate the troll about Usenet.
> >3: Beat the information out of the troll.
>
> >1: "Dear troll, please tell me about Usenet."
> >2: "Tell me about Usenet RIGHT NOW!"
> >3: "Tell me everything you know about Usenet before I break your nose."
>
> Either set is fine with me. I almost prefer the first set, because then
there
> is a bit of surprise when I read theactual words that I, the PC, say.
Otherwise
> it's just a boring repeat.
>
> However, I agree that tone is important. Options should always allow for
> varying emotional responses from the PC.
>
I think that I prefer the first set, as long as the actual words say come
close to what the PC's text actually says. I remember the game "The Pandora
Directive", where player responses were given as "moods" rather than the
actual response - i.e.
Woman: Do you think I look good?
1. Cruel honesty
2. Compliment her Outfit
3. Make a Joke.
The problem was, that you often had no idea how the PC would respond to any
of these choices - for instance, say that you want to stay on her good side,
so you choose 2. You might get:
2. Well, I think your outfit is so beautiful it disguises how hideously ugly
you actually are.
Or something like that - obviously, the opposite effect from what I
intended. It got to the point where I'd load and save before each
conversation (or maybe during - I don't remember if that was possible) just
to make sure I know what all options actually are and get to say what I
really want. It really annoyed the hell out of me.
On the other hand, if the responses reliably describe what you're going to
say, I think that it could be used to a more subtle effect than giving the
actual responses in the menu.
Eytan
Rikard Peterson
news:9udhcg$sff$1...@news.lth.se...
> And what about mixing actual lines and short summaries,
> like this:
>
> 0: End conversation
> 1: Thank Theron for his help
> 2: "Theron, I'll never forget what you've done for me."
> 3: Ask Theron about his views on dialectic materialism.
>
> Would such a menu be confusing?
I think so, yes. Maybe it's just me that's easily confused, but I prefer
either way. Which one doesn't matter that much (different approaches
would be suitable for different games), as long as it's consequent.
Rikard
Gregg V. Carroll
> >What if the menu system said, instead of literal quoted options,
> >something like:
> >
> >1 A troll interested in Usnet? What's that all about?
> >2 Give an ugly, vile troll money? Yeah right.
> >3 [End conversation.]
> >
> >> 2
> >"Stand aside foul creature...!"
> >Blah blah blah...
> I'm a bit in two minds about this.
>
> On one hand, it may make the player feel less constrained, since the
> alternatives can be made more general (in the extreme case, something
> like "Tell Baldrick that his plan sounds OK"/"His plan sounds like
> utter rubbish"); my own experience with conversation-menu games is that
> sometimes you feel like "My character would never say *any* of those
> horrible lines".
Right. When I suggested the above (they are bad examples, just off the
top of my head late last night) I was thinking that maybe it could
play off the player's (not the PC's) internal dialogue as s/he was
thinking about what to say, rather than giving explicit quotes.
(I see your point about repeating the lines for the transcript, btw,
I'd forgotten about that - I tend not to use transcripts all that
often. I'm using them now for AMFV but only because it's almost
mandatory for that game. Anyway.)
This has been brought up before by others, but I tend to dislike
conversation menus. Mind you, I won't refuse to play a game that uses
them, or not enjoy one that does, but I prefer the
>TALK TO [NPC]
>ASK [NPC] ABOUT
>TELL [NPC] ABOUT
>SAY, "SOMETHING"
method of interacting with NPCs. It *is* a simplistic trigger, really,
in many ways not any more "sophisticated" than a menu, because you'll
almost never type,
>ASK ROBERT ABOUT THE DIFFERENCE BETWEEN A R-12 CONVERTER DRIVE AND AN
R-15 CONVERTER DRIVE.
"Well, the R-12 is sufficient most of the time, but..." etc.
It's more along the lines of,
>ASK ROBERT ABOUT CONVERTERS.
"So what's the difference between these
converters anyway?" you ask.
"Well, the R-12 is sufficient most of the time, but..." etc.
but it still puts the burden of figuring out what to talk about on me,
rather than handing options to me on a menu. I use "burden" in the
kindest way, meaning that the NPC conversations are a kind of meta
puzzle.
So (for me) any time a conversation menu is presented in such a way as
to make it feel less like it is railroading me into a specific line of
dialouge, the better. Or makes it less explicit what conversation path
I'm going down is by presenting specific lines of dialogue. In real
life, sometimes what you think, and what comes out your mouth in an
attempt to express what you're thinking, don't exactly mesh 100% of
the time.
In other words, make the menu descriptions (not quotes) vauge enough
that they don't, in the mind of the player, predict a path ahead of
time, but specific enough that they clearly convey a general direction
of conversation. Some would argue that's making it unfair for the
player, whereas I would argue that's making it a *challenge* rather
than a spoon fed conversation. Coversation menus (again, for me) have
the tendency to feel like a block of conversation text that is being
segmented and presented as an illusionary interaction.
OR, mix and match, as in the one example you had, in the cases where a
specfic quote is effectively required to express intent,
"So I hear you're moving out of your apartment," Robert says.
1 How does he know I'm moving? I haven't told anybody.
2 "Yeah, the neighbors are getting on my nerves."
3 Why can't Robert mind his own damn business for a change?
3 "Yep."
4 [End conversation]
> 1
You pause for a moment, taken by surprise at Robert's
knowledge of your personal life. "No, I'm not moving," you
lie, trying to appear unfazed, "What ever gave you that idea?"
"Oh, well," he says, looking slightly guilty, "you know how
small towns are."
>2
"Yeah, the neighbors are getting on my nerves," you say, the
bitterness in your voice clear.
"Aren't your neighbors the Johnsons? He used to be a good
friend of mine in college, I can't imagine that you wouldn't get
along with him." Robert says, giving you a look like you had
suddenly sprouted a second head.
>3
"I'd rather not talk about it, if you don't mind."
"No problem," Robert says, "Just curious, that's all."
>4
"Yep." you say, rather curtly.
"Any particular reason why?" Robert presses, and you simply shrug.
He raises and eyebrow at your indifference, but seems reluctant to
force an answer out of you.
To contrast two of Emily's games, I spent a lot of time looking at the
details of the conversation in Galatea in order to figure out how I
wanted to guide the conversation, and figure out what looked like an
interesting line of questioning, and so on. I had to really "listen"
to Galatea and what she was saying and how she said it. In Best Of
Three, however, I flew through the conversation menus, and ultimately
felt it was a rather unengaging expereince. I don't remember a lot of
it, but I still remember minute details about Galatea, her subtle
physical movements, what she talked about, specific references, etc.
(and I would add that this is not a criticism of Emily's work or her
writing, but only an account of my subjective experience as a player,
nothing more).
Just by virtue of having to read and re-read dialouge made Galatea a
better experience for me. But then, I was *interested* to begin with,
so it was a welcome challenge and not a point of frustration. Perhaps
conversation menus make the game more accessible to people who
wouldn't ordinarily be bothered, but I'm not sure that is a good
thing.
I'm of the opinion that anything that can be done to menus to help the
player pay more attention (and to be more involved in the game) is a
good thing, but at the same time, I wouldn't want to imply via the
above that all NPCs should be as well-rounded as in Galatea. That, of
course, would be beyond the scope of any reasonably sane game
designer. :)
And I feel myself drifting off topic, so I'll stop there. :)
Gregg
Magnus Olsson
M. D. Krauss <MDKraus...@home-nospam.com-nospam> wrote:
>> > Yes. If I am in a no-menu state (such as the beginning of the
>> >game), and I then begin conversation and the menu opens, and I then
>> >UNDO or restore to a no-menu state, the menu remains on the screen.
>>
>> Strange - I thought I had got that to work. Must be some last-minute
>> change where I undid my own bug fixes.
>
I you're running release 2 of "Troll Bridge" then I've fixed it; if
you're running release 1 then I'm confused.
Magnus Olsson
OKB -- not okblacke <bren...@aol.comRemove> wrote:
>>the PC is out doing something with his sidekick NPC. Most of the time,
>>both are quite busy, so there's no time for conversation, and the
>>menu is empty. Then there suddenly is a real reason for the player
>>to ask the sidekick something ("Suddenly it struck me that Gabrielle
>>may have heard something about the feeding habits of chimerae").
>>
>>If there suddenly appears a line in the previously empty conversation
>>menu, the game is essentially prompting me "Look, now would be
>>a good time to ask Gabrielle something". But what if I want the
>>player to think of this herself? I may even want to make this an
>>actual puzzle, and then SITTM-style conversation would give the solution
>>away.
>
> Hmm, an interesting point. Personally, my approach would be to not leave
>the menu empty, so that that contrast doesn't arise. Even at the busy times
>when the two pals are pulling the ears off a Gondark,
Hey! Wrong universe! :-)
>I'd tend to throw in
>buddy-buddy lines or something ("All in a day's work, eh?" "This reminds of
>that time in Ecuador.").
Yes, this may very well work, at least if
chatting-while-pulling-ears-off-gondarks is in character for the PC.
> More generally, I think this kind of thing runs perilously close to guess
>the verb (the verb in this case being TALK TO).
As usual, we're running between the Scylla of over-prompting and the
Charybdis of guess-the-verb. The fact that talking to the NPC is
suddenly going to be a good idea would probably have to be
clued. Something like this: PC and Gabrielle are exploring the lair of
the chimaera, when Gabrielle makes the offhand remark, "This reminds
me of this guy I met, who actually kept a pet chimaera."
In this situation, I think "talk to gabrielle" would be a fairly
obvious thing to try - but not so blindingly obvious as if the
already-open conversation menu changed.
I think it would be preferable
>to show the menu, but not make it obvious that that is what should be done --
>either as I described above, or by hinting at other possible actions apart from
>conversation ("Out of the corner of my eye, I saw the sun gleam on something in
>the underbrush.") so that the player is left to choose between several
>apparently equally useful actions.
I think this, by the way, is a good way of giving the player the
illusion of choice: there was one moment in _All Roads_ where the game
hinted (or seemed to hint) at several possible actions, but there
simply wasn't time to try more than one of them - not time as in the
PC running out of time, but because the plot then devloped in another
direction and pulled the PC along with it.
>>The thing is that the problem is one of modelling knowledge. Real-life
>>conversations don't exhibit much tree-like structure. My feeling is
>>that a more realistic NPC would have a pool of subjects and associated
>>knowledge and interests. Which topics are available at any time
>>depends on the history of the conversation, and on the current
>>knowledge and interests of the NPC, and on her knowledge of the
>>conversation partner's knowledge and interests (I think there's a lot
>>to the idea that modelling the internal state of other sentient beings
>>- empathy, if you like - is at the core of intelligent behaviour). But
>>there's no tree structure.
>
> I don't think that topic availability of the type you refer to necessarily
>excludes a tree structure or requires knowledge modelling.
Well, what I meant was that by imposing a tree structure, we do some
knowledge modelling, only that the model is rather crude.
>To be sure, it
>excludes a STRICT tree structure, but this is why GxScript provides things like
>adopting and sharing nodes.
Sharing nodes is nice. I'll have to look at how you do that.
> We may be using different definitions here. When I say that I don't think
>knowledge modelling is necessary, I mean I don't think there needs to be an
>abstract sense of the NPC's "knowing kung fu" or "being able to speak Italian".
Agreed.
> Certainly, keeping of track of what has and hasn't been said, and when and
>where it was or wasn't said, could be considered a primitive type of knowledge
>modelling, but that's not what I use the term to mean.
I see.
> Now, I'm not saying we should throw "real" knowledge-modelling out the
>window. I'm just saying that, fundamentally, I'm not sure whether trying to
>have the NPC keep track of abstract ideas like "do I know whodunnit?" or "how
>did I feel when Eric dumped me?" is going to be more fruitful than simply
>keeping track of who said what when.
I think it depends on the complexity of what we're trying to achieve.
I think that the simpler approach will sooner or later run into a
quagmire of exponentially increasing complexity, as the complexity of
the NPCs increases.
>At the current stage of development, I
>think this question is largely moot, since the majority of attempts at
>knowledge modelling are still being used in such limited scenarios that one
>could hardly tell whether it was "real" or "fake" knowledge modelling.
--
M. D. Krauss
m...@df.lth.se (Magnus Olsson) wrote:
> In article <20011202123817.490e...@home-nospam.com-nospam>,
> M. D. Krauss <MDKraus...@home-nospam.com-nospam> wrote:
> >> > Yes. If I am in a no-menu state (such as the beginning of the
> >> >game), and I then begin conversation and the menu opens, and I then
> >> >UNDO or restore to a no-menu state, the menu remains on the screen.
> >>
> >> Strange - I thought I had got that to work. Must be some last-minute
> >> change where I undid my own bug fixes.
> >
> >Interesting. I just tried it and this is not a problem. Have you fixed it?
>
> I you're running release 2 of "Troll Bridge" then I've fixed it; if
> you're running release 1 then I'm confused.
Right, okay, it says release 2, so I guess that clears that up. :)
-M
> --
> Magnus Olsson (m...@df.lth.se, m...@pobox.com)
> ------ http://www.pobox.com/~mol ------
--
To email me convert my address to something resembling reason
Gregg V. Carroll
> In article <a5560027.01120...@posting.google.com>,
> Gregg V. Carroll <gr...@midcoast.com> wrote:
> >bren...@aol.comRemove (OKB -- not okblacke) wrote in message
> >news:<20011201165819...@mb-fc.aol.com>...
> >> m...@df.lth.se (Magnus Olsson) wrote:
> >I say trash
> >the "Talking to" part altogether. If you don't remember who you're
> >talking to, you probably shouldn't be playing interactive fiction.
>
> As long as there's only one NPC around, it shouldn't be a problem,
> but what if there are several, and the player has just restored
> from an old save file?
I suppose. See, the thing is, I see a lot of those kinds of design
problems as being somewhat dated, in light of UNDO. If you didn't
remember who you were talking to from a saved file, you could just
experiment with a few entries and then undo them if it's not what you
were planning to do when you saved. If you were playing an Infocom
game, which would require an additional RESTORE to correct a mistake,
yes, that would be tedious.
(I guess it could be argued that an interface design that depends on
UNDO is similar to forcing the player to die before figuring out the
solution to a puzzle, but on the other hand, one wouldn't nesc.
*always* forget who they were talking to inbetween saves, so it's not
a complete dependency, IMHO.)
As far as having multiple NPCs in the room, I should think that the
display text should make it clear enough who you are engaged in a
conversation with, and/or what NPCs are talking to each other. I found
this to be something of a problem in Best Of Three, mostly because of
formatting choices. I found it difficult at times to follow who was
saying what, and that's with only one NPC and one PC (the waiter
nonwithstanding).
It requires the use of "you say" and "[Name] says" type clarifications
to the dialogue. Even more so than in static fiction, because there,
you can get away with "he says," but if you're restoring from a save
in IF, and there's multiple male NPCs in the room, than which "he" is
being talked about? It might make the dialogue top heavy with "Bill
says" or "Robert says" type phrases, but on the other hand, it's such
a convention in fiction that it might not even be noticable.
In any case, all of the above aside, there's nothing particularly
wrong with making it easier on the player by explicitly displaying who
they are in a conversation with. I guess what I'm concerned with is
less to do with interface design and more about game design, and that
the "talking to [NPC]" status line might accidentaly encourage IF
writers to write conversation display text that is not as clear as it
should be (as clear as it would need to be if it were static fiction)
> >If
> >people insist on having it, a more graceful option might be to put it
> >in the actual status line. Ex. "Chasm (talking to the Troll)" or
> >"Chasm, talking to the troll."
Someone else pointed out that it seems to be that few of us pay
attention to the status line, and I am one of those people, but I *do*
use it, when I need clarification about something. So even if people
don't normally look up there, they will if the "Talking to [NPC]" is
supplied information there, but only when they need that information
(ie, they have forgotten who they are talking to).
On the other hand, "Bedroom, in bed" is put in the status line because
"in bed" is directly tied to your loction, whereas you could talk to
the same NPC in any number of rooms. Conversation is not a function of
your location, it is a function of which NPC you're talking to,
therefore, it should be a part of the menu text.
So I guess it depends on which you find more appealing; keeping things
neat and in one location on the screen (the status line) or where the
"talking to [NPC]" should logically go (in the menu).
> >The flow of conversation sort of implies that "Doesn't that get
> >boring" is refering to the fact that "Everything has its price,"
> >rather than the guarding of the bridge. I *know* it doesn't but it
> >still is a bit awkward, which is part of the problem I have with this
> >kind of system to begin with.
> This is simply a matter of bad phrasing on my part, and it's of
> course a symptom of the usual old trap that designers fall into:
> forgetting that people may do things in a different order from
> that in which they were designed.
Right, and I apologize for taking up so much text to point out a flaw
in the example game, which was fairly irrelevent in light of the fact
that it was a quickie example to showcase the interface.
> I've changed the wording in release 2 of "Troll Bridge", to
> "Doesn't guard duty get boring?". A more sophisticated solution
> would be have several alternative formulations of the same
> question depending on what has been said before.
I noticed. :)
Gregg
OKB -- not okblacke
>I suppose. See, the thing is, I see a lot of those kinds of design
>problems as being somewhat dated, in light of UNDO. If you didn't
>remember who you were talking to from a saved file, you could just
>experiment with a few entries and then undo them if it's not what you
>were planning to do when you saved. If you were playing an Infocom
>game, which would require an additional RESTORE to correct a mistake,
>yes, that would be tedious.
I disagree. . .
>(I guess it could be argued that an interface design that depends on
>UNDO is similar to forcing the player to die before figuring out the
>solution to a puzzle, but on the other hand, one wouldn't nesc.
>*always* forget who they were talking to inbetween saves, so it's not
>a complete dependency, IMHO.)
. . . and I would argue this. The meta-game action of save/restore
shouldn't disadvantage the player any more than necessary. Obviously,
forgetting information from the last session is always possible, but I don't
see that that's any reason to withhold the NPCs name
>I guess what I'm concerned with is
>less to do with interface design and more about game design, and that
>the "talking to [NPC]" status line might accidentaly encourage IF
>writers to write conversation display text that is not as clear as it
>should be (as clear as it would need to be if it were static fiction)
I generally try to mention the NPC's name at least once in each brief bit
of player/NPC dialogue. Also, any author who is concerned about the
comprehensibility of the transcript (which has been brought up here before)
will be paying enough attention to make this clear.
Magnus Olsson
>conversation menus. Mind you, I won't refuse to play a game that uses
>them, or not enjoy one that does, but I prefer the
>
>>TALK TO [NPC]
>>ASK [NPC] ABOUT
>>TELL [NPC] ABOUT
>>SAY, "SOMETHING"
>
>method of interacting with NPCs.
Well, while I share most people's dissatisfaction with menus, I'm also
not very fond of ask/tell. In most cases, it turns the entire
conversation into a huge guess-the-noun puzzle. And it's depressing to
have an otherwise well-crafted NPC give the same stock answer to all
the topics you can think of.
The exception is games like _Galatea_, where a) an enormous amount of
effort has been put into the conversation, and b) the player has
time to do what you describe here:
>I spent a lot of time looking at the
>details of the conversation in Galatea in order to figure out how I
>wanted to guide the conversation, and figure out what looked like an
>interesting line of questioning, and so on. I had to really "listen"
>to Galatea and what she was saying and how she said it.
This works well in _Galatea_, because there isn't anything to do
but converse. But will this work in a fast-paced, plot-driven game?
If you've got three turns to talk to the an NPC, you don't want
to spend those three turns on exploration of dialogue space.
I've played games where I had, say, three turns to extract information
with an ask/tell system. "Rats, those topics didn't work. Restore, try
again. Mimesis, don't let the door hit you on the way out."
Gregg V. Carroll
>> This has been brought up before by others, but I tend to dislike
>> conversation menus. Mind you, I won't refuse to play a game that uses
>> them, or not enjoy one that does, but I prefer the
>>
>>> TALK TO [NPC]
>>> ASK [NPC] ABOUT
>>> TELL [NPC] ABOUT
>>> SAY, "SOMETHING"
>>
>> method of interacting with NPCs.
> Well, while I share most people's dissatisfaction with menus, I'm also
> not very fond of ask/tell. In most cases, it turns the entire
> conversation into a huge guess-the-noun puzzle. And it's depressing to
> have an otherwise well-crafted NPC give the same stock answer to all
> the topics you can think of.
I would argue that it's just as depressing to get a nice, clear answer for
every menu item, yet be prevented from the option of trying, "TELL GRANT TO
BITE ME" in Best Of Three. It may very well be that it isn't implemented,
but at least I have the satisfaction of trying it. And if the game spits
back "Grant doesn't know how to bite you," I have something amusing to
offset the fact that he's being snotty to my PC.
> The exception is games like _Galatea_, where a) an enormous amount of
> effort has been put into the conversation, and b) the player has
> time to do what you describe here:
Right, except that the *only* reason I picked Galatea was to contrast it
with another one of Emily's games, Best Of Three, and my subjective
experience of both. They are the only two games of hers I've played. I said
in my closing lines, something to the effect that only an insane person
would try and implement 5 "Galatea"s in a game that other things to it
besides conversation.
Yes, my enjoyment of Galatea might have been tempered some if she had said,
"I don't know anything about that" more often than not, but I still say that
Emily's writing in that game helped guide the conversation via the display
dialogue, and that should be of a prime concern to any NPC code, regardless
of complexity, or menu / ASK/TELL interface.
[PLANTETFALL spoiler in the next paragraph]
Should there be room to ask off the wall questions not in line with the
topic being discussed? Sure. If I ask Floyd (a robot) how to use the food
dispenser, should I get all bent and cry "guess the verb!" when he says he
doesn't know anything about it? No. If I ask him about the elevators, and
get "Floyd's face lights up. 'Hey! I have a card for one of those!'" and
pops a nice, useful key into your hand, I would call that a success.
>> I spent a lot of time looking at the
>> details of the conversation in Galatea in order to figure out how I
>> wanted to guide the conversation, and figure out what looked like an
>> interesting line of questioning, and so on. I had to really "listen"
>> to Galatea and what she was saying and how she said it.
> This works well in _Galatea_, because there isn't anything to do
> but converse. But will this work in a fast-paced, plot-driven game?
> If you've got three turns to talk to the an NPC, you don't want
> to spend those three turns on exploration of dialogue space.
At that point in a game, either it should be crystal clear what information
you need to get from the NPC, or you should have more than three moves to do
it in.
> I've played games where I had, say, three turns to extract information
> with an ask/tell system. "Rats, those topics didn't work. Restore, try
> again. Mimesis, don't let the door hit you on the way out."
Well, the parser of IF is a tool, but a limited one. If you have a rather
long, but somewhat flimsy board to use as a lever, you don't make someone
lift a boulder with it. That's just bad game design.
In the above example, using a menu, it is customary to obscure the intended
line of questioning with other quotes that don't produce anything useful. So
what's to say you don't just pick menu items / restore until you get hit the
right answer? And would you even know why it's the right answer, or would
you just be plugging in numbers until something worked? And how does that
enhance the game experience? Mimesis is getting its ass smacked with the
door again, IMO.
And if you're forced into that kind of situation, with a limited amount of
moves to extract vital information from a NPC with an obviously flawed
system of interaction (manual or menu) it begs the question, "And the NPC is
important/in my way here because...?"
...the game isn't designed well, the menu, ASK/TELL debate nonwithstanding.
As far as I'm concerned, this all boils down to people getting upset that
the NPCs aren't AI (even if it's just on a subconscious level). Which is the
same as being upset that you can't take a shard of glass in the game,
intended to cut a rope and free your bindings, and stick it in your eye just
because you want to try it and see how "clever" the game is.
When you talk to someone in real life, either it clicks, and you become
friends, or it doesn't and you never see that person again. I think the same
goes for NPCs and interacting with them, you either want to interact with
them or you quit the game, and I'm not sure it has a whole lot to do with
using a menu or ASK/TELL. Or I'm just new, and don't know what the hell I'm
talking about.
To bring up Floyd again, an NPC that does next to nothing except follow you
around until a specific point in the game... plenty of people had a strong
reaction to him. A strong *emotional* reaction, and I have to wonder if
getting bogged down in discussing the pros and cons of menus versus ASK/TELL
makes it more or less likely that we'll see an NPC that can provoke that
kind of reaction in people again.
Gregg
Gregg V. Carroll
> dispenser, should I get all bent and cry "guess the verb!" when he says he
Eh. That should read "guess the noun." My bad.
Gregg
Kathleen
further down>
> Well, while I share most people's dissatisfaction with menus, I'm also
> not very fond of ask/tell. In most cases, it turns the entire
> conversation into a huge guess-the-noun puzzle. And it's depressing to
> have an otherwise well-crafted NPC give the same stock answer to all
> the topics you can think of.
... knowing you are missing out on gobs of text that the author did
write but you can't think of the noun to trigger it. Agreed.
> The exception is games like _Galatea_, where a) an enormous amount of
> effort has been put into the conversation, and b) the player has
> time to do what you describe here:
>
> >I spent a lot of time looking at the
> >details of the conversation in Galatea in order to figure out how I
> >wanted to guide the conversation, and figure out what looked like an
> >interesting line of questioning, and so on. I had to really "listen"
> >to Galatea and what she was saying and how she said it.
>
> This works well in _Galatea_, because there isn't anything to do
> but converse. But will this work in a fast-paced, plot-driven game?
IMHO... No.
_Galatea_ works because the player/PC has no objectives other than
speaking with her, and the gamespace is entirely constructed around
giving you ideas of what to ask her. Every bit of scenery, every line
she says, every bit of backstory are supplied as "bait" to get her to
talk. It's a wonderful illusion (as all great IF is), but (IMO) it
would only work under those very tight constraints.
Set Galatea down in the middle of _Zork 1_ or _Dangerous Curves_ and
you've got a writing/coding/playing nightmare on your hands. The NPC
is no longer just a stand-alone object to be examined, but a potential
tool: "How can I get Galatea to do something to FOR ME - to help ME
reach goals A, B, and C." As soon as that happens, I belive the
illusion will shatter. It's hard to get absorbed in an NPC if you are
thinking all the while: "If I say this, maybe she'll tell me what I
need to know, do what I need her to do. No? Rats. Maybe if say that,
give her this, ask her about that..."
I see it as the difference between reading a book for pleasure, and
reading a book knowing there is going to be a test at end. In the
later, all the while you are reading, you are trying to pick out which
bits you will be querried on during the test and store them for future
access, determine their relevence, draw conclusions... For me, that
spoils the pure joy of reading.
A non-NPC game that produced a similiar "sit back and enjoy it"
response (for me) was SUTW. There again, the gamespace was small and I
(the PC) had no objectives other than to experience it. It was a
wonderful effect and SUTW still ranks as my all time favorite game
because of that. Do I think it would would for a larger game where you
have goals to reach? No, for pretty much the same reason.
> If you've got three turns to talk to the an NPC, you don't want
> to spend those three turns on exploration of dialogue space.
Exactly. And if you have a hundred objects to ask about, which do you
start with?
> Mimesis, don't let the door hit you on the way out."
Kathleen
Adam Thornton
Kathleen <mfis...@aol.com> wrote:
>"How can I get Galatea to do something to FOR ME - to help ME
>reach goals A, B, and C."
> ASK GALATEA TO BITE ME
Adam
Adam Cadre
> but converse. But will this work in a fast-paced, plot-driven game?
> If you've got three turns to talk to the an NPC, you don't want
> to spend those three turns on exploration of dialogue space.
Well, I'll ask: VARICELLA's NPCs had a pretty hefty range of stuff you
could talk to them about -- most of that 500K is dialogue, after all --
and a not incredibly generous amount of time allotted to chat. Did
people talk for a while and then restore, or keep conversation to a
minimum, or what?
For that matter, this thread *is* about "putting words in the PC's mouth,"
something VARICELLA did at least in part with a sentence-generating
algorithm that took into account one's selected tone. Successful?
Unsuccessful? No sense talking about these things entirely as
hypotheticals when we've actually seen them in action already.
-----
Adam Cadre, Brooklyn, NY
http://adamcadre.ac
OKB -- not okblacke
>Well, I'll ask: VARICELLA's NPCs had a pretty hefty range of stuff you
>could talk to them about -- most of that 500K is dialogue, after all --
>and a not incredibly generous amount of time allotted to chat. Did
>people talk for a while and then restore, or keep conversation to a
>minimum, or what?
This is, I think, a special case because the game as whole pretty much
mandated frequent restoring anyway. I haven't finished it, but to be honest
part of the reason I quit was that, yes, I was irritated by having to
restore-restart-restore-restart-restore to try different things. So there's
one data point.
ems...@mindspring.com
> On 12/4/01 5:19 AM, Magnus Olsson at m...@df.lth.se posted:
>
> >> This has been brought up before by others, but I tend to dislike
> >> conversation menus. Mind you, I won't refuse to play a game that uses
> >> them, or not enjoy one that does, but I prefer the
> >>
> >>> TALK TO [NPC]
> >>> ASK [NPC] ABOUT
> >>> TELL [NPC] ABOUT
> >>> SAY, "SOMETHING"
> >>
> >> method of interacting with NPCs.
>
> > Well, while I share most people's dissatisfaction with menus, I'm also
> > not very fond of ask/tell. In most cases, it turns the entire
> > conversation into a huge guess-the-noun puzzle. And it's depressing to
> > have an otherwise well-crafted NPC give the same stock answer to all
> > the topics you can think of.
>
> I would argue that it's just as depressing to get a nice, clear answer for
> every menu item, yet be prevented from the option of trying, "TELL GRANT TO
> BITE ME" in Best Of Three. It may very well be that it isn't implemented,
> but at least I have the satisfaction of trying it. And if the game spits
> back "Grant doesn't know how to bite you," I have something amusing to
> offset the fact that he's being snotty to my PC.
This is a fair complaint, but arguably not something that bears
directly on the choice of ASK/TELL vs. menu systems. The classic
Inform library, at least, provides for
ASK NPC ABOUT foo
TELL NPC ABOUT foo
NPC, do foo
(for which Irene Callaci has written a library extension to allow TELL
NPC TO DO FOO and ASK NPC TO DO FOO)
SAY foo TO NPC
GIVE foo to NPC
SHOW foo to NPC
It is not uncommon, when building an alternative conversation system,
to try to reduce all of these except perhaps the last two to a single
command or command arrangement:
TALK TO NPC with no menu covers all that ground at once while leaving
the player with virtually no control. Effective, but extremely
reductionist.
TALK TO NPC that then leads to a menu state with two-ten options again
tends to try to cover for all of thsoe except give and show.
Best of Three's menu/topic system allows for considerably more
latitude with the menu, but it doesn't deal very effectively with the
orders problem. I think it *generally* allows the player a great deal
more subtlety of input than the Inform/Infocom convention, because you
can choose both a subject of conversation and a specific phrasing. As
far as orders are concerned, though, it curtails the player. You can
get around it slightly by trying to implement various probable orders
*as* menu items, eg:
>TOPIC TEA
1: "Who in their right mind puts lemon in Earl Grey???"
2: "Stop futzing around and just drink it already."
Where (2) is a much more attitudinal form of >GRANT, DRINK TEA. But
no sane person has time to do that for all the possible combinations;
furthermore, putting some of them on the menu might give away
solutions that you had not yet wished to reveal. For instance, you
might want the player to discuss a given object extensively with an
NPC without giving away that a specific order will be useful: eg:
>TOPIC SWORDSTICK
1: "My, what a long swordstick you have."
2: [Of the swordstick] "Where did you get that, anyway?"
3: "Quick, give me the swordstick. I need to run the waiter through
like a spitted pig."
1 and 2 are conversational; 3 implies a particular chain of action
that might need to be *available* to the player at any time, but which
we might not want to flag so unmistakeably.
So the trick that I'm trying to deal with now is figure out how to
offer the refinements of a Bo3-type system with the freedom of an
ASK/TELL system. Some of the conclusions I've drawn from the
discussion so far:
1. If you select a new topic explicitly, there should *always* be at
least one menu item available there, even if it is just, "So, have any
more thoughts you'd like to share about metempsychosis?"
2. Where certain kinds of speech-acts are concerned, there should be
multiple ways of achieving the desired effect, at least one of which
should be "natural" in a natural-language sense and another should be
consistent with the structure of the topic/menu system. What I mean
by that is that, for instance, after a yes/no question, the player
ought to be able to get a response from
>YES
>SAY YES
>TOPIC YES
The first two are natural; one doesn't think of changing the topic
*to* "yes", since that's obviously not an object of discussion
precisely. But someone using the system might think that >TOPIC YES
was a possibility, so it should also be accounted for. Things to say
selected in this way should not have a different conversational status
than anything else you select to say, however: they should still give
the PC's conversational squib and the NPC's response in full, and
should have the same effect on the flow of conversation as anything
else.
3. Orders should similarly function within the conversation system,
not as a separate modality.
4. If a player has given sufficiently explicit instructions that there
is nothing to be gained from offering a menu, the menu item should be
executed immediately. So if, eg, there are multiple ways to say yes,
you might get
>YES
1: "Sure, I'd love to!"
2: [Grudgingly] "OK, I guess."
But if only "Sure, I'd love to!" is implemented, then
>YES
should produce not
1: "Sure, I'd love to!"
but
"Sure, I'd love to!" you say to Grant. Inwardly you smile about the
effect it will have on his ego when you stand him up at Portland's
most expensive restaurant.
Grant beams.
---
5. In order to deal with some of these effects, there needs to be
default behavior for order-giving that works reasonably within the
topic/menu system.
As a general principle, it seems to me that I need to combine the kind
of tailor-writing that appeared in Best of Three with a better system
of underlying semi-automatic behavior. It's just a fact, sad but
inevitable, that no amount of effort on my part will produce the full
spectrum of possibilities that a person could want to try. But if I
have better defaults, then there will be fewer instances of people
trying things and getting no response from the game at all, which is
frustrating. Tailor-writing of everything gives you this:
>GRANT, GIVE ME THE SWORDSTICK
Instantly understanding your intent, Grant hands you the rosewood
swordstick. You draw the blade and leap from your seat, skewering the
waiter. Green blood oozes from his wound.
"A-HA," you shout triumphantly. "I always suspected this town was the
crash-site of aliens from the planet Klaxtor!"
>GRANT, JUMP
Grant doesn't seem interested in doing what you tell him.
>GRANT, LOOK AT THE FICUS
Grant doesn't seem interested in doing what you tell him.
>GRANT, DRINK YOUR TEA
Grant doesn't seem interested... (you get the idea)
Whereas with sufficiently clever defaulting you can get things more
like:
>GRANT, JUMP
[ Default: jump falls into a category of verbs that would be random
and silly in most contexts.]
"Hey," you suggest. "Why not jump around a bit?"
Grant looks aghast. "I have my dignity, girl."
>GRANT, LOOK AT THE FICUS
[ Default: look verbs draw attention to their objects; equivalent to
SHOW.]
"Grant, look at the ficus."
He glances over at it. "Is that remarkable for some reason?"
---
FWIW, I was working on, but never really completed the refinement of,
a system of verb classification for Bo3. The idea there was not so
much to handle what happened if the player gave Grant orders as it was
to deal with how he might react to demonstrably strange behavior, such
as fiddling with inanimate objects for no reason, etc. The plan was
to divide up all possible verbs into categorizations (DESTRUCTIVE,
MANIPULATIVE, RANDOM, etc) and have him give responses based on that.
I still think that a systematization like this is probably useful for
the kinds of applications you're talking about.
For this reason it would be convenient if verbs in Inform were
represented by objects, so that they could inherit from classes; that
would make my life considerably simpler. But there are still ways of
handling the problem, which is really this: how do you represent in
the game model the theoretical understanding that the player already
brings naturally to bear on it? My current solution for WIP is to
have, in addition to the simulationist categories of material and so
on, categories of use -- weapons, clothing, food, etc. -- which is
convenient not only for teaching the game which objects are valid
second arguments for
>KILL GRANT WITH <thing>
but also for teaching Grant that if, during a hostile situation, you
show him a knife, he's to interpret that as a threat and not as an
invitation to comment upon the craftsmanship that went into decorating
the hilt.
[No, I am not writing another game about Grant. This has been An
Example.]
In any case: the original point of this all was that I am interested
in expanding the range of options the player has available, not in
curtailing it, and that where the topic/menu system and the other
aspects of my NPC libraries actually offer the player less control
than a traditional ASK/TELL/orders system, I consider that a
deficiency and am working to correct it as cleanly as possible.
ES
Dan Schmidt
| > This works well in _Galatea_, because there isn't anything to do
| > but converse. But will this work in a fast-paced, plot-driven game?
| > If you've got three turns to talk to the an NPC, you don't want
| > to spend those three turns on exploration of dialogue space.
|
| Well, I'll ask: VARICELLA's NPCs had a pretty hefty range of stuff you
| could talk to them about -- most of that 500K is dialogue, after all --
| and a not incredibly generous amount of time allotted to chat. Did
| people talk for a while and then restore, or keep conversation to a
| minimum, or what?
'Not incredibly generous', yeah, you could say that.
I talked everyone's ear off, restored, and was then as silent as a
little mouse except when necessary when I actually tried to put
together a winning playthrough.
| For that matter, this thread *is* about "putting words in the PC's
| mouth," something VARICELLA did at least in part with a
| sentence-generating algorithm that took into account one's selected
| tone. Successful? Unsuccessful? No sense talking about these
| things entirely as hypotheticals when we've actually seen them in
| action already.
I found the tone-inflected PC sentences entertaining, if that counts
as success.
Adam Thornton
Dan Schmidt <df...@harmonixmusic.com> wrote:
>I talked everyone's ear off, restored, and was then as silent as a
>little mouse
...a little, EVIL, mouse.
Adam
I.R.
> I'm working on an Inform library for menu-driven conversation, called
> MOOCE... I'm planning to release the source of MOOCE as a library
> contribution,
> but first I'd like to see how it works in a reasonably practical
> context. [snip]
I've checked it out, and it looks - and "feels" - terrific. Great job!
I've got two questions:
1) Could you implement a mode in which the menu would be at the bottom
of the screen? The general consensus seems to be (and I agree) that it
is less strenuous for the eyes that way, and provides for better
integration with the main game text.
2) When do you plan to release the library :)? I'd love to mess around
with it.
-I.R.
Magnus Olsson
>> not very fond of ask/tell. In most cases, it turns the entire
>> conversation into a huge guess-the-noun puzzle. And it's depressing to
>> have an otherwise well-crafted NPC give the same stock answer to all
>> the topics you can think of.
>
>I would argue that it's just as depressing to get a nice, clear answer for
>every menu item, yet be prevented from the option of trying, "TELL GRANT TO
>BITE ME" in Best Of Three.
Yes. I'm not arguing that conversation menus are better than ask/tell
in some absolute sense, just that both approaches have their
weaknesses.
>> The exception is games like _Galatea_, where a) an enormous amount of
>> effort has been put into the conversation, and b) the player has
>> time to do what you describe here:
>
>Right, except that the *only* reason I picked Galatea was to contrast it
>with another one of Emily's games, Best Of Three, and my subjective
>experience of both.
OK. But then *I* would like to pick Galatea as an example of a game
where the ask/tell interface works unusually - almost uncannily -
well.
>I said
>in my closing lines, something to the effect that only an insane person
>would try and implement 5 "Galatea"s in a game that other things to it
>besides conversation.
And even if somebody were insane enough to do that, I'm not sure that
the game would be very enjoyable to play.
>Yes, my enjoyment of Galatea might have been tempered some if she had said,
>"I don't know anything about that" more often than not, but I still say that
>Emily's writing in that game helped guide the conversation via the display
>dialogue, and that should be of a prime concern to any NPC code, regardless
>of complexity, or menu / ASK/TELL interface.
Yes, definitely.
>Should there be room to ask off the wall questions not in line with the
>topic being discussed? Sure.
Yes; this is a matter of fleshing out the world, similar to
implementing descriptions of scenery objects. If only important things
can be examined *or* talked about, the game world feels dry and
incomplete.
>If I ask Floyd (a robot) how to use the food
>dispenser, should I get all bent and cry "guess the verb!" when he says he
>doesn't know anything about it? No.
No, unless there's a reason to think that Floyd knows anything
important about it. "Guess the verb/noun" only applies when there's
something you really want to do, but can't, because the game has too
small a vocabulary.
>If I ask him about the elevators, and
>get "Floyd's face lights up. 'Hey! I have a card for one of those!'" and
>pops a nice, useful key into your hand, I would call that a success.
A comment here is that both these examples concern asking Floyd about
items in the game. It is pretty clear that a good NPC in an ask/tell
game should respond at least to all important items in the game.
The tricky part is when you want to discuss abstarct concepts. I don't
necessarily mean asking Floyd if he believes that robots go to the big
backup store in the sky when they are scrapped, but rather things
like, in a game where the NPC heroine has just told a rather
complicated story and then rushed off to slay some monster,
"Gabrielle, what on earth was Xena talking about?"
You can do *some* amount of this with simple keywords - perhaps, in
this case, "Ask Gabrielle about Xena's story", but only so much.
>> This works well in _Galatea_, because there isn't anything to do
>> but converse. But will this work in a fast-paced, plot-driven game?
>> If you've got three turns to talk to the an NPC, you don't want
>> to spend those three turns on exploration of dialogue space.
>
>At that point in a game, either it should be crystal clear what information
>you need to get from the NPC, or you should have more than three moves to do
>it in.
>
>> I've played games where I had, say, three turns to extract information
>> with an ask/tell system. "Rats, those topics didn't work. Restore, try
>> again. Mimesis, don't let the door hit you on the way out."
>
>Well, the parser of IF is a tool, but a limited one. If you have a rather
>long, but somewhat flimsy board to use as a lever, you don't make someone
>lift a boulder with it. That's just bad game design.
Yes, it's bad game design. But bad game design can be caused by a
sloppy writer, or by insufficient tools.
My point is really that the ask/tell interface at the current state of
the art is insufficient for doing certain things which are very easy
to do with conversation menus. *And vice versa*.
>In the above example, using a menu, it is customary to obscure the intended
>line of questioning with other quotes that don't produce anything useful. So
>what's to say you don't just pick menu items / restore until you get hit the
>right answer?
>And would you even know why it's the right answer, or would
>you just be plugging in numbers until something worked? And how does that
>enhance the game experience? Mimesis is getting its ass smacked with the
>door again, IMO.
I'd say that this is bad game design, too.
>And if you're forced into that kind of situation, with a limited amount of
>moves to extract vital information from a NPC with an obviously flawed
>system of interaction (manual or menu) it begs the question, "And the NPC is
>important/in my way here because...?"
>
>...the game isn't designed well, the menu, ASK/TELL debate nonwithstanding.
It's bad game design if the player runs into the wall because of it. I
don't think it's necessarily bad game design in all cases.
Speaking as a game designer, I'd say that it's very easy for you to
say "Avoid bad game design." Sure, we all want to do that. But if my
problem is that I want to achieve certain goals with certain tools,
and the tools aren't sufficient, maybe I should consider other tools
rather than just giving up.
>As far as I'm concerned, this all boils down to people getting upset that
>the NPCs aren't AI (even if it's just on a subconscious level).
No, I don't think so.
As players, we accept all kinds of strange things in games, as long as
we can suspend disbelief. Granted, it's harder to suspend disbelief in
an NPC than in many other things, but I don't think we're going to be
upset because Floyd doesn't pass the Turing test. We *are* going to be
upset, though, if Floyd doesn't respond to input that *the game leads
us to believe* he should respond to. If everything about the situation
seems to cry out "Ask Floyd how to get into the reactor," but the game
won't let us, *then* mimesis leaves via the back door.
>Which is the
>same as being upset that you can't take a shard of glass in the game,
>intended to cut a rope and free your bindings, and stick it in your eye just
>because you want to try it and see how "clever" the game is.
No, it's more like if there is a piece of glass with a warning label
"CAUTION: Keep out of eyes" and the game then responds to "Put glass
in eye" with the response "That can't contain anything" or "I don't
know the word 'eye'".
>When you talk to someone in real life, either it clicks, and you become
>friends, or it doesn't and you never see that person again. I think the same
>goes for NPCs and interacting with them, you either want to interact with
>them or you quit the game, and I'm not sure it has a whole lot to do with
>using a menu or ASK/TELL. Or I'm just new, and don't know what the hell I'm
>talking about.
I think you know what you're talking about, it's just that you're
talking about a different problem than that I'm trying to address.
>To bring up Floyd again, an NPC that does next to nothing except follow you
>around until a specific point in the game... plenty of people had a strong
>reaction to him. A strong *emotional* reaction, and I have to wonder if
>getting bogged down in discussing the pros and cons of menus versus ASK/TELL
>makes it more or less likely that we'll see an NPC that can provoke that
>kind of reaction in people again.
I sincerely hope not.
M. D. Krauss
just new, and don't know what the hell I'm talking about" -- can't second
that, you definitely know what your talking about.
-M
On Tue, 04 Dec 2001 15:29:34 GMT
"Gregg V. Carroll" <gr...@midcoast.com> wrote:
Gregg V. Carroll
>> Well, while I share most people's dissatisfaction with menus, I'm also
>> not very fond of ask/tell. In most cases, it turns the entire
>> conversation into a huge guess-the-noun puzzle. And it's depressing to
>> have an otherwise well-crafted NPC give the same stock answer to all
>> the topics you can think of.
>
> ... knowing you are missing out on gobs of text that the author did
> write but you can't think of the noun to trigger it. Agreed.
I can kind of see your point, but it also seems like it's just short of
taking the "interactive" out of IF. I don't want to be argumentative or
force my point, however.
> Set Galatea down in the middle of _Zork 1_ or _Dangerous Curves_ and
> you've got a writing/coding/playing nightmare on your hands. The NPC
[snip Galatea discussion]
I almost regret bringing that game up now, because I was afraid this was
going to happen. The only point I was trying to illustrate was contrasting a
ASK/TELL system and a menu system of conversation in two different games by
the same author. I picked Emily's, well, because they're good.
I wasn't hinting at all that IF should be populated with Galatea's. God no.
That would get really tedious really quickly, *unless* conversation is the
only objective. But really, who want's to talk to NPCs all day? As I
understand it, both projects were derivative of a larger, unmanagable one
(for reasons that we're putting forth here), and were somewhat experimental
in nature.
> I see it as the difference between reading a book for pleasure, and
> reading a book knowing there is going to be a test at end. In the
> later, all the while you are reading, you are trying to pick out which
> bits you will be querried on during the test and store them for future
> access, determine their relevence, draw conclusions... For me, that
> spoils the pure joy of reading.
Well, this starts getting into the territory of "playing for story" (reading
for pleasure) or "playing for puzzles," (reading for analysis) which is a
matter of individual taste.
> A non-NPC game that produced a similiar "sit back and enjoy it"
> response (for me) was SUTW. There again, the gamespace was small and I
> (the PC) had no objectives other than to experience it. It was a
> wonderful effect and SUTW still ranks as my all time favorite game
> because of that. Do I think it would would for a larger game where you
> have goals to reach? No, for pretty much the same reason.
All I'd like to say here is that I wish people would at least mention the
full title to a game once before using acronyms. I see it a lot, and 75% of
the time I have no idea what game is being put forth as an example. Not
griping to you personally, Kathleen, just speaking in general.
>> If you've got three turns to talk to the an NPC, you don't want
>> to spend those three turns on exploration of dialogue space.
>
> Exactly. And if you have a hundred objects to ask about, which do you
> start with?
Like I said, if you've got 100 possible objects to find some info about from
an NPC in 3 moves, something else entirely is worng with the game that has
nothing to do with how one is interacting with the NPC. In my opinion,
anyway.
Gregg
Magnus Olsson
>> This works well in _Galatea_, because there isn't anything to do
>> but converse. But will this work in a fast-paced, plot-driven game?
>> If you've got three turns to talk to the an NPC, you don't want
>> to spend those three turns on exploration of dialogue space.
>
>Well, I'll ask: VARICELLA's NPCs had a pretty hefty range of stuff you
>could talk to them about -- most of that 500K is dialogue, after all --
>and a not incredibly generous amount of time allotted to chat. Did
>people talk for a while and then restore, or keep conversation to a
>minimum, or what?
IIRC, I tended to explore conversation space via save-and-restore.
But then the game itself builds on exploring all kinds of stuff via
save-and-restore, so having to do this with conversations didn't
bother me at all.
>For that matter, this thread *is* about "putting words in the PC's mouth,"
>something VARICELLA did at least in part with a sentence-generating
>algorithm that took into account one's selected tone. Successful?
>Unsuccessful?
I think that feature of _Varicella_ was underused: there were few
instances where it mattered.
Magnus Olsson
I.R. <quickan...@yahoo.com> wrote:
>m...@df.lth.se (Magnus Olsson) wrote in message
>news:<9ub6vm$c62$1...@news.lth.se>...
>> I'm working on an Inform library for menu-driven conversation, called
>> MOOCE... I'm planning to release the source of MOOCE as a library
>> contribution,
>> but first I'd like to see how it works in a reasonably practical
>> context. [snip]
>
>I've checked it out, and it looks - and "feels" - terrific. Great job!
>I've got two questions:
>
>1) Could you implement a mode in which the menu would be at the bottom
>of the screen? The general consensus seems to be (and I agree) that it
>is less strenuous for the eyes that way, and provides for better
>integration with the main game text.
I assume you played the Z-code version (troll.z5)? The Glulx version
does put the menu at the bottom, because I totally agree with you
that that's better.
Unfortunately, you simply can't do that in V5 Z-code, because only
the upper window (i.e. the status line) allows things like moving
the cursor around.
>2) When do you plan to release the library :)? I'd love to mess around
>with it.
Real Soon Now (tm) - that is, as soon as I have had some time to
clean up the source a little.
Magnus Olsson
>> response (for me) was SUTW.
>All I'd like to say here is that I wish people would at least mention the
>full title to a game once before using acronyms. I see it a lot, and 75% of
>the time I have no idea what game is being put forth as an example.
SUTW = _Space Under The Window_, by Zarf.
>>> If you've got three turns to talk to the an NPC, you don't want
>>> to spend those three turns on exploration of dialogue space.
>>
>> Exactly. And if you have a hundred objects to ask about, which do you
>> start with?
>
>Like I said, if you've got 100 possible objects to find some info about from
>an NPC in 3 moves, something else entirely is worng with the game that has
>nothing to do with how one is interacting with the NPC. In my opinion,
>anyway.
A more interesting situation would be if you have *one*
think you want to ask the NPC, but it's not a simple thing like an
object, but something more complicated. You can do probably do this
with ask/tell, but it may be hard to do it in three moves. If the
plot absolutely requires fast pacing here, you don't want to
give the player more than three moves. In such a case, it may be
better to use a menu system.
Gregg V. Carroll
>>> Well, while I share most people's dissatisfaction with menus, I'm also
>>> not very fond of ask/tell. In most cases, it turns the entire
>>> conversation into a huge guess-the-noun puzzle. And it's depressing to
>>> have an otherwise well-crafted NPC give the same stock answer to all
>>> the topics you can think of.
>> I would argue that it's just as depressing to get a nice, clear answer for
>> every menu item, yet be prevented from the option of trying, "TELL GRANT TO
>> BITE ME" in Best Of Three.
> Yes. I'm not arguing that conversation menus are better than ask/tell
> in some absolute sense, just that both approaches have their
> weaknesses.
Agreed, completely.
>>> The exception is games like _Galatea_, where a) an enormous amount of
>>> effort has been put into the conversation, and b) the player has
>>> time to do what you describe here:
>> Right, except that the *only* reason I picked Galatea was to contrast it
>> with another one of Emily's games, Best Of Three, and my subjective
>> experience of both.
> OK. But then *I* would like to pick Galatea as an example of a game
> where the ask/tell interface works unusually - almost uncannily -
> well.
It's such a special-case example that it's almost counterproductive to
include it in "NPCs in general" discussions. It's like saying all films
should be like Citizen Kane, even ones starring Jim Carrey. :)
>> I said
>> in my closing lines, something to the effect that only an insane person
>> would try and implement 5 "Galatea"s in a game that other things to it
>> besides conversation.
> And even if somebody were insane enough to do that, I'm not sure that
> the game would be very enjoyable to play.
It could be if the game was open-ended enough, but then that's a task in
itself. Maybe, say, a 30MB text game that takes about a 20secs to compute a
reasonable response to a command. :)
>> Should there be room to ask off the wall questions not in line with the
>> topic being discussed? Sure.
> Yes; this is a matter of fleshing out the world, similar to
> implementing descriptions of scenery objects. If only important things
> can be examined *or* talked about, the game world feels dry and
> incomplete.
>> If I ask Floyd (a robot) how to use the food
>> dispenser, should I get all bent and cry "guess the verb!" when he says he
>> doesn't know anything about it? No.
> No, unless there's a reason to think that Floyd knows anything
> important about it.
Right, that was my exact point. "There's a reason to think" is tied directly
to display text, and its subsequent quality, rather than the conversation
interface. The above is a bad example, because it's not quite ambigious
enough (My point was that asking a robot about a food dispenser, seeing as
how robots have little use for food, and therefore not much knowledge about
how to use food dispensers - assuming the robot isn't a database - and
getting ticked because you got a stock response is not an argument for using
menus over ASK/TELL. I could be wrong about that, but that was my
argument.).
>> If I ask him about the elevators, and
>> get "Floyd's face lights up. 'Hey! I have a card for one of those!'" and
>> pops a nice, useful key into your hand, I would call that a success.
> A comment here is that both these examples concern asking Floyd about
> items in the game. It is pretty clear that a good NPC in an ask/tell
> game should respond at least to all important items in the game.
>
> The tricky part is when you want to discuss abstarct concepts. I don't
> necessarily mean asking Floyd if he believes that robots go to the big
> backup store in the sky when they are scrapped, but rather things
> like, in a game where the NPC heroine has just told a rather
> complicated story and then rushed off to slay some monster,
> "Gabrielle, what on earth was Xena talking about?"
>
> You can do *some* amount of this with simple keywords - perhaps, in
> this case, "Ask Gabrielle about Xena's story", but only so much.
Okay, but going back to Floyd, based on his outward behavior, you learn
quite a lot about what might be and what might not be appropriate abstract
concepts to ask him about. This isn't in the game, but just based on Floyd's
surface behavior, ASK FLOYD ABOUT THE MEANING OF LIFE isn't likely to
produce anything useful, and that's determinable *before even trying*
because it's cued by the display text regarding Floyd and what he's like.
So I would say that *perhaps* defficiencies in conversation interfaces
should be headed-off before they can happen by appropriate cues, be they
dialogue, behavioral, or "other."
This whole thread is making me extremely nervous about submitting a game to
Comp with an NPC in it. I am apt to get my virtual butt kicked upon review
posting time. "Well, Gregg, *you* said it yourself..." :)
> My point is really that the ask/tell interface at the current state of
> the art is insufficient for doing certain things which are very easy
> to do with conversation menus. *And vice versa*.
Right, which comes back to what we agreed (I think) on before, and that
neither one is intrinsically bad or good, just different and useful
depending on the context of the game.
I wasn't really trying to build a case for or against either method of
interaction (beyond saying which one I personally prefer), if it sounded
that way that's my fault.
> Speaking as a game designer, I'd say that it's very easy for you to
> say "Avoid bad game design." Sure, we all want to do that. But if my
> problem is that I want to achieve certain goals with certain tools,
> and the tools aren't sufficient, maybe I should consider other tools
> rather than just giving up.
Absolutely.
>> As far as I'm concerned, this all boils down to people getting upset that
>> the NPCs aren't AI (even if it's just on a subconscious level).
> No, I don't think so.
>
> As players, we accept all kinds of strange things in games, as long as
> we can suspend disbelief. Granted, it's harder to suspend disbelief in
> an NPC than in many other things, but I don't think we're going to be
> upset because Floyd doesn't pass the Turing test. We *are* going to be
> upset, though, if Floyd doesn't respond to input that *the game leads
> us to believe* he should respond to. If everything about the situation
> seems to cry out "Ask Floyd how to get into the reactor," but the game
> won't let us, *then* mimesis leaves via the back door.
Okay, I see what you're getting at more clearly here. Agreed.
>> Which is the
>> same as being upset that you can't take a shard of glass in the game,
>> intended to cut a rope and free your bindings, and stick it in your eye just
>> because you want to try it and see how "clever" the game is.
> No, it's more like if there is a piece of glass with a warning label
> "CAUTION: Keep out of eyes" and the game then responds to "Put glass
> in eye" with the response "That can't contain anything" or "I don't
> know the word 'eye'".
Ditto.
>> When you talk to someone in real life, either it clicks, and you become
>> friends, or it doesn't and you never see that person again. I think the same
>> goes for NPCs and interacting with them, you either want to interact with
>> them or you quit the game, and I'm not sure it has a whole lot to do with
>> using a menu or ASK/TELL. Or I'm just new, and don't know what the hell I'm
>> talking about.
> I think you know what you're talking about, it's just that you're
> talking about a different problem than that I'm trying to address.
I wandered OT there a bit, yes. Apologies. :)
Gregg
Gregg V. Carroll
posted:
> Just jumping in to second everything Gregg said here. Except one: "Or I'm
> just new, and don't know what the hell I'm talking about" -- can't second
> that, you definitely know what your talking about.
For a second there I thought I was going to have to change the subject to
"Putting Gregg's foot in his mouth (was: A constructive argument but isn't
anymore)." :)
Gregg
Gregg V. Carroll
>>> A non-NPC game that produced a similiar "sit back and enjoy it"
>>> response (for me) was SUTW.
>> All I'd like to say here is that I wish people would at least mention the
>> full title to a game once before using acronyms. I see it a lot, and 75% of
>> the time I have no idea what game is being put forth as an example.
> SUTW = _Space Under The Window_, by Zarf.
Ah. I've been sidestepping his games, so I haven't gotten to that one just
yet.
>>>> If you've got three turns to talk to the an NPC, you don't want
>>>> to spend those three turns on exploration of dialogue space.
>>> Exactly. And if you have a hundred objects to ask about, which do you
>>> start with?
>> Like I said, if you've got 100 possible objects to find some info about from
>> an NPC in 3 moves, something else entirely is worng with the game that has
>> nothing to do with how one is interacting with the NPC. In my opinion,
>> anyway.
> A more interesting situation would be if you have *one*
> thing you want to ask the NPC, but it's not a simple thing like an
> object, but something more complicated. You can do probably do this
> with ask/tell, but it may be hard to do it in three moves. If the
> plot absolutely requires fast pacing here, you don't want to
> give the player more than three moves. In such a case, it may be
> better to use a menu system.
That makes sense. The thing I'm working on now has some time-based elements
to it (not quite as tight as 3 moves, but tight enough) but they don't
involve NPCs, so I hadn't considered that problem myself. I might have
subconsciously avoided it in my game design, in fact, since my plate is full
enough the way things are.
Gregg
Magnus Olsson
> A comment here is that both these examples concern asking Floyd about
> items in the game. It is pretty clear that a good NPC in an ask/tell
> game should respond at least to all important items in the game.
>
> The tricky part is when you want to discuss abstarct concepts.
To exemplify this, here's an example where I don't really see how to
implement a dialogue with ask/tell (The hypothetcial game can be
assumed to be quickly paced and packed with action, and not really a
game where conversation plays a large part):
You're in a tavern in a city torn by civil unrest and on the brink of
war, trying to locate some magic McGuffin or other, when a stranger
sits down at your table and starts discussing the local
ruler. Politics is a sensitive matter, so what you say may have very
important consequences.
Now, if you want to know what the stranger thinks about the king, it's
easy to "ask stranger about king".
But what if the stranger asks you, "Well, what do you think of king
Hippokephalos?"
With conversation menus, you can get a choice
1: "King whatsisname? Never heard of him (hic)"
2: "I don't really have an opinion - I'm a stranger here"
3: "The greatest ruler that ever lived! May he continue to do so!"
4: "He's a tyrant! Death to him, and long live the revolution!"
5: "He's a good ruler, but who will succeed him?"
- but how do you do this with an ask/tell interface?
And this is a simple example - what if the NPC was a priest asking you
about your philosophical position regarding his religion?
You could, conceivably, have the NPC ask you a series of simple
questions, but that would smack of interrogation and I think it would
be rather clumsy; not the effect I'm after.
Sean T Barrett
>something VARICELLA did at least in part with a sentence-generating
>algorithm that took into account one's selected tone.
Whoa, sentence-generating, really? I'm always keeping an eye out
of examples of synthetic text; I'd asumed the Varicella texts
were hand-authored, but I didn't play it for long.
SeanB
Adam Cadre
> I wasn't hinting at all that IF should be populated with Galatea's. God
> no. That would get really tedious really quickly, *unless* conversation
> is the only objective.
Er, what? How can a greater range of responses be a detriment? Like,
let's say you're playing a game where you have to get into a particular
club. There's a bouncer out front who won't let you pass. Show the
bouncer a ticket, and he'll let you in. That's all the interaction the
games requires you to have with the bouncer for the plot to proceed.
Now, as a bonus, the game author decides to add a Galatea-like web of
responses to the bouncer so that you *can* spend three hours talking with
him about the meaning of life. You don't have to, but you can. How does
that possibility make a game tedious? The only downside I can think of
is that games would take longer to download.
Adam Cadre
> Whoa, sentence-generating, really? I'm always keeping an eye out
> of examples of synthetic text; I'd asumed the Varicella texts
> were hand-authored, but I didn't play it for long.
There's a routine in VARICELLA called GenSen() that mixes and matches
phrases to create a sentence of the appropriate tone; GenSen() in turn
calls the routine Epithet() to pick an appropriate slur or honorific
to plug into the sentence, and also makes use of the variable "substr"
when it needs to refer to the topic at hand. The variation isn't huge,
but when asking a question, there are usually about a dozen different
ways it might come out.
Joe Mason
ems...@mindspring.com <ems...@mindspring.com> wrote:
>
>>GRANT, JUMP
>[ Default: jump falls into a category of verbs that would be random
>and silly in most contexts.]
>"Hey," you suggest. "Why not jump around a bit?"
>
>Grant looks aghast. "I have my dignity, girl."
Now I have the urge to replay Chix Dig Jerks and just repeat
>GIRL, JUMP AROUND
over and over again.
Joe
Kathleen M. Fischer
>Gregg Carroll wrote:
>> I wasn't hinting at all that IF should be populated with Galatea's. God
>> no. That would get really tedious really quickly, *unless* conversation
>> is the only objective.
>
>Er, what? How can a greater range of responses be a detriment? Like,
>let's say you're playing a game where you have to get into a particular
>club. There's a bouncer out front who won't let you pass. Show the
>bouncer a ticket, and he'll let you in. That's all the interaction the
>games requires you to have with the bouncer for the plot to proceed.
>
>Now, as a bonus, the game author decides to add a Galatea-like web of
>responses to the bouncer so that you *can* spend three hours talking with
>him about the meaning of life. You don't have to, but you can. How does
>that possibility make a game tedious?
Because the player might not realize all they have to do is hand the ticket
to
the bouncer and enter? Instead, they sit there for the next two (player)
hours
typing in all nouns and concepts known to the game (and a few choice ones
that
aren't) to try to discover everything the NPC knows - after all, it must be
important or the author wouldn't have spent so much time on it. Two hours
later, the disenchanted player scores the game a 1, grumbling that they
never
made it past the first scene. :) :) :)
Another way of looking at it: It would be like including in your game a
completely implemented 5 story department store - with all items implemented
fully - wearable, breakable, usable, whatever.. but not removable from the
store and of no use in any way. All the player really needs to do is read
the
sign on the door that says "Closed on Sundays." Unless you call the store
"The Red Herring", I bet a certain number of players would grumble at its
"unfairness."
I do think NPC's should have personality, should speak when spoken to :)
and
should have a reasonable grasp of their lot in life and be able to relate
it... briefly... But I also think they should know their bounds and the
players should be able to fathom those bounds as well. (unless you are
writing
an Art Show Portrait-like game, where the whole point is the NPC, in which
case all of the above should be ignored)
Kathleen
-- Prized Possession (Comp2001)
--
http://www.ifarchive.org/if-archive/games/competition2001/inform/possess
-- Masquerade (Comp2000, nominated for Best Story (XYZZY's))
-- http://www.ifarchive.org/if-archive/games/inform/Mask.z5
-- The Cove - http://www.ifarchive.org/if-archive/games/inform/Cove.z5
-- Excuse me while I dance a little jig of despair
Rikard Peterson
news:9ujkqn$e4o$1...@news.lth.se...
>
> But what if the stranger asks you, "Well, what do you think of king
> Hippokephalos?"
>
> With conversation menus, you can get a choice
>
> 1: "King whatsisname? Never heard of him (hic)"
> 2: "I don't really have an opinion - I'm a stranger here"
> 3: "The greatest ruler that ever lived! May he continue to do so!"
> 4: "He's a tyrant! Death to him, and long live the revolution!"
> 5: "He's a good ruler, but who will succeed him?"
>
> - but how do you do this with an ask/tell interface?
>
> And this is a simple example - what if the NPC was a priest asking
> you about your philosophical position regarding his religion?
>
> You could, conceivably, have the NPC ask you a series of simple
> questions, but that would smack of interrogation and I think
> it would be rather clumsy; not the effect I'm after.
I think this is the major difference between ask/tell and conversation
menus. If you want lengthy dialogue with a lot of back and forth
interaction (like a real conversation), then it seems clear to me that a
menu system would be more suitable to the game. Ask/tell would be very
likely to end up with "guess the noun".
If you on the other hand want to be sparse with dialogue, ask/tell would
be a logical choice. I think which system is the best depends on what
kind of game you are making.
Rikard
Daryl McCullough
>How can a greater range of responses be a detriment? Like,
>let's say you're playing a game where you have to get into a particular
>club. There's a bouncer out front who won't let you pass. Show the
>bouncer a ticket, and he'll let you in. That's all the interaction the
>games requires you to have with the bouncer for the plot to proceed.
>
>Now, as a bonus, the game author decides to add a Galatea-like web of
>responses to the bouncer so that you *can* spend three hours talking with
>him about the meaning of life. You don't have to, but you can. How does
>that possibility make a game tedious?
It's only tedious if you find out, after spending three hours talking
with the bouncer, that he really doesn't know anything that is any use
to you.
--
Daryl McCullough
CoGenTex, Inc.
Ithaca, NY
Mike Duncan
<greenG...@MailAndNews.com> wrote:
>>
>>Now, as a bonus, the game author decides to add a Galatea-like web of
>>responses to the bouncer so that you *can* spend three hours talking with
>>him about the meaning of life. You don't have to, but you can. How does
>>that possibility make a game tedious?
>
>Because the player might not realize all they have to do is hand the ticket
>to
>the bouncer and enter? Instead, they sit there for the next two (player)
>hours
>typing in all nouns and concepts known to the game (and a few choice ones
>that
>aren't) to try to discover everything the NPC knows - after all, it must be
>important or the author wouldn't have spent so much time on it. Two hours
>later, the disenchanted player scores the game a 1, grumbling that they
>never
>made it past the first scene. :) :) :)
Sometimes games are about a bit more than just moving forward. Say in
this game you are Diogenes, lantern in hand, and your goal is to sneak
into an elite watering hole in Athens so you can throw your own dung
on the floor and make some pithy remarks about the smell. The bouncer
turns out to be a wannabe Cynic, so you can trade sarcastic rejoinders
with him for hours before venturing inside. Totally unnecessary but
entertaining.
Random idea - Let's say Diogenes' lantern was implemented (You have a
lantern, which is lit.) Aiming it at various NPCs produces a slightly
different description.. Unnecessary to finish the game, but a bonus
for those that try things like SHINE LANTERN ON BOUNCER. As Kathleen
has alluded, some people might say 'the lantern lights things up but
doesn't seem to do anything special. I couldn't even set the dung on
fire!'.
Mike Duncan
http://www.boston.quik.com/mduncan/
Andrew Plotkin
> Gregg Carroll wrote:
>> I wasn't hinting at all that IF should be populated with Galatea's. God
>> no. That would get really tedious really quickly, *unless* conversation
>> is the only objective.
> Er, what? How can a greater range of responses be a detriment?
The same way putting five thousand extra rooms into a map can be a
detriment. (In fact, I was just posting about _Starship Titanic_ on
the PC adventure newsgroup...)
The most important thing for a game to communicate is -- what range of
action is available for the player to explore? If it's too small, the
game is uninteractive; if it's too large, the player futzes around for
hours and never makes progress. If the player is searching the wrong
area of the action range, the game has to indicate somehow that he's
flailing in the shallows.
If the huge, explorable conversation web is actually part of the game
-- if the half-hour-long dialogue amounts to the player having *gotten
somewhere* -- then it's a benefit.
If it leaves the game state unchanged, then it's a detriment, because
the player is going to figure that out (players not being morons,
unfortunately for us authors :-) and then (1) feel resentful and (2)
not bother talking to the next character he meets.
--Z
"And Aholibamah bare Jeush, and Jaalam, and Korah: these were the borogoves..."
*
* Make your vote count. Get your vote counted.
Adam Cadre
> If the huge, explorable conversation web is actually part of the game
> -- if the half-hour-long dialogue amounts to the player having *gotten
> somewhere* -- then it's a benefit.
>
Well, here's an example. In VARICELLA, Miss Sierra brusquely cuts off
all attempts to ask her about her personal life. And you don't need to
know any of it to accomplish the ostensible goal of winning the regency.
But there is a fairly involved sequence of things you can do elsewhere
in the palace that will allow you to return to her room and learn her
life story. You can also get pieces of it without doing a lot of work
by asking her about non-obvious things, like various scenery objects. I
don't know if you can still accomplish the stated goal of the game if you
do this.
Does learning about Sierra's childhood count as "getting somewhere"? Does
this change "the game state"? Would she be a better NPC if most of her
dialogue were dropped, such that she more effectively channeled the
player's efforts toward the PC's narrow goal?
Gregg V. Carroll
>>> I wasn't hinting at all that IF should be populated with Galatea's. God
>>> no. That would get really tedious really quickly, *unless* conversation
>>> is the only objective.
>> Er, what? How can a greater range of responses be a detriment?
> The most important thing for a game to communicate is -- what range of
> action is available for the player to explore? If it's too small, the
> game is uninteractive; if it's too large, the player futzes around for
> hours and never makes progress. If the player is searching the wrong
> area of the action range, the game has to indicate somehow that he's
> flailing in the shallows.
Yeah, what he said.
I think what Kathleen is getting at is that complexity for its own sake
enhances the experience. When you think about it, though, it's one thing if
the bouncer in question is the only point of complexity, but if the entire
game was that way, including any number of items that can be manipulated in
any number of ways, the game starts being less like a game and more like a
simulation.
In a fictional game that takes place in my house, in which I'm required to,
say, fix a proken water pipe, and there's only a handful of objects that
will allow me to do it, that's a challenge. But if the complexity is so high
that it's a simulation, I'm more likely to type in
> OPEN PHONEBOOK, CONSULT PHONEBOOK ABOUT PLUMBER, CALL PLUMBER.
It's great that the game knows what I'm talking about, even without
prompting me beforehand that it knows what a plumber is, but it's not very
satisfying as a game.
Or even
> TALK TO PLUMBER ABOUT POLITICS.
[3,456 moves later...]
The plumber suddenly stops in mid-sentence and says, "Don't take this the
wrong way, this is a great discussion, but shouldn't I take care of this
broken pipe of yours in the basement?"
> X BASEMENT.
The basement is currently 8 feet deep with water.
Gregg
MFischer5
>
>>>> I wasn't hinting at all that IF should be populated with Galatea's.
>God
>>>> no. That would get really tedious really quickly, *unless* conversation
>>>> is the only objective.
>
>>> Er, what? How can a greater range of responses be a detriment?
>
>> The most important thing for a game to communicate is -- what range of
>> action is available for the player to explore? If it's too small, the
>> game is uninteractive; if it's too large, the player futzes around for
>> hours and never makes progress. If the player is searching the wrong
>> area of the action range, the game has to indicate somehow that he's
>> flailing in the shallows.
>
>Yeah, what he said.
>
>I think what Kathleen is getting at is that complexity for its own sake
>enhances the experience.
Uhmm... no. I think that's what Adam was getting at. I agree with Zarf.
Kathleen (checking nametag again, just to be sure)
-- Prized Posession (Comp 2001):
http://www.ifarchive.org/if-archive/games/competition2001/inform/possess
-- Masquerade (Comp 2000):
http://www.ifarchive.org/if-archive/games/zcode/Mask.z5
-- The Cove (Art Show 2000):
http://www.ifarchive.org/if-archive/games/zcode/Cove.z5
Gregg V. Carroll
>> I think what Kathleen is getting at is that complexity for its own sake
>> enhances the experience.
> Uhmm... no. I think that's what Adam was getting at. I agree with Zarf.
>
> Kathleen (checking nametag again, just to be sure)
Whoops! Okay, now I'm the one that's blushing.
Gregg
Sean T Barrett
>Would she be a better NPC if most of her
>dialogue were dropped, such that she more effectively channeled the
>player's efforts toward the PC's narrow goal?
It's worth remembering that you personally have a particular disinterest
in immersion and more interest in "lots of neat responses" even
if those responses don't bear on the game/overarching story;
your review of "Got ID?", I think, was explicit about this.
So at a certain point (I think almost immediately), this
discussion becomes just a matter of taste. I think for a lot of
people the focus is on the story-of-what-happens-to-the-player,
and the elements of the backstory that have an immediate or
causal relationship with what the player does or will do are
important, but once you move away from those things the information
may be entertaining or character-revealing but those are less
significant than they would be in a static work (which has
no interactive linchpin).
But that all depends on the point of the work. If you set up
the work to not be about the player achieving some goal (reaching
some fixed end point)--e.g. as most (all?) of the IF Art Show entries
have done ("not done"?), the focus now shifts to exploring a space or
a world or what have you, and at that point most of the aforementioned
players presumably wouldn't have any problem with it.
SeanB
Sam Barlow
> > If the huge, explorable conversation web is actually part of the game
> > -- if the half-hour-long dialogue amounts to the player having *gotten
> > somewhere* -- then it's a benefit.
> >
> > If it leaves the game state unchanged, then it's a detriment [...]
>
> this change "the game state"? Would she be a better NPC if most of her
> dialogue were dropped, such that she more effectively channeled the
> player's efforts toward the PC's narrow goal?
>
Varicella belongs to the puzzle methodology shared by Suspect and
Corruption that means the gameworld is *more* fleshed out that one
would expect. The puzzle is in sorting out and optimizing a route
through the world that brings about the optimum ending. As such there
are many play-throughs that result in no game progress but add to the
player's knowledge of the world. The cool thing about this is that it
often means the writer adds more depth to his world -- incidental to
the puzzles perhaps, though as the puzzles mainly arise from plot not
necessarily so -- which is why Corruption, Suspect and Varicella are
among my favourite IFs.
So the agreement between player and writer is 'do anything, but do not
expect progression to arise from interaction'.
I think Andrew's point is that overloading a NPC with 100s of
puzzle-redundant conversation topics would be to place
disproportionate emphasis on the rendering of that NPC in the midst of
a shallow gameworld. If the writer feels that the NPC needs such depth
then perhaps the game should be rewritten so that the NPC is more
important within its puzzle structure -- puzzles being the carrot and
stick of the fiction. Otherwise the NPC stands out like a sore thumb
declaring 'The art, here'. Of course if the NPC is just a puzzle--i.e.
the writer has added him as a locked door, not as Art--adding extra
conversation topics _just because_ is going to further draw attention
to the NPC's shallowness.
A character who can coverse on a 1,000 topics but has no discernable
impact on the gameworld is a conversation machine not a character. A
still-life with 50% of its brushstrokes dedicated to one piece of
fruit amongst the ten in its bowl draws attention to the poor
rendering of the other nine rather than causing an illusion of Real
Fruit in the one.
Miss Sierra _does_ impact the gameworld and so adding depth to her
character is not misplaced. And on the subject of the elaborate
sequence of events in order to ask about her personal life -- I doubt
anyone followed this sequence in order to ask those questions; more
likely in the interaction/exploration necessary in a save/restore game
like Varicella, the player chanced upon this added depth whilst
experimenting. A save/restore game must reward experimentation
otherwise it becomes very dull -- if only the final playthrough
results in success, you have to give encouragement in the other
playthroughs.
Yep.
--Sam.
Magnus Olsson
>> If the huge, explorable conversation web is actually part of the game
>> -- if the half-hour-long dialogue amounts to the player having *gotten
>> somewhere* -- then it's a benefit.
>>
>> If it leaves the game state unchanged, then it's a detriment [...]
>
>Well, here's an example. In VARICELLA, Miss Sierra brusquely cuts off
>all attempts to ask her about her personal life. And you don't need to
>know any of it to accomplish the ostensible goal of winning the regency.
>But there is a fairly involved sequence of things you can do elsewhere
>in the palace that will allow you to return to her room and learn her
>life story.
(...)
>Does learning about Sierra's childhood count as "getting somewhere"? Does
>this change "the game state"?
I was going to write a lot about backstory, providing depth to the
character, etc, but I think it's better to use another angle:
My goal - or at least one, major goal - when playing "Varicella" was
to find out what was going on behind the scenes and how everything fit
together, including what drove the NPCs to behave the way they did.
So for me, learning about Sierra's childhood definitely was "getting
somewhere".
>Would she be a better NPC if most of her
>dialogue were dropped, such that she more effectively channeled the
>player's efforts toward the PC's narrow goal?
In one word, no. And the game would be worse off for it.
As for the more general argument: I think I must take back, or at
least modify, what I wrote before. Casting a Galatea-type NPC as, say,
the bouncer outside a club could detract from a game, but it could
also be a positive thing. I think it depends on what the player
expects to do and what the author wants to say.
Magnus Olsson
Joe Mason <jcm...@student.math.uwaterloo.ca> wrote:
>Now I have the urge to replay Chix Dig Jerks and just repeat
>
> >GIRL, JUMP AROUND
>
>over and over again.
What happens when you do that?
Joe Mason
>In article <9ujnto$9si$1...@watserv3.uwaterloo.ca>,
>Joe Mason <jcm...@student.math.uwaterloo.ca> wrote:
>>Now I have the urge to replay Chix Dig Jerks and just repeat
>>
>> >GIRL, JUMP AROUND
>>
>>over and over again.
>
>What happens when you do that?
I dunno, I don't have a copy handy to check.
Joe
Kathleen M. Fischer
>I think this is the major difference between ask/tell and conversation
>menus. If you want lengthy dialogue with a lot of back and forth
>interaction (like a real conversation), then it seems clear to me that a
>menu system would be more suitable to the game. Ask/tell would be very
>likely to end up with "guess the noun".
>
>If you on the other hand want to be sparse with dialogue, ask/tell would
>be a logical choice.
Unless you were writing _Galatea_, in which case the reverse would be true.
:)
I think ASK/TELL works best when it is clear to the player what they are
supposed to be asking/telling, and it can be expressed in simple, singular,
obvious thoughts with little cause of confusion.
ASK JOE ABOUT TREE
The more objects you have, the more rooms, the more concepts (love, loss,
whatever) the less workable it becomes unless the nature of the NPC is such
that it bounds the discussion (the ticket taker at the theatre whose sole
job
really is to take tickkets)
As Magnus pointed out (I think), complex metaphysical discussions with fine
gradations are, from what I've seen so far, still best handled by menus or
menu-like (TALK TO, TOPIC, ...etc) systems, or better still a combination of
both.
Kathleen (IMHO, of course)
M. D. Krauss
On 4 Dec 2001 23:04:55 GMT
m...@df.lth.se (Magnus Olsson) wrote:
That depends. Are we talking about a more simulationist game where the
player gets to define the PC? Or a more structured narrative type story
where the PC is well defined? In the latter case (and that's the direction
of most modern IF), you wouldn't even want to allow those options. Either
the PC's dialogue will be given in response to "tell stranger about king"
(or "talk to stranger" or whatever) or more likely, immediately without
interactivity. In the example given, you'd have to decide how you want to
manage coding the *entire rest of the game* in such a way as to make sense
and flow naturally from this nexus no matter which of those radically
different responses the player selects before we even consider whether to
use ask/tell or a menu.
> And this is a simple example - what if the NPC was a priest asking you
> about your philosophical position regarding his religion?
Exactly the same. Either the PC has an opinion, in which case we don't
*want* the player to choose a response (although we may want the player to
choose whether to respond, which is a different matter and handled well by
ask/tell) or the player is expected to actually affect the course of the
game here, in which case the complexity of human behaviour will lead
inevitably to a combinatorial explosion.
> You could, conceivably, have the NPC ask you a series of simple
> questions, but that would smack of interrogation and I think it would
> be rather clumsy; not the effect I'm after.
Agreed that this would be Very Bad.
--
To email me convert my address to something resembling reason
Jon Ingold
"Adam Cadre" <gri...@drizzle.com> wrote in message
news:9uk68t$qga$1...@drizzle.com...
> Andrew Plotkin wrote:
> > If the huge, explorable conversation web is actually part of the game
> > -- if the half-hour-long dialogue amounts to the player having *gotten
> > somewhere* -- then it's a benefit.
> >
> > If it leaves the game state unchanged, then it's a detriment [...]
<snip Varicella example>
It's more that - it's a detriment *if the player has done it expecting it to
change the game state*. So in the Varicella example, no-one expected it to
lead the player suddenly being able to access the Great Hall of Wanderings.
In the night-club game example above, the player may well expect that his
chat with the bouncer would let him into the club.
So the game really needs to flag (somehow) that the extra
conversation-web-stuff is in the "neat backstory" category and not the
"vital to proceed" one. How you do that -- is a matter of design.
Jon
M. D. Krauss
Andrew Plotkin <erky...@eblong.com> wrote:
> Adam Cadre <gri...@drizzle.com> wrote:
> > Gregg Carroll wrote:
> >> I wasn't hinting at all that IF should be populated with Galatea's. God
> >> no. That would get really tedious really quickly, *unless* conversation
> >> is the only objective.
>
> > Er, what? How can a greater range of responses be a detriment?
>
> The same way putting five thousand extra rooms into a map can be a
> detriment. (In fact, I was just posting about _Starship Titanic_ on
> the PC adventure newsgroup...)
>
> The most important thing for a game to communicate is -- what range of
> action is available for the player to explore? If it's too small, the
> game is uninteractive; if it's too large, the player futzes around for
> hours and never makes progress. If the player is searching the wrong
> area of the action range, the game has to indicate somehow that he's
> flailing in the shallows.
I disagree. There's a difference between limiting the range of action and
communicating what action is needed to progress. In some of the best IF I
find that the immersion in to the plot brings me in sync with the PC, and
leads me to know just what action is necessary (the action the PC would
take, generally), which in turn draws me deeper in to the plot and
atmosphere of the piece; and having a richer fuller world of
irrelevant-but-interesting things doesn't detract at all.
Of course, providing such extras is a kind of simulationism, so I guess
I'm partly contradicting what I said elsewhere in this thread... But as
ol-whatsisname said, "of course I contradict myself; I am wide and contain
multitude" (or something like that).
> If the huge, explorable conversation web is actually part of the game
> -- if the half-hour-long dialogue amounts to the player having *gotten
> somewhere* -- then it's a benefit.
>
> If it leaves the game state unchanged, then it's a detriment, because
> the player is going to figure that out (players not being morons,
> unfortunately for us authors :-) and then (1) feel resentful and (2)
> not bother talking to the next character he meets.
If the game is generally well written enough that the player trusts it
they will be confident that nothing useful is going to come of a
conversation with a bouncer unless there's something to indicate that it
will. If I have reason to be concerned that I'll be getting in a fight and
the bouncer might be able to give some tips on self defense, so I wander
in to a massively complex conversation web only to find out hours later
that it was a waste of time, I might be miffed (or I might say it was an
interesting sidetrack and just go on with things); but if I have seen no
particular reason to talk to this bouncer, and I start to, and he actually
responds in some appropriate fashion, I'll be pleased.
In a more traditional puzzle game where I'm going around fiddling with
everything and picking everything up to stuff in my appearantly massive
backpack because you never know what will be usefull, red herrings are a
nuisance. In a more natural feeling fiction, they are fun.
Dropping Galatea in Zork would be Bad. Dropping Galatea in Anchorhead
might be Very Cool.
-M
>
> --Z
>
> "And Aholibamah bare Jeush, and Jaalam, and Korah: these were the borogoves..."
> *
> * Make your vote count. Get your vote counted.
Magnus Olsson
M. D. Krauss <MDKraus...@home-nospam.com-nospam> wrote:
>I'm Really Trying not to get involved in this one. :)
>
>On 4 Dec 2001 23:04:55 GMT
>m...@df.lth.se (Magnus Olsson) wrote:
>> To exemplify this, here's an example where I don't really see how to
>> implement a dialogue with ask/tell
>> But what if the stranger asks you, "Well, what do you think of king
>> Hippokephalos?"
>>
>> With conversation menus, you can get a choice
>>
>> 1: "King whatsisname? Never heard of him (hic)"
>> 2: "I don't really have an opinion - I'm a stranger here"
>> 3: "The greatest ruler that ever lived! May he continue to do so!"
>> 4: "He's a tyrant! Death to him, and long live the revolution!"
>> 5: "He's a good ruler, but who will succeed him?"
>>
>> - but how do you do this with an ask/tell interface?
>
>That depends. Are we talking about a more simulationist game where the
>player gets to define the PC? Or a more structured narrative type story
>where the PC is well defined? In the latter case (and that's the direction
>of most modern IF), you wouldn't even want to allow those options.
To begin with, you've forgotten one possibility: The PC may have a
well-defined personality, a complete set of opinions, and so on - but
he might want to lie to the stranger, or at least not give those
opinions away. In this case, the game could very well have told the
player that the PC hates king Hippokephalos like the plague - but the
player would still have to figure out which response would serve him
the best in this situation. In other words, a plain, old-fashioned
puzzle.
Secondly, I think you're wrong about the "direction of modern IF", but
I won't go into this now; it's too late at night.
Anyway, I wasn't really asking you for advice on how to define the
PC's character; this was just a constructed example where conversation
menus work and ask/tell doesn't seem to work.
M. D. Krauss
m...@df.lth.se (Magnus Olsson) wrote:
Right, I understood your intent. What I'm saying, in fact is just that --
this is a constructed example. A bit to constructed. In reality, there
aren't likely to be such a wide variety of responses that the author will
want to code for. I mean, come on... Assuming the responses are actually
significant, this kind of thing will quickly lead to a completely
unmanageable combinatorial explosion. If I didn't make that meaning clear,
I do apologize.
As to the "direction of modern IF" -- perhaps I generalized a bit to much
there. Still, I think most of what we've all been impressed by in the past
few years has been work that's less simulationist, more structured. Not to
say, non-interactive; but following forms and conventions that give the
author more control and allow an actual story to be told.
-M
> --
> Magnus Olsson (m...@df.lth.se, m...@pobox.com)
> ------ http://www.pobox.com/~mol ------
Magnus Olsson
M. D. Krauss <MDKraus...@home-nospam.com-nospam> wrote:
>On 5 Dec 2001 23:23:13 GMT
>m...@df.lth.se (Magnus Olsson) wrote:
>> >> With conversation menus, you can get a choice
>> >>
>> >> 1: "King whatsisname? Never heard of him (hic)"
>> >> 2: "I don't really have an opinion - I'm a stranger here"
>> >> 3: "The greatest ruler that ever lived! May he continue to do so!"
>> >> 4: "He's a tyrant! Death to him, and long live the revolution!"
>> >> 5: "He's a good ruler, but who will succeed him?"
>> Anyway, I wasn't really asking you for advice on how to define the
>> PC's character; this was just a constructed example where conversation
>> menus work and ask/tell doesn't seem to work.
>
>Right, I understood your intent. What I'm saying, in fact is just that --
>this is a constructed example. A bit to constructed. In reality, there
>aren't likely to be such a wide variety of responses that the author will
>want to code for.
You sound awfully certain of what authors want to do. I wouldn't
be so certain if I were you; keep your eyes open, there might be an
example of this kind of conversation in the next Comp.
>I mean, come on... Assuming the responses are actually
>significant, this kind of thing will quickly lead to a completely
>unmanageable combinatorial explosion.
You're tacitly making a couple of unwarranted assumptions here.
Even if the answers are significant, they do not have to lead to
major plot branching. And even if they do, there does not have to
be lots and lots of these kinds of branches. And even if there
are, the choices need not be independent.
You only get a combinatorial explosion if all of those assumptions
are met. I said nothing of the sort.
>As to the "direction of modern IF" -- perhaps I generalized a bit to much
>there.
With all respect, you generalized far too much.
>Still, I think most of what we've all been impressed by in the past
>few years has been work that's less simulationist, more structured.
An aside: You're using a rather uncommon definition of "simulationist"
- it usually seems to refer to IF where the game environment is
simulated from first principles rather than scripted, and, by
extension, of (hypothetical) IF where plot isn't scripted, but emerges
from the simulated environment ("simulate a world, a villain and a
cause, and plot will emerge by itself"). But you seem to use it
to refer to IF where the PC is defined by the player's actions.
>Not to
>say, non-interactive; but following forms and conventions that give the
>author more control and allow an actual story to be told.
But what you were saying actually seemed to be that the direction IF
is taking is towards non-interactivity.
I'm all for having well-defined PCs under authorial control. But if
authorial control means that the player can't even choose which answers
to give during a questioning, what's left of interactivity? Why stop
at conversations - doesn't every action the PC takes serve to define
his/her personality? If this is the direction modern IF is taking, we'll
end up with "games" where every action is under authorial control,
even picking up things or examining them:
|>take sculpture
|Thievery would be against your moral code.
|
|>buy sculpture
|The possession of worldly goods would be against your vows.
|
|>examine sculpture
|You're not interested in art.
|
|>break sculpture
|Such a senseless act of vandalism would wreak havoc with your
|karmic record.
|
|>quit
|Why quit now, now that things are just getting interesting?
Daryl McCullough
>The PC may have a well-defined personality, a complete set
>of opinions, and so on - but he might want to lie to the
>stranger, or at least not give those opinions away. In this
>case, the game could very well have told the player that
>the PC hates king Hippokephalos like the plague - but the
>player would still have to figure out which response would
>serve him the best in this situation. In other words, a plain,
>old-fashioned puzzle.
Thanks for the examples. I think that what it boils down
to is this:
Making a statement can be an action in its own right.
Lying, insulting, flattering, etc. are linguistic actions
that are not completely characterized by the information
being exchanged. The say/tell mechanism, however, views
language purely in terms of information exchange. In a
game where speech acts are important, I don't think it
is possible to accomplish everything with say/tell.
(Except, as somebody pointed out already, when the
speech acts can be as simple as "yes" and "no").
However, I'm a little uncomfortable with the menu system,
as well. The problem is that any small number of choices
seems too limited. Giving the player the choice of saying
1. The king is a big fat jerk.
2. The king is a wise and benevolent leader.
3. The king looks pretty sexy in that robe.
is like having a rock in a game, and giving the player
a menu of options
1. Break the window with the rock.
2. Distract the guard by tossing the rock into the garbage can.
3. Hit the ogre on the head with the rock.
A menu with a small number of choices makes it more like
a "choose your own adventure" game, rather than text-based
IF.
If you want to allow the user to treat language as action,
on a par with physical actions, then there needs to be a
rich environment for creating speech acts that is comparable
to the environment for creating physical acts. I don't really
know how to do that in a practical way.
What I'd like (though I don't know how to do it) is for
the player to be able to construct speech acts from primitives
in the same way that physical actions can be constructed out
of primitive actions such as "get object", "put object in container",
"give container to NPC".
ems...@mindspring.com
> As for the more general argument: I think I must take back, or at
> least modify, what I wrote before. Casting a Galatea-type NPC as, say,
> the bouncer outside a club could detract from a game, but it could
> also be a positive thing. I think it depends on what the player
> expects to do and what the author wants to say.
This is a fair assessment, of course, but it still turns around a
semi-traditional view of how NPCs can be used in games. One of my
reasons for wanting to develop a conversation system of some
complexity in the first place, of course, was not just to be able to
spruce up the backstory of my games ("I'll still have a door-guarding
NPC, but this time he'll be an angst-ridden Siberian albino refugee
with Tourette's Syndrome...") but to be able to write games where the
trick was to get the NPC into a certain state of mind or attitude.
I still want to see a *real* intrigue game where you have to persuade
NPCs to trust you and play them off each other...
ES
Magnus Olsson
ems...@mindspring.com <ems...@mindspring.com> wrote:
>news:<9ul1a9$nqu$1...@news.lth.se>...
>> As for the more general argument: I think I must take back, or at
>> least modify, what I wrote before. Casting a Galatea-type NPC as, say,
>> the bouncer outside a club could detract from a game, but it could
>> also be a positive thing. I think it depends on what the player
>> expects to do and what the author wants to say.
>
>This is a fair assessment, of course, but it still turns around a
>semi-traditional view of how NPCs can be used in games. One of my
>reasons for wanting to develop a conversation system of some
>complexity in the first place, of course, was not just to be able to
>spruce up the backstory of my games ("I'll still have a door-guarding
>NPC, but this time he'll be an angst-ridden Siberian albino refugee
>with Tourette's Syndrome...") but to be able to write games where the
>trick was to get the NPC into a certain state of mind or attitude.
Yes, of course. Both _Galatea_ and _Best of 3_ are a very different
type of IF from what we were discussing here; but the point was
how a Galatea-type NPC would work in a more "traditional" game.
>I still want to see a *real* intrigue game where you have to persuade
>NPCs to trust you and play them off each other...
A _Varicella_ where you have to persuade the NPC's to do your bidding?
Yes, that would really be interesting.
I think that we've so far only scratched the surface of IF with
multiple interacting NPC's.
Peter Ketting
<ems...@mindspring.com> wrote in message
> I still want to see a *real* intrigue game where you have to persuade
> NPCs to trust you and play them off each other...
Just out of curiousity, has anyone ever tried to implement a sort of
Nural-Network
for Artificial Inteligence in Interactive Fiction?
Cheers,
Peter
Magnus Olsson
Not that I know of.
And I don't think neural networks are very well suited for IF (but
it would be interesting to be proved wrong!)
To begin with, NNs aren't really artificial intelligence; they are
a way of implementing systems that are "trained" to do certain things,
such as pattern recognition, but it is on a much lower level than AI.
I suppose you could, in principle, use a NN to implement an NPC
that observed patterns in the PCs actions and in some way learned
to interpret the PC, or anticipate the player's actions.
The problem with this is that NNs are slow learners. It takes
a lot of input data to train an NN to do something, and I think this
is simply too slow for IF applications.
M. D. Krauss
m...@df.lth.se (Magnus Olsson) wrote:
Not certain at all. I said "likely". If in fact there is such an example
in the next comp I'll be very pleased. Certainly I'm not saying menu
systems aren't useful or that there could never be a case that calls for
them... just that such a case is, in my opinion, rare.
> >I mean, come on... Assuming the responses are actually
> >significant, this kind of thing will quickly lead to a completely
> >unmanageable combinatorial explosion.
>
> You're tacitly making a couple of unwarranted assumptions here.
> Even if the answers are significant, they do not have to lead to
> major plot branching. And even if they do, there does not have to
> be lots and lots of these kinds of branches. And even if there
> are, the choices need not be independent.
>
> You only get a combinatorial explosion if all of those assumptions
> are met. I said nothing of the sort.
Good point there. Sorry.
> >As to the "direction of modern IF" -- perhaps I generalized a bit to
much
> >there.
>
> With all respect, you generalized far too much.
As you wish.
> >Still, I think most of what we've all been impressed by in the past
> >few years has been work that's less simulationist, more structured.
>
> An aside: You're using a rather uncommon definition of "simulationist"
> - it usually seems to refer to IF where the game environment is
> simulated from first principles rather than scripted, and, by
> extension, of (hypothetical) IF where plot isn't scripted, but emerges
> from the simulated environment ("simulate a world, a villain and a
> cause, and plot will emerge by itself"). But you seem to use it
> to refer to IF where the PC is defined by the player's actions.
I'm not using an uncommon definition. I was speaking in terms of degrees.
Only a true simulationist approach can reasonably accomodate any action a
player chooses to take. Usually, the author picks and chooses what player
choices to allow based on what story the author wants to tell. The most
well-recieved work these days seems to limit player choices by
circumstances rather then by simply saying "you can't do that", but the
limit is still there, and would not be present in a true simulationist
work. I think it's fairly clear that in a simulationist work, the PC is
defined by player actions, and that this is a "simulationist trait" (so to
speak), and that IF where the player is less free to take actions outside
the spectrum of what the author thinks the PC would do is "less
simulationist".
> >Not to
> >say, non-interactive; but following forms and conventions that give the
> >author more control and allow an actual story to be told.
>
> But what you were saying actually seemed to be that the direction IF
> is taking is towards non-interactivity.
Certainly if I gave that impression I apologize... Oh hell, no I don't. I
said nothing of the sort.
> I'm all for having well-defined PCs under authorial control. But if
> authorial control means that the player can't even choose which answers
> to give during a questioning, what's left of interactivity? Why stop
> at conversations - doesn't every action the PC takes serve to define
> his/her personality? If this is the direction modern IF is taking, we'll
> end up with "games" where every action is under authorial control,
> even picking up things or examining them:
>
> |>take sculpture
> |Thievery would be against your moral code.
> |
> |>buy sculpture
> |The possession of worldly goods would be against your vows.
> |
> |>examine sculpture
> |You're not interested in art.
> |
> |>break sculpture
> |Such a senseless act of vandalism would wreak havoc with your
> |karmic record.
> |
> |>quit
> |Why quit now, now that things are just getting interesting?
Well, um, putting aside the burning question of whether players should be
allowed to quit...
This actually isn't completely unreasonable, assuming the PC is some sort
of extremist monk or something. I wouldn't want to be prevented from
examining something, of course, but it's reasonable even there for the
description to be filtered through the eyes of a PC who doesn't find it
interesting. Personally, I find it horribly destructive of mimises when
I'm allowed to (or in the worst cases, required to) do things that the PC
would never even consider. It really represents, I think, one of the most
difficult challenges of the medium -- how much do we limit the player?
I have no good answers there, but just think that we need to keep the
question open and not immediately assume that all limitation is bad.
-M
> --
> Magnus Olsson (m...@df.lth.se, m...@pobox.com)
> ------ http://www.pobox.com/~mol ------
Magnus Olsson
M. D. Krauss <MDKraus...@home-nospam.com-nospam> wrote:
>On 6 Dec 2001 11:36:20 GMT
>m...@df.lth.se (Magnus Olsson) wrote:
>
>> In article
><20011205194338.7c3d...@home-nospam.com-nospam>,
>> M. D. Krauss <MDKraus...@home-nospam.com-nospam> wrote:
>> You sound awfully certain of what authors want to do. I wouldn't
>> be so certain if I were you; keep your eyes open, there might be an
>> example of this kind of conversation in the next Comp.
>
>Not certain at all. I said "likely". If in fact there is such an example
>in the next comp I'll be very pleased. Certainly I'm not saying menu
>systems aren't useful or that there could never be a case that calls for
>them... just that such a case is, in my opinion, rare.
Just to make it clear what I'm reacting to: it's not your opinions
on menu systems. As I said, I share your reservations about conversation
menus, I understand why people don't like them and so on. It's just
that I can't think of any other way to achieve certain things that
I want to have in my WIP.
This, I must confess, is as much a failing on my part as it is on
the ask/tell system. It is quite possible that somebody else could
achieve those things with an ask/tell system, but personally I don't
know how to do that.
My example with the conversation about the king was really just an
attempt to show the kind of thing that I *can* do with menus, but
don't know how to do with ask/tell.
But then you responded by saying, in effect, that authors aren't
likely to want to do that kind of thing, and that modern IF is moving
away from that anyway.
I'm sure you didn't mean to, but what you did was actually to tell me
that I really didn't want to do what I was doing. Again, you probably
didn't mean to say that, but as with all communication, it's not
intent that counts, but what's actually said.
A general observation (which is not about your post, but in general) is
that you often get an exchange on this group and other groups involved
with creative efforts where A says "How would you do X in a game (or
novel, painting, computer language)" and B replies "Noone in his right
mind would want to do that". In some cases, B is actually right, in
some cases B is wrong, but in most cases this is simply communication
failure - A fails to communicate the entire context, and B
misunderstands. So I think we should all be very careful about saying
that something won't work, or "you don't really want to do that".
>> You're tacitly making a couple of unwarranted assumptions here.
>> Even if the answers are significant, they do not have to lead to
>> major plot branching. And even if they do, there does not have to
>> be lots and lots of these kinds of branches. And even if there
>> are, the choices need not be independent.
>>
>> You only get a combinatorial explosion if all of those assumptions
>> are met. I said nothing of the sort.
>
>Good point there. Sorry.
Well, you don't have to
>apologize, really; acknowledging that the point exists is enough.
>> >As to the "direction of modern IF" -- perhaps I generalized a bit to
>much
>> >there.
>>
>> With all respect, you generalized far too much.
>
>As you wish.
No offence intended, but what is "as you wish" supposed to mean in
this context? I'm just puzzled - it sounds like the kind of thing
you say when you're relinquishing a point just to humour an opponent.
>> >Still, I think most of what we've all been impressed by in the past
>> >few years has been work that's less simulationist, more structured.
>>
>> An aside: You're using a rather uncommon definition of "simulationist"
>I'm not using an uncommon definition.
You aren't? Well, I suppose I must have missed something, because
I've never seen anybody use the word "simulationist" applied to
the PC before.
Or perhaps this is just a matter of semantics, so let me backtrack:
Perhaps you used the usual definition of "simulationist", but you
just applied it in an unusual context.
>Only a true simulationist approach can reasonably accomodate any action a
>player chooses to take.
Agreed, for a reasonable definition of "reasonably accommodate".
>I think it's fairly clear that in a simulationist work, the PC is
>defined by player actions,
Could be.
For one kind of simulationist work, where the PC is simulated as well
as the world around her, this is true.
But I can think of other kinds of simulationist works as well.
You could have a pre-defined PC put in a simulationist world -
I imagine it would be like live role-playing, where you can do
anything you like as long as you stay in character. How to implement
this without destroying the simulation is a tricky problem - but at
least one can imagine the *player* imposing restrictions on the PC.
I can also imagine a simulationist work that more or less dispenses
with the PC; a return to the old "nameless, featureless generic
adventurer". In essence, the player interacts directly with the game
world, not via a PC intermediary, and the player's actions are the
player's actions, not actions defining the PC.
>> But what you were saying actually seemed to be that the direction IF
>> is taking is towards non-interactivity.
>
>Certainly if I gave that impression I apologize... Oh hell, no I don't. I
>said nothing of the sort.
Note that I wrote "seemed". You did write that the direction IF is
taking is towards non-interactivity in dialogue; I took the liberty
of extrapolating a bit.
>> |>take sculpture
>> |Thievery would be against your moral code.
>> |
>> |>buy sculpture
>> |The possession of worldly goods would be against your vows.
>> |
>> |>examine sculpture
>> |You're not interested in art.
>> |
>> |>break sculpture
>> |Such a senseless act of vandalism would wreak havoc with your
>> |karmic record.
>> |
>> |>quit
>> |Why quit now, now that things are just getting interesting?
>
>Well, um, putting aside the burning question of whether players should be
>allowed to quit...
>
>This actually isn't completely unreasonable, assuming the PC is some sort
>of extremist monk or something.
Agreed. It is not at all unreasonable; it's just that I don't think
IF is going in that direction or that IF *should* go in that direction.
For individual works to go in that direction would be interesting.
For the entire genre... well...
Jon Thompson
<peterk...@myrealbox.com> started to type:
Not a neural network, but I am playing about with the idea of a large
number of "attitude" variables and modifiers for an NPC. When a
request is received by the NPC the state of the attitude variables is
inspected in conjunction with the modifiers and a decision is made as
to whether or not the NPC will comply, do nothing, or even something
contrary. The modifiers are there as sort of neuro transmitters that
either inhibit or promote a certain attitude (give an NPC a drug and
see how much he will comply with your request while stoned, etc). The
complexity of the character's mind comes from the logic you use to
evaluate these variables and modifiers when you ask the character to
do something. Interacting with the character can change his attitudes
to you over time. The art comes in designing the initial variable
states and underlying logic for each NPC to give them individual
"personalities".
Even with a fairly small set of attitude variables with a range of 0 -
10 and no modifiers the range of possible states is very large indeed.
You can also introduce probabilities to inject a bit of chance into
responses so that the underlying logic is not 100% predictable for a
given game state.
--
齯滌`偕爻,虜,齯滌`偕爻,虜,齯滌`偕爻,虜,齯滌`偕爻,虜,齯滌`偕爻,虜,
"I never deal with the common man. The common man has no
spirituality. The common man thinks Ganesha is Dennis
The Menace's Dog." - Hugo Rune
齯滌`偕爻,虜,齯滌`偕爻,虜,齯滌`偕爻,虜,齯滌`偕爻,虜,齯滌`偕爻,虜,
Alan DeNiro
How much work has been done with this? I think this is a really
interesting avenue of NPC interaction. Emily Short touched on this
briefly in her essay on NPCs
(http://emshort.home.mindspring.com/NPC3.htm), and that (and Daryl's
post) got me thinking. If there's a list of 5-12 verbs for
conversation (from the more standard GREET to FLATTER, TAUNT, BADGER,
whatever), then there can be various shades of meaning, depending on
the response of the NPC. Suddenly the different between these speech
acts and HUG, KISS, SHOW, and other body language clues don't seem
that far apart--which might be a good way to go, if you're of the
opinion that communication is more than dialogue.
Some more advantages:
1. The verbs are active and kinetic, and perhaps vivid words in their
own right. It ties conversation into one of the key strengths of IF in
the first place--a direct succinct ability to impact a virtual
environment (counting the NPC as part of that environment, here).
2. It could conceivably create an easier shorthand--or perhaps a
stronger emotional bond--between the game-player and the player
character. Active conversation verbs ARE emotionally loaded. If the
game designer has an NPC acting in a certain way to create a
feeling-based effect on the player, then the player, with these tools,
is closer to the action. If a PC's lover is on the verge of a
breakdown, then SOOTHE would have one effect, FLATTER quite a
different effect, and BADGER might have disasterous consequences. This
can be done with menus, of course, but the difference is that, with
the right type of IF, here the ball is more in the PLAYER's court to
volley or lob back as he or she sees fit, depending on the reactions
elicited.
Note that these verbs would only deal with the ATTEMPT to flatter,
soothe, etc. The NPC might be coded to have different ideas (>SOOTHE
LOVER leading to "Stop trying to butter me up! Where were you when I
needed you most?"). This in itself might be an interesting mimetic
side-affect. And in the spirit of "I'm not a linguist but I play one
on USENET", I don't even know if this is the way language acts _work_,
that is:
a) I see someone who may or may not say something
b) I decide the effect that I want to have on this person
c) Language act is formed, and spoken
I would guess that in real life, language is much more hardwired, that
(b) would be a very subconscious step. I defer to others on this one.
But even if this isn't "realistic" in a linguistic sense, this model
of conversation might be realistic in a game-model sense.
Disadvantages are obvious--you can't really interrogate people for
specific information (unless with a hybrid system). The different
layers of responses would be hard to code. There would be a whole new
lexicon of verbs to learn that might not at all be standardized from
game to game. However, if a list of new verbs seems to long, how is it
quantitatively AND qualitatively different from, say, the spellcasting
system in Enchanter?
Also, I'm not sure if this is exactly what you were talking about
Daryl, but this is what struck me. And keep in mind this is coming
from a person whose only released work of IF had precisely zero NPCs,
notwithstanding stuff I'm working on now. :) But I wanted to throw
this all into the mix and see what people thought of it, especially
people who might have tried their hand at it.
Alan
ems...@mindspring.com
> > What I'd like (though I don't know how to do it) is for
> > the player to be able to construct speech acts from primitives
> > in the same way that physical actions can be constructed out
> > of primitive actions such as "get object", "put object in container",
> > "give container to NPC".
>
> How much work has been done with this?
Not much, to the best of my knowledge. But I'm not sure how much this
idea corresponds to the set of verbs you outline next.
If I understand the 'primitives' notion, what it really requires is a
simulationist system of NPC behavior/reaction: an underlying model of
NPC emotions and acts that will respond in a predictable, yet
interestingly complex, way to a variety of possible inputs. And it
seems to ask for more degrees of complexity than you get from just a
single verb: >INSULT NPC, >COMPLIMENT NPC, etc., give you a fairly
small option set to work with, compared to the kind of freedom and
complexity made possible by the object containment system.
Here's my basic rule of thumb: the more trivial something is
(story-wise) the easier it is to build a simulationist model around
without breaking the game. What I mean by this is that the game keeps
track of a world state and generates the descriptive prose itself, and
allows for a wide range of player tinkering.
This in itself is not necessarily the best approach to NPC building.
At best, it seems, what you would get is a purely emotive NPC, who
shades through various states (angry, lustful, curious, etc) according
to your input and the order in which such things were received.
> I think this is a really
> interesting avenue of NPC interaction. Emily Short touched on this
> briefly in her essay on NPCs
> (http://emshort.home.mindspring.com/NPC3.htm), and that (and Daryl's
> post) got me thinking. If there's a list of 5-12 verbs for
> conversation (from the more standard GREET to FLATTER, TAUNT, BADGER,
> whatever), then there can be various shades of meaning, depending on
> the response of the NPC. Suddenly the different between these speech
> acts and HUG, KISS, SHOW, and other body language clues don't seem
> that far apart--which might be a good way to go, if you're of the
> opinion that communication is more than dialogue.
I am, of course; and I'd like to find a way to offer the player the
freedom to act in those ways when it seems appropriate. I think a
similar intention underlies the tone system in Varicella, since saying
something in a hostile tone usually involved a gratuitous insult or
two, for instance.
> 1. The verbs are active and kinetic, and perhaps vivid words in their
> own right. It ties conversation into one of the key strengths of IF in
> the first place--a direct succinct ability to impact a virtual
> environment (counting the NPC as part of that environment, here).
This is a good point. One of the annoying things about menus is that
it reminds the player that he is at one remove from the action;
likewise disambiguation.
> Disadvantages are obvious--you can't really interrogate people for
> specific information (unless with a hybrid system). The different
> layers of responses would be hard to code. There would be a whole new
> lexicon of verbs to learn that might not at all be standardized from
> game to game.
And:
> However, if a list of new verbs seems to long, how is it
> quantitatively AND qualitatively different from, say, the spellcasting
> system in Enchanter?
These two points were killer the last time I tried something like
this. It was too hard to remember what all the options were, too
difficult to guess what results there would be. My system included
some verbs like >SMILE and >LAUGH, where it was easy enough to
describe the PC as behaving that way, but very hard without
case-by-case coding to predict how the NPC would respond. At the time
I hadn't come up with good general ways of representing various
information about things like NPC mood and expectations, and so there
was no level of simulation, just hand-written everything. Which meant
that 99 times out of a 100 SMILE didn't do anything remotely
interesting, and then once in a 100 it produced some vitally important
plot event.
My sister adjudged this a dead loss, reasonably enough; I abandoned it
in favor of refining the topic and speech representation.
So my conclusions are more or less as follows:
Verbs like this basically ask for a simulationist approach to NPC
behavior. So far we don't have one, really; what we have, for the
most part, is varyingly complex structures of hand-coding. It's true
that something like Galatea or Best of Three does a certain amount of
statistical mood-tracking to determine whether the NPC is angry,
interested in you, etc., but that information is never used *raw* to
determine the results of a player action. Mood affects the
sentence-generators that describe gestures; it makes the NPC laugh,
sometimes, or causes a pause in conversation if things have gone
awkward, but it's never the sole determining factor in what happens
when a player says something.
In order to make these verbs really useful -- consistently
interesting, and thus fun to play, because no one ever likes playing
with a complicated set of verbs that only work once per game, and
Enchanter would be a lot less fun if you couldn't FROTZ inappropriate
objects when bored -- you need a way to represent their behavior that
doesn't require the author to hand-code them in addition to everything
else. (Or you need to make them the *only* thing that the author is
hand-coding -- so, for instance, instead of a TALK TO system that
produces one canned gobbet of text each time the player chooses to use
it, you'd have a choice of BADGER, CAJOLE, or FLATTER, which would
produce one of three canned gobbets.)
NPC as purely simulated emotion box doesn't interest me much. There
was some description a long time ago, and I can't remember in what
context or where I read this, of an NPC named "Kim", who could be
moved through a sort of emotional maze: you could, by the right
combination of verbs in the right order, get her from cautious to
friendly to seduce-able. But at best that seems like a toy, at worst
a combination lock; there's no conversation there, and I like
conversations (not that you could've guessed); and I don't see it as
very useful for building cool narratives.
So then-- how do you get NPCs who are partly simulation and partly
hard-coded? I'm not completely sure, but I think that's where we need
to go anyway, partly because it provides the freedom for verbs like
this and partly because simulation spackles over the spaces that
hand-coding hasn't managed to fill.
One thing that I'm trying to do is make a system that is able to
relate verbs like this to existing conversational statements. If you
think about it, a lot of existing standard verbs already suggest a
certain amount of conversational interaction -- GIVE, SHOW, and BUY
all seem to imply interactions that really ought to affect the course
of a conversation if one is in progress. So for instance in my WIP I
have fixed it so that if you want, you can ask the price of something
by doing >TOPIC FOO and selecting the "So, how much for foo?" menu
option, and then you can buy it by selecting the "Ok, I'll take it."
menu option. Or you can type >BUY FOO, and the appropriate
conversation will be narrated around the action. So the more direct
verb is translated internally into a set of other actions that the
program already knows about as statements.
Likewise, it's possible to encode certain pieces of conversation as
being both 1) things that should appear on the conversation menu in
certain places and 2) valid automatic results to commands like
>THREATEN MAN.
This is all kind of experimental, though. I wrote the BUY routines
last Saturday, and I haven't finished working out a lot of the other
implications of this system.
The more I work on the topic-menu thing, the more I like its freedom,
and the more I recognize what a pernicious thing it is. Typing (1)
just is NOT as gratifying as typing ASK MARIUS ABOUT COSETTE, and
*certainly* not as great as TAUNT MARIUS ABOUT BED-WETTING would be,
if it would only work.
So I'm coming back around to being interested in some of those more
direct conversation-action verbs, this time with (I hope) a somewhat
more mature concept of how to handle them.
ES
Kathleen
> > conversation (from the more standard GREET to FLATTER, TAUNT, BADGER,
> > whatever), then there can be various shades of meaning, depending on
> > the response of the NPC.
I think you would be hard pressed to create a "standard" list with
just 12 items on it, though I would be more than pleased to be proved
wrong.
The other problem that I've personally seen is that some people don't
like not knowing the difference between choices ahead of time. In
_Prized Possession_ for example, at one point TALK TO produces a list
with two opposite responses to choose from (LIE/TELL THE TRUTH). I
received some feedback indicating people were uncomfortable not
knowing up front what the implications of each would be before they
selected.
Finally - BADGER about what? There could be many things that could
refer to - though disambiguation might be useful to narrow it down:
>BADGER MAN
... about hair, about wife, or about dog?
Of course, each of those requires new text to be written by an author.
12 different reply options for 3 different topics is a whopping 36
different text blobs to write, and that's just for a single move. If
each of those 36 had 3 topics with 12 different options you've just
created 1296 text blobs and you're only on move 2.
> > Suddenly the different between these speech
> > acts and HUG, KISS, SHOW, and other body language clues don't seem
> > that far apart--which might be a good way to go, if you're of the
> > opinion that communication is more than dialogue.
<snip>
> > However, if a list of new verbs seems to long, how is it
> > quantitatively AND qualitatively different from, say, the spellcasting
> > system in Enchanter?
>
> These two <kmf:now 1> points were killer the last time I tried something like
> this. It was too hard to remember what all the options were, too
> difficult to guess what results there would be. My system included
> some verbs like >SMILE and >LAUGH, where it was easy enough to
> describe the PC as behaving that way, but very hard without
> case-by-case coding to predict how the NPC would respond.
I also think it is a problem of scaling. Most objects in IF have
relatively simple implied usage, even to the uninitiated. A button
begs to be pushed, a door/box to be opened, a coat to be worn, a rock
to be thrown; but unless your NPC is a glorified prop (ie. the doorman
in _Masquerade_), the number of possible verbs for even moderately
implemented NPC's is quite expansive. It would be like having a button
where PRESSED, PUSHED, SHOVED, TOUCHED or POUNDED each produced
slightly (or wildly) different effects (and that's just 5, imagine if
there were 12!). While it might be fun (and completely appropriate)
for an object art show entry (or an NPC entry like _Galatea_), I think
it would quickly annoy players of an Annual Comp entry.
Kathleen - and no, I don't have any better idea. Sorry. Wish I did.