One-on-One IF
fiziwig
and some notes. He tells you the beginning of a story and then invites
you to ask questions and participate in the unfolding of the plot.
"What if I talked to the dragon?" you might ask you.
"Well, the dragon is not in a very talkative mood right now," he
replies.
"Suppose there there a way to convince the dragon to talk," you wonder
aloud. "I wonder how I might go about that."
"I think you might be barking up the wrong tree," he says. "There's
something much more important in this room."
And so it goes. Interactive, person-to-person IF; where statements made
are actually understood and discussed, and the answers are helpful and
encouraging to the player.
THAT is how IF should work; as if you were chatting with the author,
and as if the author really cared that you were interested in playing
the game, and made an effort to encourage you can keep you interested
and on track.
In other words, the IF engine ought to have enough intelligence and
natural language ability to hold a sustained and meaningful
conversation with the player. Of course that's a pipe dream right now,
but maybe some day...
--gary
Stephen Bond
> Imagine this scenario. You sit down with a friend. In his lap, a map
> and some notes. He tells you the beginning of a story and then invites
> you to ask questions and participate in the unfolding of the plot.
>
> "What if I talked to the dragon?" you might ask you.
> "Well, the dragon is not in a very talkative mood right now," he
> replies.
> "Suppose there there a way to convince the dragon to talk," you wonder
> aloud. "I wonder how I might go about that."
> "I think you might be barking up the wrong tree," he says. "There's
> something much more important in this room."
>
> And so it goes. Interactive, person-to-person IF; where statements made
> are actually understood and discussed, and the answers are helpful and
> encouraging to the player.
>
> THAT is how IF should work; as if you were chatting with the author,
> and as if the author really cared that you were interested in playing
> the game, and made an effort to encourage you can keep you interested
> and on track.
This is one model of how IF could work, and not one appropriate
for every game.
I think you're bundling in a few concepts together here. The
first is the intelligent parser, the parser that understands every
natural language command you could type, which is desirable,
but, as you say, a pipe dream. But an intelligent parser is not
necessarily the chatty parser you've described, and a chatty
parser not necessarily the helpful best friend parser you've
described either. Chatty and helpful parsers have already been
implemented in various games.
Different games assign the parser a different role -- parser as
character (PC or NPC), parser as authorial voice, parser as
best friend, parser as transparent interface -- or sometimes
a combination of these roles (e.g. "interface" messages are
surrounded by square brackets). In some games, you want the
parser to be an interface and not an interlocutor.
The helpful parser model might be suitable in some game
aimed at beginners, but in general there are better ways to
keep the player interested than direct prompts -- such as
good writing, a good story, or an interesting puzzle or game
concept. Indeed, if you need to keep prodding the player in
the right direction to keep him interested, then I suspect
there's something else wrong with your game. Often, such
hand-holding (found in a lot of 3D games today) makes me
LESS interested, completing the game becoming a matter of
following all the prompts and going through the motions.
Stephen.
fiziwig
percentage of my attempts at communication with the game come back with
"I don't have a clue what you're talking about" stated in one form or
another. I keep wanting to like IF and I keep trying to like IF, but it
just isn't working out for me. :-(
--gary
Greg Boettcher
Which games are you playing? That clearly makes a huge difference. If
you just choose an IF game more or less at random, the odds are that it
will suck. I'm stating the obvious here, but it was a relevant point in
that "why newbies don't like IF" thread.
Greg
Emily Short
If you don't feel uncomfortable sharing, your frustrating transcripts
might be a useful source of information for people who are trying to
improve IF parsing and feedback.
The Computer Dood
Eric H. Wrote
I always thought that was what the BETA TEST process is all about. It
is always good to have someone else read through your program, and to
see wheither or not your thoughts match the typical users thoughts. I
am writing an ADRIFT adventure. I have a small panel rusted shut. I
had to put all the verbs PULL, PUSH, PRY, HIT, etc. as described to
try to get the panel open. I could have used the *Panel command, that
any verb would open the panel, but what if some were to talk to the the
panel. I would not want the panel to open up. So there are varying
degrees that I want the program to assume what I meant.
I too have felt frustration on some games. The verb Look after a
description. The program should be smart enough if they just said
"There is a small object here". I should not have to type "examine
object". I should be able to type "examine" and it should know that I
meant the last thing described. I guess that would require a more
robust parser, or it puts the burden on the story writer to properly
describe what they meant.
Emily Short
The Computer Dood wrote:
> Emily Short wrote:
> > fiziwig wrote:
> > > Very good points! I guess my biggest frustration is having such a large
> > > percentage of my attempts at communication with the game come back with
> > > "I don't have a clue what you're talking about" stated in one form or
> > > another. I keep wanting to like IF and I keep trying to like IF, but it
> > > just isn't working out for me. :-(
> > >
> > > --gary
> >
> > If you don't feel uncomfortable sharing, your frustrating transcripts
> > might be a useful source of information for people who are trying to
> > improve IF parsing and feedback.
>
> Eric H. Wrote
> I always thought that was what the BETA TEST process is all about.
Smoothing out parsing is one of many things a good beta test can do --
but beta testers are often experienced with IF, and therefore less
likely to get hung up on things that cause trouble to novice players.
I'm not knocking this, since it is important to have at least a couple
of testers who are familiar with the genre and can give feedback on
things like puzzle design and game pacing based on some real
experience; also because often the only testers one has available are
others from the community. The ideal tester pool would probably include
some experienced testers and some novice players, but you can only rope
your friends and family in so many times before they stop counting as
novices.
Brian Wh.
back with "I don't have a clue what you're talking about"
Quite right, fiziwig. I have the same feeling. It doesn't matter which
game, if it's parser-based, it shows the same damned behaviour. More
than a few people here on rgif actually believe that this is how IF
_should_ behave, that IF that doesn't go "Huh?" at you it's not really
IF.
This problem has been talked to death here (see IFwiki -- past raif
topics). It pains me to see fine authors like Zarf and Emily Short
crippling their work by clinging to a parser paradigm that was tried,
found wanting and dumped by the rest of the world two decades ago.
Zarf, Emily and you others, I love your work, hate your parser.
As an example, look at Dreamhold, an IF intended for beginners. Not
even an exhaustive list of commands. _WHY?!?_ As far as I played it,
nothing would have been given away by simply giving a complete command
list. (By the way, I know many authors work from a map. Would it kill
you to include it with your IF? INFOCOM [pauses to genuflect] included
maps in their IFs even though they cost money to print. Modern IF
authors can just include a .GIF or .JPG with their distribution, at no
cost to themselves. Maps, people, if your IF is map-oriented!)
Let's face facts, the "parser as NLP" paradigm doesn't work. It's what
killed IF in the INFOCOM [pauses to make sign of the cross] era. That
and puzzles.
The parser doesn't work the way you wish it would. It's not NLP, not
even close.
Be honest and straightforward with your players. If you make your
command set known, your players will thank you for it. Keep your
command line interface, but open the kimono and show us your command
set. You'll be truly free! You know you want it! Hide in the shadows no
longer! Throw off the shackles of the oppressor and abandon the parser!
Embrace the command set! "We are the future, Charles, not them! They no
longer matter."
Samwyse
> "a large percentage of my attempts at communication with the game come
> back with "I don't have a clue what you're talking about"
>
> Quite right, fiziwig. I have the same feeling. It doesn't matter which
> game, if it's parser-based, it shows the same damned behaviour. More
> than a few people here on rgif actually believe that this is how IF
> _should_ behave, that IF that doesn't go "Huh?" at you it's not really
> IF.
To echo Emily's reply to fiziwig, please send us a transcript of this
behavior. There are a lot of people who want the parser to behave
better. Not full NLP, but able to recognize when it's over it's head
and provide meaningful assistance to the player.
> As an example, look at Dreamhold, an IF intended for beginners. Not
> even an exhaustive list of commands. _WHY?!?_ As far as I played it,
> nothing would have been given away by simply giving a complete command
> list. (By the way, I know many authors work from a map. Would it kill
> you to include it with your IF? INFOCOM [pauses to genuflect] included
> maps in their IFs even though they cost money to print. Modern IF
> authors can just include a .GIF or .JPG with their distribution, at no
> cost to themselves. Maps, people, if your IF is map-oriented!)
Which games would those be? I only recall one of the first twelve games
(specifically the folio editions) having a map (Suspended). I won't
claim to own all of the Grey Box releases, but of the ones I do own, I
think that only LGoP had a map, and that was only of the "maze".
> Let's face facts, the "parser as NLP" paradigm doesn't work. It's what
> killed IF in the INFOCOM [pauses to make sign of the cross] era. That
> and puzzles.
What killed Infocom was their ill-fated expansion into business
software. What wounded IF was the graphic adventure.
Autymn D. C.
> better. Not full NLP, but able to recognize when it's over it's head
its head, illiterate retard!
Autymn D. C.
> I too have felt frustration on some games. The verb Look after a
> description. The program should be smart enough if they just said
> "There is a small object here". I should not have to type "examine
> object". I should be able to type "examine" and it should know that I
> meant the last thing described. I guess that would require a more
What's wrong with "x it" or "l at it"?
Andrew Plotkin
>
> As an example, look at Dreamhold, an IF intended for beginners. Not
> even an exhaustive list of commands.
Sure it has one. Not "exhaustive" as in "covering every verb you will
see in any IF game ever", but it covers what you need to reach the end
of the game.
It doesn't shove them at the player right on the first move. That
would be... exhaustive. I mean, *exhausting*. Shoving a six-page
manual at the user is a great way to render him completely
unconscious.
If you type "help" right off, the parser starts by introducing the
most common commands (L, I, X, moving) and then points you at the help
menu. This has a bunch of topics, including "help commands".
Or, if you *don't* start by typing "help", the game leads you through
the same basic commands one at a time -- in the course of actually
playing the first few rooms. I suspect that's what people actually do.
See above, re manuals.
--Z
--
"And Aholibamah bare Jeush, and Jaalam, and Korah: these were the borogoves..."
*
"Bush has kept America safe from terrorism since 9/11." Too bad his
job was to keep America safe *on* 9/11.
Andrew Plotkin
> Here, Brian Wh. <bri...@sympatico.ca> wrote:
> >
> > As an example, look at Dreamhold, an IF intended for beginners. Not
> > even an exhaustive list of commands.
>
> Sure it has one. Not "exhaustive" as in "covering every verb you will
> see in any IF game ever", but it covers what you need to reach the end
> of the game.
Footnote: I don't recall offhand whether it's enough to reach every
optional discovery.
In any case, I *didn't* present the verb list as "this is the list of
verbs you need to win." It being a newbie game, I figured the illusion
of a broad command set was as good as actually having one. :)
fel...@yahoo.com
> What's wrong with "x it" or "l at it"?
The same thing. The pronouns aren't set automatically to the last
object described. The author can do it manually in some
circumstances, but it's a burden (another one?!) so it's not often
done. It stumps me all too often as a player.
Felix
P.S. I think I have a technical solution for this one: devise some kind
of semantic markup for authors to embed in descriptions, indicating
which words represent an object in the room. TADS3 already uses a
similar method for changing conversation states.
Brian Wh.
"Which games would those be?[that have maps] I only recall one of the
first twelve games
(specifically the folio editions) having a map (Suspended). I won't
claim to own all of the Grey Box releases, but of the ones I do own, I
think that only LGoP had a map, and that was only of the "maze". "
These INFOCOM [touches forelock in respect] have at least partial maps
in the Lost Treasures distribution:
Ballyhoo
Deadline
Enchanter
Lurking Horror
Moonmist
Infidel
Planetfall
Spellbreaker
Starcross
Stationfall
Trinity
Sorcerer
Beyond Zork
Zork I
Not an exhaustive list <g>.
Andrew Plotkin
>
> "Which games would those be?[that have maps] I only recall one of the
> first twelve games
> (specifically the folio editions) having a map (Suspended). I won't
> claim to own all of the Grey Box releases, but of the ones I do own, I
> think that only LGoP had a map, and that was only of the "maze". "
>
> These INFOCOM [touches forelock in respect] have at least partial maps
> in the Lost Treasures distribution:
Those were spoiler maps, not feelie maps. In effect, they were part of
the game *hint* package, not the game *documentation*.
If you want to argue that all games should be distributed with full
hints, that's a different argument.
--Z
--
"And Aholibamah bare Jeush, and Jaalam, and Korah: these were the borogoves..."
*
If the Bush administration hasn't shipped you to Syria for interrogation, it's
for one reason: they don't feel like it. Not because of the Eighth Amendment.
The Wanderer
> Here, Andrew Plotkin <erky...@eblong.com> wrote:
>
>> Here, Brian Wh. <bri...@sympatico.ca> wrote:
>>
>>> As an example, look at Dreamhold, an IF intended for beginners.
>>> Not even an exhaustive list of commands.
>>
>> Sure it has one. Not "exhaustive" as in "covering every verb you
>> will see in any IF game ever", but it covers what you need to reach
>> the end of the game.
>
> Footnote: I don't recall offhand whether it's enough to reach every
> optional discovery.
I'll just note that, although I don't particularly agree with your
antagonist's argument, I would not consider even that to necessarily
qualify as an "exhaustive" list.
An "exhaustive" list of verbs in IF, to me, means "a list of every verb
recognized by the game" - no exceptions. Some games have, I believe,
provided this, albeit AFAIR no non-commercial ones; I still find myself
typing VERBS early on in most IF I play, just in case there happens to
be such a list there. IIRC, some red-herring verbs were included -
"recognized" in the sense that they would produce some non-generic
output in some context, or perhaps even some appropriately generic
output in every context, but never actually useful at any point.
Unfortunately I don't remember what games I've played which *did*
implement the VERBS verb...
(As far as games which would be ruined by the availability of a verb
list, how about Ad Verbum? Around two-thirds of the puzzles in that game
are guess-the-verb, obviously and by design, and I for one found it
quite enjoyable... though I'm still annoyed at having No Clue where
there might be left to look for the last lousy point.)
--
The Wanderer
Warning: Simply because I argue an issue does not mean I agree with any
side of it.
Secrecy is the beginning of tyranny.
Andrew Plotkin
> Andrew Plotkin wrote:
>
> >> [Dreamhold]
> >> Sure it has one. Not "exhaustive" as in "covering every verb you
> >> will see in any IF game ever", but it covers what you need to reach
> >> the end of the game.
> >
> > Footnote: I don't recall offhand whether it's enough to reach every
> > optional discovery.
>
> I'll just note that, although I don't particularly agree with your
> antagonist's argument, I would not consider even that to necessarily
> qualify as an "exhaustive" list.
>
> An "exhaustive" list of verbs in IF, to me, means "a list of every verb
> recognized by the game" - no exceptions.
That's a straightforward list to compile. But wouldn't that tend to
*both* spoil the game *and* mislead the player? If I document that
Dreamhold supports "HANG ON TO XX" and "HANG XX ABOVE YY", is the
player really going to become more comfortable in his approach to the
game? Or is he just going to assume that the game is obsessed with
slang and pendulums?
In fact the former is an obscure synonym for "HOLD XX", and the
latter was thrown in to handle "HANG PICTURE ABOVE FIREPLACE", as
there happens to be a hook there. Both exist because of feedback from
betatesters; neither is necessary to win the game.
This, I note, is my indication that the truly exhaustive verb list is
an actively bad idea. The beta-testing process always gives you a list
of strange synonyms for existing actions -- command phrasings that you
never thought of. You add (some of) them to the game because they
will improve responsivity for the few players that come up with them,
and they don't hurt the game for anyone else.
However, does this process really imply that *everyone* should be
seeing these obscure synonyms, as part of the game documentation? It
seems to be that that would more than negate the point; you'd be
confusing everybody. And the more synonyms you add, the more complex
and intimidating the exhaustive command list would get. That can't
possibly be right.
> (As far as games which would be ruined by the availability of a verb
> list, how about Ad Verbum? Around two-thirds of the puzzles in that game
> are guess-the-verb, obviously and by design, and I for one found it
> quite enjoyable... though I'm still annoyed at having No Clue where
> there might be left to look for the last lousy point.)
For a more traditional example: Enchanter. Much of the fun of this
game is finding new spells; each spell is a new verb.
--Z
--
"And Aholibamah bare Jeush, and Jaalam, and Korah: these were the borogoves..."
*
If the Bush administration hasn't subjected you to searches without a warrant,
it's for one reason: they don't feel like it. Not because you're innocent.
Rubes
> What killed Infocom was their ill-fated expansion into business
> software. What wounded IF was the graphic adventure.
Could you elaborate? I don't necessarily disagree; I'm just interested
in the argument.
The Wanderer
> Here, The Wanderer <inverse...@comcast.net> wrote:
>
>>Andrew Plotkin wrote:
>>> Footnote: I don't recall offhand whether it's enough to reach
>>> every optional discovery.
>>
>> I'll just note that, although I don't particularly agree with your
>> antagonist's argument, I would not consider even that to
>> necessarily qualify as an "exhaustive" list.
>>
>> An "exhaustive" list of verbs in IF, to me, means "a list of every
>> verb recognized by the game" - no exceptions.
>
> That's a straightforward list to compile. But wouldn't that tend to
> *both* spoil the game *and* mislead the player? If I document that
> Dreamhold supports "HANG ON TO XX" and "HANG XX ABOVE YY", is the
> player really going to become more comfortable in his approach to the
> game? Or is he just going to assume that the game is obsessed with
> slang and pendulums?
Actually, this would not be necessary - because the verbs list, by my
conception of such a thing, would simply include the verb HANG.
Individual usages of listed verbs need not be given (and indeed, for the
form of the list I remember, would be difficult to display without
introducing confusion).
(I've spent a little time thinking the matter over, and I haven't been
able to come up with any examples of two-word "verbs" which I would say
should be included in such a list. If anyone cares to suggest
possibilities, feel free to do so.)
> In fact the former is an obscure synonym for "HOLD XX", and the
> latter was thrown in to handle "HANG PICTURE ABOVE FIREPLACE", as
> there happens to be a hook there. Both exist because of feedback from
> betatesters; neither is necessary to win the game.
>
> This, I note, is my indication that the truly exhaustive verb list is
> an actively bad idea. The beta-testing process always gives you a
> list of strange synonyms for existing actions -- command phrasings
> that you never thought of. You add (some of) them to the game because
> they will improve responsivity for the few players that come up with
> them, and they don't hurt the game for anyone else.
>
> However, does this process really imply that *everyone* should be
> seeing these obscure synonyms, as part of the game documentation? It
> seems to be that that would more than negate the point; you'd be
> confusing everybody. And the more synonyms you add, the more complex
> and intimidating the exhaustive command list would get. That can't
> possibly be right.
It would be possible to argue for eliminating from the list verbs which
exist only as true synonyms of other verbs - although it would, I think,
also be possible to argue against it.
It would really help my position here, I think, if I could *remember*
one of the games which implemented the kind of list I'm thinking of, and
give its verb list here... the only thing which is coming to mind as a
possibility is Leather Goddesses of Phobos, and I don't have a copy of
that in an accessible state (we own it on Apple //c floppy disk, and our
Apple //cs are rather packed away at the moment).
>> (As far as games which would be ruined by the availability of a
>> verb list, how about Ad Verbum? Around two-thirds of the puzzles in
>> that game are guess-the-verb, obviously and by design, and I for
>> one found it quite enjoyable... though I'm still annoyed at having
>> No Clue where there might be left to look for the last lousy
>> point.)
>
> For a more traditional example: Enchanter. Much of the fun of this
> game is finding new spells; each spell is a new verb.
Very good point. One could theoretically argue for not listing those
verbs separately, on the grounds that each is actually a shorthand for
the form CAST <SPELL NAME>, but that would not be compatible with my
initial statement. (I do not rule out that my initial "no exceptions"
may have been too strong. I try to speak precisely, but I do still flub
things from time to time.)
(Oh! I just remembered something obvious. Yes, eliminating such verbs
from the list would make it no longer a truly exhaustive list - but all
that means is that a *truly* exhaustive list, as opposed to one of the
more restricted forms (such as the "direct synonyms need not apply" form
mentioned above), probably is in fact a bad idea.)
I'm picturing in my head a syntax such as VERBS (prints all non-hidden,
non-synonym-only verbs in the game), VERBS KNOWN (prints all known verbs
- i.e., the list in Enchanter would grow to include each new spell as
you come across it), VERBS ALL (asks for confirmation, similar to
invoking a hint system, then prints all recognized verbs without
exception), and possibly a few others... but that's starting to get a
bit more involved than is probably justified, here.
Sophie Fruehling
> It would really help my position here, I think, if I could *remember*
> one of the games which implemented the kind of list I'm thinking of, and
> give its verb list here...
_The Mulldoon Legacy_ has a verb list. It does understand other verbs,
too, though.
--
Sophie Frühling
The cube tastes like sugar. You are suddenly surrounded by a herd
of moose. They start talking to you about a moose-load of things.
Samwyse
You can read an entire term paper on the subject here:
http://web.mit.edu/6.933/www/Fall2000/infocom/
or skip to the conclusion here:
http://web.mit.edu/6.933/www/Fall2000/infocom/failure.html
Brian Wh.
"If you really wanted them to change, you'd use other tools than
withering sarcasm, so I guess there was no reason for you to post at
all. "
Sorry, Poster, that wasn't sarcasm at all, let alone the withering
variety. I meant it sincerely.
So I guess there was no reason for you to post at all.
-Brian
The Wanderer
> The Wanderer writes:
>
>> It would really help my position here, I think, if I could
>> *remember* one of the games which implemented the kind of list I'm
>> thinking of, and give its verb list here...
>
> _The Mulldoon Legacy_ has a verb list. It does understand other
> verbs, too, though.
Since I've never played that, it can't be one of the ones I'm thinking
of (although it may fit the same form - I haven't looked yet). Thanks
for the suggestion, though.
Michel Nizette
> The pronouns aren't set automatically to the last
> object described. The author can do it manually in some
> circumstances, but it's a burden (another one?!) so it's not often
> done. It stumps me all too often as a player.
Personally, I'm not sure I would like that. As a player, I prefer having
pronoun antecedants set to the last thing *I* referred to. This makes the
handling of pronouns more predictable, and gives me more control over it.
--Michel.
Rubes
Samwyse wrote:
> You can read an entire term paper on the subject here:
> http://web.mit.edu/6.933/www/Fall2000/infocom/
> or skip to the conclusion here:
> http://web.mit.edu/6.933/www/Fall2000/infocom/failure.html
Thanks. Nice read.
Baggins
> It would really help my position here, I think, if I could *remember*
> one of the games which implemented the kind of list I'm thinking of, and
> give its verb list here... the only thing which is coming to mind as a
> possibility is Leather Goddesses of Phobos, and I don't have a copy of
> that in an accessible state (we own it on Apple //c floppy disk, and our
> Apple //cs are rather packed away at the moment).
While not implemented in the game itself, Windham Classics' "The Wizard
of OZ" had a complete list of recognized words at the end of the game
manual. You can check it out at
http://files.the-underdogs.org//games/w/woz/files/woz.txt
As a young player new to text adventures, that list was an absolute life
saver. I never would have completed the game without it and it was quite
challenging enough with it.