June 10, 2005

Bernstein’s Bait Redux

by Andrew Stern · , 11:59 pm

Echoing the debates we had two years ago in both blog and book, Mark Bernstein has recently restated his argument, with the coda, “As far as I’m aware, this argument has been essentially ignored.” Actually, I responded to this argument in my First Person book essay response to his and Diane Greco’s essay (see below).

Before you stop reading, thinking this is just pointless academic jousting, let me say I think Bernstein and Greco’s argument is a good and very useful one, pushing important design issues to the fore, that I rarely see done. And although I’m not wanting to rehash the debate, it would be interesting to hear others’ take on the issue, if anyone has any new thoughts to contribute. (Mark includes a link to a good new essay from WRT on frustration with IF and HTF.)

Here’s an excerpt from my response, written four years ago, taken from the book (that you won’t find in the abbreviated version at EBR):

“Even if we could experience Hamlet on the holodeck, it wouldn’t work. Tragedy requires that the characters be blind…” I agree, it seems likely that certain types of stories such as traditional tragedy may not work as an interactive story, for the reasons Bernstein and Greco describe. Instead authors will need to tell the kinds of stories that do work interactively. Façade is a more open-ended, explorative, psychological situation. Is this drama anymore? We hope to understand this better once we get a chance to play with the finished work.


A bit more:

“Illusions that place the reader on stage necessarily founder when promised freedom of action is contradicted by the limitations of the simulated environment.” This is hard to accept. Is it hopeless to give the player the freedom to say whatever she wishes, to freely express herself? I feel it is too soon to answer this as definitively and devastatingly as Bernstein and Greco do. Certainly this kind of freedom is the dream for many players and authors. Bernstein and Greco seem to think that because we cannot fully deliver on this expectation, that we should discard the approach all together. That’s their choice, and their proposals are excellent alternatives. However this dream should not be abandoned because it is technically and artistically challenging to achieve. Instead let’s try to work within the limitations of the technology, push on them, use them as artistic constraints. The player’s expectations must be set at the appropriate level so she avoids struggling against a “necessarily recalcitrant world-model”. Some of us believe that today’s computational environment can in fact match at least some of our aspirations, and as technology inevitably improves over time, it will be able to match even more.

15 Responses to “Bernstein’s Bait Redux”


  1. Bacon Says:

    Saying that Interactive storytelling is impossible because one cannot envision an interactive recreation of Hamlet is completely pointless. This is encompassed in the ‘low-hanging fruit’ discussions that happened earlier on this site.

    The techniques Shakespeare used to express his themes in that play are what should be studied to advance this burgeoning art form, instead of Hamlet’s particular series of events. An interactive story must be able to coherently express the author’s themes at the same time as allowing player agency, so the ground-up rethinking of dramatic principles that is occuring here (among other places) is the way to go.

    The neccesarily closed system of a computer game is an ideal environment to explore such experiments. If the player has only a limited amount of possible actions and the subjects of these actions are also limited, then the laborious task of encoding the dramatic effects of each of their combinations is just that, laborious, but not infinite or intractable. Especially if classifictions of dramatic phenomena are involved such that a certain amount of abstraction from specific game events can occur.

  2. noah Says:

    It’s certainly not the case that Mark’s argument has been ignored. It’s been aired in several high profile places, considered by many people in the field, and then left aside. Those are the characteristics of an argument that people don’t find interesting, rather than one that is failing to attract initial consideration. I hope we won’t re-blog this every time that Mark does.

  3. Dirk Scheuring Says:

    As somebody who is sure that Bernstein’s argument holds an ocean of water, I have difficulty in believing that people don’t find it interesting. Rather, I think that people find it scary. Because: what if he’s right?

    If he’s right, then Bacon, above, is also right, and the dramatic effects of every combination of actions possible for the player to take has to be encoded “by hand”, using human storytellers. This is not what most people in the AI business think AI should be about. Usually, AI researchers and developers want to create AI-driven characters that can figure out for themselves what the consequences of the player’s actions should be, without having to encode for every case what – from the POV of each of the characters in question – makes sense, and what doesn’t, and how to enable the character(s) to motivate each of those distinctions in communication with the player, so the resulting text can make sense as a whole. In short, they still believe that it is possible to give the machine a limited set of parameters from which an unlimited amount of meaning can be created (or will “emerge”), to satisy an unlimited amount of players and their different styles and approaches to solving what each individual player sees as being the story’s problem(s).

    The alternative to this view of what AI can and should do is to acknowledge that the way in which storytellers have been creating successful characters for ages is the only viable option: the creator has to identify with every action that each character takes, in order to motivate it. Now the significant quality of an interactive story is that the player is a character, too, who can follow up each action of each other character with an inter-action, which is probably a counter-action. But the motivation of the player character is completely unknown to the creators, and if they don’t want to resort to IF’s “This is not a good idea” catch-all, they are looking at a huge amount of work, and, as Bacon points out, the necessity of a “ground-up rethinking of dramatic principles”.

    However, it seems that the psychological strain that will enforce this paradigm shift is not yet large enough, so people still comfort themselves with the illusion that simply sidelining Bernstein’s argument with the claim that it’s “uninteresting” to them will work. He can’t be right, he mustn’t be right, since if he’s right, the bottom of the whole AI science barrel is in severe danger of dropping out.

    Me, I turn to Mahatma Ghandi for a description of the pattern that I believe to be in effect here.

  4. noah Says:

    Dirk, there are other ways for it to be uninteresting. For more of my point of view, please see the long thread from last time (linked in Andrew’s post).

  5. Espen Aarseth Says:

    The argument has *largely* been ignored by IF critics, proponents and practitioners (except, of course, in their practical efforts) a notable exception among the latter is Selmer Bringsjord, who has published extensively on this issue, based on more than a decade of exploratory research.

    Both Selmer and I presented critical papers on this problem at the 1994 conference on computer generated literature in Paris; for my own formulation of the argument, see Cybertext, pp 139-41. For Bringsjord, see e.g. http://www.technologyreview.com/articles/98/03/bringsjord0398.4.asp

  6. Dirk Scheuring Says:

    Thanks for the hint, Noah; I see what you mean now. But I also see why I get more mileage from Bernstein’s post than you: I’m not interested in his agenda as a stakeholder in interactive media, but I am interested in the character he thinks he should be able to play in an interactive Hamlet: a character that thinks that, for Hamlet to solve his problems, “it’s absolutely obvious that he should go back to school, get roaring drunk, get laid, and await his opportunity”. I went and tried to play it that way:

    PALACE (BEDROOM) SCORE: 0%

    I am in my bedroom in the palace. There is a four-poster bed, and not much else. A portrait hangs on the wall.
    An exit leads north.

    >Get laid.

    I'm not sure what you want me to pick up.

    >Pick up a girl.

    I'm not sure what you want me to pick up.

    >Damnit! Go down to the pub, get drunk, grab a broad and get your rocks off! Is that so hard to understand?!

    I can't see a way down from here.

    >You mean you don't know how to get to the pub?

    Sorry, I can't do that.

    >Why not?

    Sorry, I didn't understand that.

    >Aaargh!

    Sorry, I didn't understand that.

    So as far as his argument is that his particular character can’t be part of the story, he’s right – for now. Instead of just ignoring him, giving him all the more reason to repeat his “You can’t do that!” schtick like he’s a piece of IF himself, why not treat his challenge as a viable feature request? Especially since the reason why his “solution” won’t work on Hamlet is already implicit in the play as it is: “getting laid” is exactly what Hamlet sees his mother doing with the murderer of his father, and he detests her for it. Making this motivation, and the values behind it, explicit is making it accessible for the Hamlet character to use in the interaction – he needs to be enabled to repudiate Bernstein’s suggestion himself. I think what needs to be demonstrated is that, while boundaries like those in the above interaction are arbitrary, boundaries that arise from characters are not – all characters are limited, each one in specific, and therefore characteristic, ways. The problem is not having a character say “I can’t do that” – the problem is him not being able to argue about why he can’t.

  7. Dirk Scheuring Says:

    Thanks for bringing up Selmer Bringsjord, Espen; I’ve all but forgotten about Brutus.1 as of late. It’s true, Bringsjord has a lot of things to say that is relevant in the context of interactive storytelling, like this remark about subjective awareness:

    Renowned human storytellers understand this concept. For example, playwright Henrik Ibsen said: “I have to have the character in mind through and through, I must penetrate into the last wrinkle of his soul.” Such a modus operandi is forever closed off to a machine.

    Identification. Human authors can do it, machine authors can’t. However, human authors can, by ways of identification, create motivated interactive characters, which are then rendered by machines. Brutus.1 only showed that you can’t let machines invent characters and their stories; that is the work for human authors who have that Ibsen quote pinned above their workdesks.

  8. andrew Says:

    Just to clarify, Mark has made two arguments in the past, that I and others have responded to: 1) tragedy, for example, is incompatible with player-as-protagonist, and 2) that games to date have largely not addressed broader issues of the human condition. I’m sort of conflating the two in my links at the top of this post; of course, the two arguments stem from a similar root — a general skepticism / cynicism / what-have-you about games and/or player-as-protagonist experiences (ignoring the intepretation of this all as merely a turf war). I can dish it out too, in the reverse direction, so to me, it’s all good; obviously I don’t mind being trolled now and then, for a good cause.

    On the topic of player-as-protagonist: one of the failures we alluded to early on about Facade (pdf from 2002, see bottom of page 30) and expanded upon recently (pdf, bottom left of page 2) is that it turns out in Facade the player is less a protagonist in the drama than an antagonist. (It’s a failure in the sense that our initial design goals hoped otherwise.) But how’s that for a solution?

    Player-as-antagonist is framing things in a similar way as Dirk is suggesting, I think. (Although I probably shouldn’t encourage him — now we’ll get 5 or 6 comments in a row, I fear ;-)

    Also, Marie-Laure Ryan’s recent paper, for which she had the benefit of a sneak-peek of Facade that she requested, does a bit of addressing the issue of what interactive drama will be good at or good for.

  9. Nicolas Szilas Says:

    [Ed note. – this comment was written last night but got stuck in our spam filters]

    Concerning Bernstein’s second argument about freedom: I believe I have solved (or avoided) this problem two years ago (sorry, not in 1994!) when I replaced freedom by reciprocity in the definition of Interactive Drama. Reciprocity means that instead of giving the user the full freedom Bernstein describes, it is interesting to confine him/her to the limitations of the simulated world, including the other characters. In other terms, the keypoint of Interactive Drama might be to limit the freedom of non playing characters, so that the user can have the (limited) freedom to do whatever the other character can do. Then, one can really say that the user is a character.
    Is that interestings to play with limited characters? What is sure is that given the limited interactivity with characters in current video games, there is plenty of room for innovation. In IDtension for example, when a character says something to the user, s/he can repeat it to another character, who can repeat it to a third character, etc. This seems rather basic, but it goes far beyond current video games (and it raises plenty of issues regarding the belief updates in agents…)
    That is why in a previous post, I have claimed that free interfaces (free text, free speech) might not be the only path to Interactive Drama. Indeed, as you know, free interface makes the user believes s/he has freedom, while s/he has not.
    Note also that this approach also alows some incremental improvement of Interactive Drama: once one piece will be created, with limited communicaiton abillities for the characters (including the user), some richer drama will be possible, with the improve of AI.
    It is not sure Hamlet can be adapted to that context. Let us find some drama that fit the medium…

  10. Kenneth Stein Says:

    Interactive narrative will obviate tragedy.

  11. WRT: Writer Response Theory » Blog Archive » Frustration, Irony, and Sanity Says:

    […] casion for his comment, so I’m trying to catch this wave of interest passing through Grand Text Auto. Mark Bernstein was generous about my post, and I appreciate h […]

  12. andrew Says:

    > Interactive narrative will obviate tragedy.

    That I don’t quite see… how so? While some pleasures of traditional stories may not be feasible in an interactive medium — at least without serious transformation of the way they operate or are achieved — I think the artistic ambition to create (perhaps even more amplified versions of) these pleasures still makes sense to strive for, or at least explore… and discover new ones along the way.

    Are you saying that the experience of tragedy can somehow fall out of, or emerge from, interaction? That I don’t see either, without a lot of help from the system and the system’s authors.

  13. andrew Says:

    Mark B. offers a lengthy response today on his blog, leaving it to me to link to it from here. Besides the title of this post, why Mark says I and Noah are only responding to his “Bernstein’s Challenge” is confusing, since the substance of this entire post and all the comments are focused on the “My Friend Hamlet” argument — a distinction further clarified by my comment above from 2 days ago.

    Anyhow, hopefully I’ll get a chance to contribute more to this discussion before I head off to Vancouver, because I’m finding it productive (even if flames are licking at our heels), but it may have to wait until I get back.

    Actually, come to think of it, I’d like to use this as a response for now.

  14. andrew Says:

    Read more at WRT, where Jeremy has a great follow-up post and discussion, Frustration, Irony and Sanity.

  15. noah Says:

    Oh, and it looks like I should apologize. For some reason I thought the part Mark was re-blogging was the “games don’t help us understand anything important about being human” argument. Puzzling why I thought that, given Andrew’s post and comment being focused more on the “my friend Hamlet” argument… Anyway, hence my disconnect from Andrew’s, Dirk’s, and other comments.

Powered by WordPress