February 19, 2009

Reinventing Dialogue

This Blog has moved to

http://www.gamedesignreviews.com/.

This particular Page has moved to

http://gamedesignreviews.com/scrapbook/reinventing-dialogue/.

If you are a recurring visitor, please update your bookmarks & rss feeds. We apologize for the inconvenience.

14 comments:

Kylie Prymus said...

Ok, seriously, did you take the woman in the first image from an anti-depressant advertisement? That's exactly what it looks like.

I know so very little of actual game design and my programming skills stopped at using PASCAL in high school "Computers" class (yes, it was officially called "Computers") so anything I say is armchair at best. Still I appreciate what you're trying to do here by making dialogue options more transparent. Not sure if it really replicates the sense of Flow though. But then I'm not entirely sure what it is you mean by achieving flow through dialogue choices.

I get that you don't want the choices to be simple and obvious (which is not the same as transparent) and perhaps most importantly the player shouldn't feel like every dialogue choice has game-changing consequences (like the casual players of ECC). Real dialogue flow should be such that no specific choice matters too much, but the sum of the choices does. But as you've mentioned this is just a nightmare from a design perspective.

The problem I had with Indigo Prophecy/Fahrenheit is that sometimes I would make a choice and the character would say something entirely the opposite of what I had intended. I suppose that's what you mean by choices being not transparent enough. The question is what's supposed to be transparent in the choice: what the character is going to say, or what reaction the interviewee is going to have?

One thing we have to realize is that in any sort of dialogue heavy game the player has to make a sharp distinction between themselves and the character (unless the player can directly type/say dialogue but that's another thing entirely). What I like about Fahrenheit and what I've seen of Mass Effect (though I haven't played it myself) is that the full dialogue choice isn't there, so the player still gets to be a "spectator" of sorts. I want to respond angrily, so I chose to respond angrily, but then I get the joy of experiencing how that character responds angrily. This is different from, say, KOTOR where the full dialogue choices were there.

Long and muddled comment. My brain is still thinking about Street Fighter. Really great post though!

Tangential anecdote - I find it amusing that one of your graphics says fact: Shoeprints match shoes worn by Lise. If I recall you mentioned in the bortcast that you've not actually read The Brothers Karamazov so you wouldn't know that Lise is wheelchair bound for the entire novel. Guess that drives home your idea of the investegator fudging the facts to reach her own conclusions!

Digital Tools said...

Interesting ideas. And you are right - in a way. Leading to a good "rule of thumb", re-think design-issues, while focus on flow on interaction - that's good.

But on the other side I think, there is also more. Even on "poor interaction" beautiful things can be made. Who will not remember the games of Indiana Jones and Day of the Tentacle in a bliss, that are (of course not entirely!!) build around multiple-choice dialogs and made us so laugh with that charming humor. I generally are on your side, but good writing skills can also do wonder- maybe not for everyone, and maybe also nothing for "getting fast into it". Constraints can be good.

Digital Tools said...

Well, just like the concept you given. =)

emshort said...

Thanks for a really interesting article.

One thing that occurs to me, though, that may complicate Playfirst's evidence: the first dialogue options in Emerald City Confidential take place in a dangerous, high-tension situation, in which it's easy to feel that some of the options may lead directly to the protagonist's death.

I wonder whether the casual players would have had the same reaction to a lower-stakes kind of interaction as the first dialogue sequence of the game.

Anonymous said...

I would't diss multiple choice dialogue so quickly. Has anyone of you ever played the Fallout RPG series? Dialogue is very important and multi-choice, and choices and consequences are actually important. And your stats affect dialogue quite a lot. A must play for any gamer.

Krystian Majewski said...

Why is everybody always excusing themselves they were no game designers? Also, programming and game design are different things.

most importantly the player shouldn't feel like every dialogue choice has game-changing consequences (like the casual players of ECC).

In a way, but also if there are consequences, there should be immediate and clear feedback.

The problem I had with Indigo Prophecy/Fahrenheit is that sometimes I would make a choice and the character would say something entirely the opposite of what I had intended.

Yeah, I've experienced it also. This one problem I was referring to: because the verbs are always new and different, it's difficult to anticipate what effects they will cause. You can lessen the impact by spelling out the beginning of the next conversation in the choice menu itself but it's not a fix.

The separation between player and character is a good point. I've seen this work very well in Phoenix Wright. When using "Press" or "Push" I wasn't always exactly clear on what the problems with the testimonies were. I often just had a vague idea that something was fishy and was glad if the main character figured it out for me.

did you take the woman in the first image from an anti-depressant advertisement?

Haha, close enough. I took it from GettyImages. They tend to look very commercially. I missed the similarity because we don't have those advertisements in Germany. But I can imagine what you mean. :-)

Shoeprints match shoes worn by Lise

Nice one! Cool mistake. Fits the concept perfectly.

Who will not remember the games of Indiana Jones and Day of the Tentacle in a bliss, that are (of course not entirely!!) build around multiple-choice dialogs and made us so laugh with that charming humor.

I love them just as well as you do. This kind of interaction was ok back then. It was a quantum leap over parser-based interfaces of that time. I think in times of Web 2.0, DS and iPhone we need to think about how to push things forward. Implementing different interaction doesn't mean you have to give up humor and good writing - on the contrary. Phoenix Wright and Hotel Dusk both have some descent writing as well.

the first dialogue options in Emerald City Confidential take place in a dangerous, high-tension situation, in which it's easy to feel that some of the options may lead directly to the protagonist's death.

Good point! I might be wrong here but I believe the test version they had started different. But they mentioned a similar problem - the sheer nature of the first puzzle lead to (different?) complications with user testing.

Dialogue is very important and multi-choice, and choices and consequences are actually important. And your stats affect dialogue quite a lot. A must play for any gamer.

I've played Fallout 1 and Fallout 2, but I don't quite understand what you think makes dialogue in Fallout different in the context of my critique. Would you mind to elaborate?

dhalgren2882 said...

I like the direction you're going here, Krystian! I think it's difficult for a lot of us (including me!), to look at the mechanics beneath the game and try to innovate those in our designs, but I think that's most important, as you've pointed out.

The idea that depending on the player's investigation, any of the suspects, including the player, could be the murderer is intriguing. Even though the responses from the NPC's are still set in stone, the player's ability to access the responses is more intuitive.

I wonder, however, if this idea could be looked at as an even more confusing multiple-ending game. Now the player can get the number of suspects + him or herself number of endings. The dialogue system is, in a way, just obscuring the multiple choice. If you applied the mathematics to your design, wouldn't there be an even more intimidating number of paths toward those endings? I realize your design isn't intended to address those issues, since the short length of the game limits the density, but I'll be interested to see how you tackle those other issues in future designs.

I love the idea and would like to see this system implemented. It will be interesting to see how Jonathan Blow tries to revamp dialogue systems, since he said he's working on a game that attempts to do that.

Kylie Prymus said...

@Krystian

You're right, design and programming are two different things and it's important to recognize that. I think for most of us the problem-solving element of game design feels a bit like program because it's trying to use the tools at your disposal in a sometimes complicated way to produce what appears to be a smooth and easily accessed result. Like making a level in LBP. Having to learn to use the tools is the programming side, having a vision of a level is the design side, making it incorporates both in such a way that it's not always easy to know where one ends and the other begins. But it's important to make the distinction in general because people may think that because they have no programming skills (me) then they don't know anything about design (but hey, I managed to italicize in html!).

Regarding the immediate and clear feedback when a dialogue choice has big consequences. For the most part I think this is appropriate and important. Just don't forget the narrative power of occasionally having important choices seem mundane at first. We've all seen movies and read books where an off-hand comment ended up being the thing that sets major wheels in motion but the audience or characters don't realize until it's too late...

@dhalgren

With regard to the possibly confusing nature of a multiple-ending game I'm going to reference my intention in having them in my "prototype" post. I know Krystian also has some reservations about multiple-endings (hence keeping the game short which I think is good) but I sort of hope for a bit of a paradigm shift on the part of the player where they don't try to "figure out" how to get different endings. Ideally you would play through a game like this your first time, seem to "naturally" arrive at a solution to the crime, see your ending and be satisfied that you solved it "correctly". Perhaps you might have some questions about the designer's intentions - why did they chose to make it a spiritual crime (if that's the ending you go) - but if you play it again or do some research you discover that the designer didn't directly intend that. There were multiple possibilities so the ending you received - the "moral of the story" if you will - is based just as much on your choices. In the end you see the ending that you want to see.

Emmett said...

Hi there!

I work as a writer and designer within the game industry, and I say this only to put my remarks in context -- I don't think we "know better", necessarily, and there should certainly always be more discussion on ways to do things.

I agree with your premise, to a point. I know that my company has done some investigation into tracking player responses to gameplay, this by an outside research company that tracks such responses via eye movement, heart rate, etc. It's actually pretty interesting stuff. I know that we encountered some of the same results, that player's investment in the action dropped significantly during long dialogues and picked up only when they were actually required to make a significant choice.

There is another side to that, however, and that is the emotional investment involved. My only concern with the system you propose is that it might end up feeling a little mechanical, like interacting with a spreadsheet. While pre-written dialogue trees certainly do have their weaknesses, the one advantage is that they do sound more natural and players relate to them as easily as they might a book.

Whether a book makes for good *gameplay* -- perhaps that's a different story, but not every player is looking to game their dialogue. Does that make sense? I'd be interested in seeing how such a system you propose would work, but I'm wondering if there isn't a happy medium to be found that wouldn't involve writing an untenable amount of dialogue.

axcho said...

I really like your point about information density. I had not too long ago come across the first post where you mentioned that, and it totally makes sense.

Thinking about information density has helped give me some insights about why the intro puzzles for Foldit are so boring.

Also, it's amazing what a few concept screen shots can do for a game idea description! :D Seeing your mockups really makes the idea come alive and seem really cool. I can see how such a game would be compelling.

More, more! :)

Davo said...

Fascinating article Krystian.

I am convinced that this is the beginning of a new field in games, where dialogue will graduate from static flow charts, to something that is dynamic and can give the player unlimited options.

janjetina said...

The premise of this article can be boiled down to the statement "redding is teh hard".Judging the quality of interaction by the quantity of bits exchanged is flawed, as it lays on a false premise of indistinguishability between the syntax and the semantics. When you carry out an actual conversation with a person, the quantity of bits exchanged is fairly low, but the quantity of information exchanged is orders of magnitudes higher, as the words and phrases have multiple, context dependent meanings. Human brain doesn't consider the syntax independently of the semantics, and it's the semantics that contains the real information, and semantics of a natural language is impossible to quantify by counting bytes (characters). Implying that "Crime and Punishment" contains a few megabytes of information is hilarious.

Once we dispensed with that fallacy, we can assert that a good dialogue in a video game should approach the level of interaction between the people in the real world. In the real world conversation, you hear what the other person has to say, and, based on that and your own disposition, you form a response, that consists of one or more full sentences - you don't think "I'm going to respond religiously" and by some magic wait for the appropriate response to form itself. Choosing between well written, natural responses that convey the intent (including at least one neutral option) is definitely preferable to choosing the keywords ("religious", "intellectual" etc.). All choices and consequences that can be achieved by a keyword system can be achieved by dialogue trees. It is good writing that makes or breaks the dialogue trees and hiring a good writer costs orders of magnitude less than the money spent on graphic enhancements in games.

It is sufficient to provide an example of a "keyword / disposition based dialogue wheel" in Mass Effect, and compare its character interaction, that ranges between uninspired, flat and hilarious (when an unintended response happens, which is quite often) and counter it with an example of Planescape: Torment, a game with quality writing, where different dialogue options convey different personalities, where those dialogue options have an actual effect on the game world, and finally, where dialogues manage to reach the player on an emotional level.

So, let's stop pandering to the lowest common denominator and start making games for the literate.

Krystian Majewski said...

Wow, so many responses. I would like to thank you very much for taking your time and sharing your thoughts with us. As always, the most insightful part are the comments, not the article itself. ^_^

@dhalgren2882: You are right, I didn't really practice what I preach here. I don't really have a solution right now. I don't think there is a simple fix for those problems. The whole game must be designed with dilemma in mind. So I guess we need to try out different concepts and see what works. And I'm looking forward to Blow's new game as well!

@Kylie Prymus Good observation of small details turning out to be important later on! I haven't really taken this into account now. I think this kind of dramaturgy difficult to plan ahead in a concept stage of a game. It is generally difficult to anticipate what details players will pay attention to in a game. But ultimately, it's directing attention what good design/writing is often about.

@Emmett & janjetina: I would like to clarify something. I feel my concept has been misunderstood. It my fault because I concentrated so much on things that are different that I haven't really stressed out what remains the same. The game I envision does include dialogue. So you would read (or hear) a conversation between characters. The difference is in how you control this conversation. Instead of selecting the spelled-out answers as choices, you merely point out the manner in which the dialogue should unfold without being to specific. After you choose, you would still get a pre-written dialogue where characters act out what you've selected.
The other difference is that you can interact with the dialogue at any point instead of pre-defined "decision points".

@Emmett The things you describe sound very intriguing. May I ask who you work for?

@janjetina

"redding is teh hard"

I hope the summery above cleared out some of the misunderstanding. I would have less problems agreeing to your one-sentence-breakdown if you wrote "just reading is not playing". ;-)

When you carry out an actual conversation with a person, the quantity of bits exchanged is fairly low, but the quantity of information exchanged is orders of magnitudes higher, as the words and phrases have multiple, context dependent meanings.

Maybe. You are mixing different understandings of what "Information" is. But either was, it misses the point. In a multiple-choice dialogue the actual answers were written by the author not by the player. So their content doesn't count as input from the player. Just as buying a book in a store is different from writing it yourself - even if you bought it because you like and understand the content.

you don't think "I'm going to respond religiously" and by some magic wait for the appropriate response to form itself.

Actually, there are various theories how speech is constructed. The conseus seems to be that that in a conversation, the conscious focus is often on high-level strategic decision. At the same time grammar and syntax is mostly performed automatically. That's why as a Bi-lingual with L1 attrition, I tend to start sentences I don't have the vocabulary for anymore.

Choosing between well written, natural responses that convey the intent (including at least one neutral option) is definitely preferable to choosing the keywords ("religious", "intellectual" etc.).

Well, it turns out it isn't. Distilling the meaning out of a spelled-out choice adds a level of uncertainty and buden. A given sentence can be interpreted in different ways and can even change it's meaning depending on emphasis which isn't conveyed in written language. In the end, you mostly can only fit the beginning of the following discussion on the screen so it represents poorly what it is to come. That's why I suggested collapsing the dialogue choice into a general intention. Not to numb down the game but to express more clearly the available choices so players can focus on the interaction instead of on interpretation.

Interesting that you've mentioned Mass Effect. I just started it and while I agree that the writing could be better, the way the interaction works already in the same direction as I've been describing here. The selection wheel is separated into 4 different regions so players can anticipate what sort of impact an answer would have without having to interpret it's meaning. So for example selecting right/up will tend to end the discussion in a peaceful manner. Good thing too, because apart from dodgy writing the dialogue suffers from poor feedback.

So, let's stop pandering to the lowest common denominator and start making games for the literate.

You are looking from the perspective of a player who obviously has a lot of experience with the established system. I would ask you to try to keep an open mind. Yes, bad writing will break dialogue but although good writing is necessary, it isn't ENOUGH if we are talking about games. Games are about interaction so we must also look into different ways how to allow players to access and interact the dialogue we prepare for them. I certainly don't think my concept is the next big thing. I'm merely trying to come up with different solutions for that lazy, stale setup as I do believe it has flaws.

Musenik said...

You should probably look at The Witch's Yarn. It avoids the problem Playfirst discovered.

www.mousechief.com

"Star Trek DS-9: Harbinger" and "Dangerous High School Girls in Trouble!" use a form of dialog trees that always moves forward. See the latter at the address above.