While nn is flexible enough to allow either side to influence the other, it doesn’t provide any sort of pre-made solution for how this should happen. Characters can interact with the underlying world and with the narration and drive it in a bottom-up way, or a drama manager could take care of everything, or some other scheme could be used. In any case, a system would have to be implemented to make this happen, so aditional work will have to be done. My hope is that for many types of systems, splitting the narration and the simulation will be very useful to creating new narrative effects.

My point exactly. :)

hm.

Nick- As for what I am talking about with NPC behavior, I guess I assume that
your god-narrator has some idea how to pull the marionette strings all over the
game. That includes what transitional (traditionally narrated) text the player
gets exposed to, and it also includes whatever plot-forcing-forward text the
player gets exposed to by way of interactions with NPCs. Sorry my original question
sounded vague. So what I am asking is whether nn specifically or non-specifically
attempts to address this.

To me, what’s at stake with narration-automation is that at some level there’s
the author, who either explicitly or implicitly holds the narrative arc in
their hands… and so with a tool/engine for automation of some of that, your
god-narrator can presumably do a lot of fancy things at a broader level. I was
asking whether nn specifically anticipates the need to treat NPC-related narrative
bits any differently than anything else in the game.

So your comment that your work focuses on narrators and not characters sort of
tells me that philosophically, you’d rather let NPCs be automatons of another sort
rather than ones that get to interact with the plot/narrative engine. Ok. That’s a
really big philosophical choice, to me, at least. Is that where you stand on this?
To me it is a little too limiting to say that the voice of the top-narrator only
gets voiced as a voiceover, that’s all.

From what I understood from this presentation, Mr. Montfort claims in his doctorate thesis to have a system for producing variation of narrative of several kinds, including by order and focalization:
“nn can produce narrative variation in all of these categories”

I’ll take it such system does exist and is verifiable for him to get his doctorate. I have no reason to think otherwise for now, specially from such a celebrated author.

Such a system is good for the advancement of IF as both a narrative and more interactive medium. And, AFAIK, there’s nothing like that out there, or at least that doesn’t involve writting some several thousand lines of custom hackish efforts by the author.

He’s talking about dynamic variation of narrative and you reply with report grouping. These are clearly not the same thing.

I am eager to see it in action and to read more of the theory behind.

> are you saying that you can’t find a need for events to be told non-chronologically[?]

No, quite the contrary. Obviously it’s sometimes good to tell events non-chronologically. (The obviousness of the value of a-/multi-chronology threatens the pretended usefulness of this claim, when the claim is made in a doctoral work. Simplifying Genette’s very sophisticated treatment of the subject — this does not make the point less inane.)

So, yes of course it’s sometimes good to tell events non-chronologically. Does this imply that chronology should be handled by an automatic narrator? No. Is it possible that automatic handling of mixed-chronology is a good idea? Yes, it’s possible. But Nick’s argument falls well short of this question.

The overriding reason for non-chronology is the same as the reason for multi-threaded story; multi-chronological and multi-threaded are structurally the same on a high level. Nick hasn’t recognized this; he hasn’t thought this through.

This thesis smacks of something thrown together at the last minute. — Which really really surprises me. I expected something much more.

noah, you’ve shown on the Indigo Prophesy thread that you’re capable of having a discussion wherein you don’t shriek “troll.”

… but that would be an equivalently trollish approach, and I think it’s just a matter of us not being properly acquainted.

Anyway, yes I’m sure Nick is the Golden Child and everything… but I did make the list of questions pretty clear. If greater clarity serves the cause:

1) misappropriation (or, charitably, naive simplification) of the principle theoretical source

2) the problem that report grouping (by report-class, and by detail) is nothing new

3) theorization of chronological rearrangement implies that it’s necessary to automate chronological narrative, but Nick hasn’t identified a need for an automated narrator to rearrange event-report, but only a desire for events to be told non-chronologically

and of course 4) the existence and interest of ‘nn’.

But I will offer this one thought. If you have a friend you trust, who you think communicates well, maybe you could show them what you’ve been posting on this thread and ask for their opinion as to whether it’s appropriate?

I’m not going to purport myself to be well-versed in narrative theory, but are you saying that you can’t find a need for events to be told non-chronologically as well, or just that Nick hasn’t made one? I can think of plenty of ways this can be used in video game AI, such as the one I mentioned almost at the top of this thread.

> I posted my slides to inform people about my work and to allow discussion, so I’m glad to answer any serious questions
> about them.

Your post is not only slides; it’s a representation of your argument. And either you represented your dissertation very badly (i.e., it’s good but you make it sound dumb), or you represent it reasonably well (and it is in fact impressive).

You say you’re interested “to answer any serious questions.” Does this include the problem of your misappropriation (or, charitably, naive simplification) of your principle (sole?) theoretical source; the problem that report grouping (by report-class, and by detail) is nothing new; that your theorization of chronological rearrangement implies that it’s necessary to automate chronological narrative, but you haven’t identified a need for an automated narrator to rearrange event-report, but only a desire for events to be told non-chronologically?

But you know as well as anyone what the serious questions/problems are. Hell, you or I or Short or Eve or Roberts or Nelson or Plotkin or Nyman or Jerz or Thornton (or a lot of other people) could have put this or a better thesis together in a couple solid weeks’ work. There’s no work here to justify a doctorate, or if there is, you really haven’t let on. You must know that you’re considered the guy who believes that we’re making progress when we translate well-known ideas into Greek. The last time your work was discussed on RAIF, it was generally concluded that your hypotheses are untestable and your theories ungrounded.

If you don’t recognize a problem, there’s no hope. You’ll applaud yourself when arising from the toilet. “Thanks for the congratulations!” indeed.

David, I posted my slides to inform people about my work and to allow discussion, so I’m glad to answer any serious questions about them. I’m not sure what you mean by “NPC handling,” but my work on this project focuses on the level of narrating, on narrators rather than characters. I haven’t really tried to improve what characters do or what things that happen in the world, just how those things are told. There is certainly a connection between what you need in order to narrate from a certain perspective and what you need to determine how a character should act. I discuss this some in the document, but didn’t cover it in any detail in the presentation.

My full dissertation will be available for free from UMI as soon as they process and post it; when I get the chance, I’ll also post a PDF myself.

The main point was simply that this work really doesn’t show a lot of promise. The course of the argument strongly suggests that the underlying project is badly thought out, which further suggests that it’s not terribly interesting on a coding level either. Altogether, that’s pretty depressing.

I should probably apologize for writing as though I’m scandalized by all this. But I really am mad about it. I guess I don’t know why I should be. Some ethic thing I guess, about how a person (especially one who has the power and indeed the faculties to change things) is supposed to do a good job and not be a nincompoop. So no I won’t apologize for scolding; it’s an honest reaction and I do think it’s the right thing to do.

I will, however, apologize in the formal sense: I apologize if I offended anybody; I do not mean to be offensive. etc.

It seems obvious enough that there is a ripe area in revising how IF handles narration. It’s extremely
important that NPC handling be tied intimately to whatever narrative scheme emerges. I’m missing whatever
nuance might exist in the dissertation itself when it comes to your take on this. It’s so key, so I hope
you’ll comment. I think it’s just poor if you don’t at least publicly make some attempt to rebut breslin,
regardless of how challenging that might be. So please fill us in a little more, and if possible, post a
link the pdf of the final dissertation so that there can be a better discussion.

Thanks for what you are trying to do.

This is not a well-earned doctorate. I would have rejected this defense. I would have rejected it with the enthusiasm that would rightly accompany an acceptance of Mateas’ thesis. The extreme to which Mateas earned his Ph.D. is the degree to which Montfort did not earn his degree. It was given, that’s all.

First of all, there’s a difference between a powerpoint presentation and a thesis. This is a powerpoint presentation. As we all know, the major work in a thesis of this kind is not in the presentation itself but the work which the presentation describes. If Mateas’ thesis is 200 pages, his efforts documented therein easily amount to ten times that. In other words, a thesis of this sort is supposed to be a critical summary of a great work. Not a fluff-job on a minor work.

== the intro ==

You would not be well advised to spend 1/4 of your time re-explaining the pedestrian conception of signifier and signified.

Cartoons to make the point. An intellectual analysis of “Heathcliff” would be more rewarding than a pseudo-intellectual analysis of a quasi-intellectual cartoon. This is all very transparent.

You really don’t understand Genette *at all*. Soundbyting a philosopher is like tasting uncooked ingredients, before the meal is prepared by a great chef. Genette’s work is proto-deconstructionist, in the sense that it proves the difference between visiting a moment in the argument and asserting a proposition. If you *must* reduce Genette — and it is criminal to do so — you would be better with the formula: x=x; y=y; x~=y; y=x; x!=y; y!~=x; therefore think.

No, Genette’s work is not this schematic-clownlike presentation you find convenient for the sake of your argument. (You should have studied him in one of the philosophy classes they have at Penn. It’s actually one of the best departments in the country for European philosophy.) Many good dissertations have been written on this. I suggest you read one or two. I think Genette is only present in this work because — if you can make it sound like he said it, it won’t sound so inane when you say it.

==

I can’t believe you dedicated a section to your “hopes and dreams” for the genre. “Oh gosh I imagine someday, for IF.” — Are you really this gormless?

==

Ok, now *more than halfway through*, FINALLY finished with the fluff, and on to the actual ideas.

Your description of modern IF architecture is a disgraceful joke. Really: there’s a parser and a world model (within which a game file)?! I just cannot *believe* how dumb a view you’re happily adopting. That’s like saying “ok, take a Playstation, there’s a controller and a TV. Oh, and the console is attached.”

The one idea that’s worthwhile here is also entirely derivative. Emily Short’s already done it. I’ve already done it. Anybody working in the field of automation-report has already done it. Even the main Tads-3 library already does it to a degree. That one idea is report grouping. BTDT. (Been There, Done That)

A doctoral thesis is supposed to contain *new* ideas. So let’s press on, hoping to find a one.

==

two-thirds done, and a conceivably-useful idea. Different ways of ordering a story, both chronologically and conceptually.

A very sketchy claim: “The Narrator has to rearrange events when they are to be told in a non-chronological order.”

You haven’t identified a need for a narrator to rearrange events, but only a desire for events to be told non-chronologically. This is abysmally sloppy thinking.

Automatically re-ordering narrative events under one of a few structures — this is a conceptually-simple algorithm, right?

There’s a bit of meat here, but scant little.

==

varying level of detail in report. I was writing about this like 4 years ago. Nice job, ace.

==

If this were the doctoral thesis of a nobody-dumbass, I wouldn’t mind. But this is billed as cutting-edge theory in IF. Which is just awful.

C-

(due to grade inflation)

The narrator needs to (a) known what the player knows, and (b) know what the player needs to know (and when).

A simple implimentation already exists in IF where a long description is provided for an object the first time it’s seen, and then only a short descriptions later times.

The concept is more generalizable. For example: The 1st year of the TV show, Heroes, is about stopping a huge explosion from happening in NYC. The viewers need to know this fact fairly early on in the narration since it glues the whole series together.

In terms of IF, the narrator would know that initially the player doesn’t know about impending explosion. If they don’t happen to talk to the fortune teller right away and hear about the impending doom, then maybe another NPC would mention it as part of a conversation (hearing it from the fortune teller). If the player had talked to the fortuneteller, then for the sake of brevity, the second NPC would never bother to mention the info.

Of course, this goes a bit beyond what your definition of narrator seems to be… more of a director role since it affects the world model.

This can be extended further into more director-like behavior, ensuring that certain activities happen in a certain order… but that’s a different discussion not entirely related to a narration module.

http://www.CircumReality.com

I like your use of Matt Madden’s comics.

PS. Can we play the game?

(Courtesy of Ian Bogost at Water Cooler Games) Nick Montfort has defended his dissertation on Interactive Fiction. It is well worth reading for those interested in interactive fiction as well as game design.