Celia Pearce is a game designer, artist, teacher and writer. She is the designer of the award-winning virtual reality attraction Virtual Adventures: The Loch Ness Expedition, and the author of The Interactive Book: A Guide to the Interactive Revolution (Macmillan, 1997) as well as numerous essays on game design and interactivity.

Past conversations:
Will Wright - Sims, BattleBots, Cellular Automata God and Go
Louis Castle - The Player with Many Faces

Author homepage:
www.cpandfriends.com

Game Noir -

A Conversation with Tim Schafer

by Celia Pearce

(Interviewed at Game Developers Conference, March 7, 2003)

Celia Pearce: How do you see the relationship and the role of story in games?

Tim Schafer: Games have always stressed story a lot. Besides being the part that interests me most creatively, stories also really motivational for the player, to pull you through the experience. Puzzles can be challenging, but I know I go through games a lot of times because I want to see the characters through, to solve their problems, find that character who was kidnapped and see them through to the end. So, it's about motivation.

CP: In many of your games, there is some sort of dilemma that the main character is presented with at the onset of the game that propels the story forward. Often, the dilemma is very "gamey" if you will. How do you establish a dilemma that will motivate the characters through the game?

TS: I think you have to do two things at once. You have to provide the character with motivation and you have to provide the player with motivation. Because the character will care about things that the player will not necessarily care about.

CP: Like what? Can you give some examples?

TS: Well, you end up doing these little bribes with the player. Like in Psychonauts, the new game, you're a kid at the Psychic Summer Camp. There's a girl, Lilly, at the Psychic Summer Camp with you, and she gets kidnapped. And Raz, the player character, really likes Lilly, and he wants to go off and save her. But you don't know if the player really cares because he could just run and jump around and explore the camp and never go off and find her. And so you want to make sure that Lilly actually gives you some cool power or some cool tool in the beginning of the game, as a way to bribe the player to strengthen their empathy. You can't just rely on the story empathy, you have to put in little gameplay bribes, to make them like that character and want to pursue her.

CP: It sounds like you think about the player as a character too, in a way.

TS: I'm always trying to think of a way to describe the model for what we do. People compare it to film or Dungeons and Dragons, or something like that. There are these plays in L.A. where the play is set in this house, and you go from room to room following the actors around. You can hear a scene or not hear a scene. You can go to it multiple times and it's always different. Because if you weren't upstairs at that moment in the story you don't know who killed so-and-so.

CP: You mean Tamara?

TS: Yes. That's a little closer in some way than a movie. You're interacting with it. It's more like The Last Express, which is very similar. Because you can do it again and again, and every time, you miss something. Some games are more like the Haunted House ride at Disneyland, where you're moving through on a track, looking around. That's more like a rail shooter. I think of the games that I work on as being more like a hunted house—the kind you go to for Halloween—something that's been arranged for your entertainment, but more sophisticated, with a plot. You go in and a mummy jumps out of the closet and grabs you, but it's strung together with a plot involving the mummy. But even that's not exactly what's going on, because when you go to that house, you're just you, not a character.

So it's kind of like you're showing up that play, Tamara, and you have a part in it, and you know some of your lines, and you're trying to wing it with the rest of the actors. And they're trained in improv, so they react to you in a way that hopefully holds the fantasy together.

CP: The characters in your games have a quality of being comically heroic, like Guybush Threepwood in Monkey Island, who is a Danny Kaye type of character. You know, sort well-intended, clumsy but endearing the same time.

TS: Actually, we went through a big change over that. The first two games I worked on, Monkey Island and Day of the Tentacle, had these characters - (Bernard Bernouli) and Guybush Threepwood who were these kind of "loveable losers" who kept messing up. That's a plot device that's used over and over again in movies - you solve a problem, but then that actually causes this huge catastrophe, and than you fix that problem, and you get kind of a snowball effect.

CP: And a lot of times the solving of the problem is itself a sort of slapstick endeavor, like the whole business in Grim Fandango where you have to put the birdseeds, and the balloons, and then get the birds to pop the balloons. Half the time it's also kind of comical in its own way.

TS: (Laughs) Maybe that's more of a personal style. (Laughs) Maybe these are all based on yourself in some way.

CP: Usually, I can only play most games for like a half an hour, and then I get bored. But these are games I really played a lot, particularly Grim Fandango and Monkey Island. What I like about them is that the story is not gratuitous to the game play, or vice versa. Everything that happens tells you about the character. It's all integrated, it's not like a story slapped on top of a shooter game, you know what I mean?

TS: Right. And that's what we strive for, because the story is separate from the plot. It's not like a boring cut-scene intro to a level, a long cinematic that you watch before the actual game. I know it's always tempting for me as a writer to put those scenes in and make them long and beautiful, like movies. But that's the biggest challenge—my goal creatively would be someday to make a game with all the story elements that didn't have any cut-scenes in it at all, which is really, really hard.

CP: I give that as a parameter to my students in their game design assignments.

TS: No cut-scenes?

CP: No cut-scenes.

TS: (Laughs) You're kidding.

CP: And they whine and carry on. But they make better games.

TS: I'd like to do that someday. But right now that there are still things I want to put in there. We try to make them shorter. I was inspired by some of the storytelling in Half-Life, where they tell little micro-stories, where you're walking down the hallway and you hear this screaming and some shots. And you go "What's going on?" and you run around the corner, and you see a dead alien, and there's a security guard crawling toward the first aid kit. And it's obvious when you look at it: "Oh, he got attacked, and he shot the alien," and he's trying to get the first aid kit, but then he dies. It's a little tiny story that you pick up. But they didn't put in any recorded dialog, they didn't do a cut-scene. They just used their existing art assets, and they arranged them in an interesting way. So the level design can actually tell the story as opposed to the media.

CP: It's a narrative visual detail. Talk a little bit about dialog. I notice in your games that how things are worded is integral. There's a puzzle to solve, and you're getting a clue, and interpreting one word in a different way can really make a difference.

TS: You want to give a hint but you don't want to ruin the puzzle, and so you have to write them just enough.

CP: Obviously you come to this with a strong focus on writing. I'm curious how you arrived at a career in game design.

TS: I got my first job at LucasArts right out of college, and I was starting to abandon computer science.  I was a computer science major, but it was getting a little boring, and I was starting to get into creative writing. And that was the job listing—half creative writing and half programming. We had to write interactive dialog. So you're not just writing a sequential list of dialog lines, you're thinking, this line of dialog could happen before or after a hundred other lines. So it has to make sense that the player has done this, or the player has not done this. Or else you have to determine whether the player has done that action or not and then have two separate lines. And you don't want to do that, because you want to make nice tight dialog—plus it's better for financial reasons to have fewer dialog lines. (Laughs) So you just learn this style of writing dialog that works for multiple situations—either you write a different line for the second time you approach the character, or write a line that works multiple times.

CP: It's interesting that you came from a cross-disciplinary background, so you already had a procedural way of thinking built into the writing process.

TS: Definitely. I mean it would be really hard to do if one person was doing the writing and one person was doing the programming. Because I've never written my dialog in a script program. I've always written it in Scumm, or whatever language—into the code itself.

CP: Did you write Scumm?

TS: Ron Gilbert made the language Scumm, and that's what we used for Monkey Island 1 & 2, and so I would write the dialog directly into Scumm.

CP: The first game job I had was writing, too, and I used to write all my game stuff in Lotus.

TS: Ugh.

CP: I developed these really complicated spreadsheets, mainly it was because everything was on laser disc. So it was kind of a database that would call up frame numbers on the disc.

TS: Is that something that was released commercially?

CP: No these were like location-based entertainment and prototypes. We would write the script in Lotus, and then when we got the disc back, we would add a column with disc addresses, so that would become the master database for the program to go and find the proper clips. It's kind of old fashioned.

TS: (Laughs) But that's the fun part to me. In Psychonauts, we have a character named Boyd who is a conspiracy theory nut. He's based on an actual interesting character we have in the alley near where we work, who is always stopping and telling us about the radio transmissions beaming into his head, and how the government is looking at him through the broken glass on the street. So we have a character in the game that's always wandering around ranting about these conspiracy theories. I wanted to have him create this conspiracy theory on the fly, and constantly change it, and stop and start over again, and get confused. So we have this kind of randomly generated conspiracy that he's always trying to like figure out, so he's got this list of conspirators, like the government, and the illuminati and the Girl Scouts or whatever. And then there's a bunch of stuff they could be doing, like they could be secretly controlling ..., or they're blocking the production of ..., or they're doing something bad to ... some group of victims.

CP: It's almost like a Mad Lib in a way.

TS: Yeah, and that's the one time I really had to make a flow chart for how this could all go, and randomize. And then every once in a while, he'll stick in something, he'll cough or something and that's a little loop, or maybe he'll just say a non sequitur and then loop back. So there's a whole logic to these things. And hopefully it won't seem really artificial.

CP: So this is a variation of what I mentioned earlier—the idea of coming up with a character premise to then facilitate a game-like behavior. So you're saying "okay," he's kind of a crazy guy, and then all of a sudden, there's a great little conceit around which to build a character that's part of the game story. So the characters, by their nature, fit into the game world. In the conversation I had with Louis Castle, we talked about how in the Blade Runner game the character starts out not knowing if he's a replicant or a human, and that just sets off the whole story. Just that one character trait propels the story forward.

So this segues well to my next question: How do you invent characters? What inspires you? A lot of your games have a certain amount of satire of certain film genres, like Grim Fandango is kind of a film noir.

TS: It's a loving homage, not a satire. (Laughs) A lot of it depends on what you're inspired by at the moment, and what you're tying to do. It ties into what you were saying about the nebish characters. That was kind of perceived as a problem at the time, because actually Monkey Island and those games didn't sell as well as they wanted them to. And management was kind of like, "you know your heroes are not very heroic." But I think we enjoyed those characters because we liked poking fun at ourselves, because the characters are making fun of themselves a lot. And maybe not everyone likes to do that. I guess you have to be pretty secure to want to do that. So then I started thinking that maybe people would want to be something they weren't in their regular life, like a fantasy, someone who's tougher and cooler than them. And that's where Full Throttle came from - "I want to be a biker, I want to lead a biker gang, I want to kick down doors." I really think that all games are wish fulfillments of some kind, so why would you want to play a character that was not as cool as you?

In Monkey Island, we thought it was because they got to say "the good come back." They got to be kind of a wise-ass and say the funny line that Guybrush thought of but they wouldn't. But I still think it's useful with any character you make up to think "How is this a wish fulfillment? Why would anybody want to be this character?" Like the say in screenwriting class, are you writing a character that an actor would fight to play? That's actually a really a good rule too for characters in games. Would any player want to play this character? It's a very similar kind of question.

            With Grim, I wanted to do film noir, but film noir had been kind of done in games, like Killing Moon, and others. But people has only explored this one side, which I thought of as a very superficial take on noir—Sam Spade, sitting on the desk, the bottle of whiskey, the gorgeous dame walking in the office. And there are so many more examples of film noire. The one I really like is Double Indemnity, which was about an insurance salesman, you know what I mean?

CP: (Laughs)

TS: And, I mean, there's nothing exciting about that, and yet it's one of the greatest movies of all time. That's what inspired me to make the main character in Grim Fandango a travel agent. This is another angle on noire—the story where a regular Joe makes one mistake, or some small error in judgement, or tries to take the short way out, and gets involved in a spiraling world of crime and corruption that they have to claw their way back out of.

That's kind of what Grim is about—where he's going to get a jump on Domino to get this one lead, and he ends up trapped in the conspiracy.

CP: Grim operates on so many levels. It has the whole idea of the "Day of the Dead" world. It harkens back to something like Cocteau's Black Orpheus, which is the Greek myth of Orpheus going into hell to rescue his lover, reset during carnival in Brazil. The theme itself has this sort of noire element built into it—Day of the Dead—and then there's the whole art deco aesthetic mixed in with the Aztec motifs. I show it to my students all the time because I think it's a great example of a narrative world with many layers built into it.

TS: A lot it we picked just because that's what we liked. If you're going to spend three years on a game, you have to build a little, as they say, sandbox. You have to build something that's going to be fun to play with for a couple of years. The artists on the team brought a lot of the deco and art nouveau style to it.

CP:  So the world itself is also a character in a way. I'm curious when you're starting a new game and inventing a new world, what's your process? How do you go about creating a world?

For example, in the theme park industry, there is a process for creating fictional worlds that's fairly consistent, that everybody learns. I was curious if there was something like that in the way that you design.

TS: Hmmm ... I'm just trying to think if there's a consistent thread. Well, often, the world is the initial inspiration for the game. One day I was listening to someone tell me their stories of spending the summer in Alaska. They had hung around this one biker bar, with these people with names like Smilin' Rick and Big Phil. And I thought, "Wow, what a crazy world that is." It's so apart from everybody's life, and yet it's right there, it's so mundane in a way. And that's where Full Throttle came from. The world was the starting point. And Grim Fandango, also, seeing the Day of the Dead art, that was the starting point too. So it wasn't so a game idea, and then "let's make a world to fit it." You sort of stumble upon some world, and thing - that's something that's never been brought to life before. Let's bring it to life. Wouldn't it be fun to run around in that world?

CP: I was reading about the new game, Psychonauts, and I was really intrigued by the fact that you used this psychic vehicle as a way to do a game. Where did you get that idea?

TS: One of the great things about games is that they kind of accidentally explore surrealism, because they're so abstract. Myomoto games are very crazy and surreal, and people just accept it. I always wanted to do a game that really went crazy with that. I remember we had this sequence in Full Throttle where Ben was going to take peyote, and we had this interactive peyote trip. But it got cut because we didn't think that would really fly at LucasArts.

CP: (Laughs)

TS: But I've always wanted to have interactive dream sequences. I thought that would be interesting because all the layers of meaning. There was this psychology of dreams class I took when I was at Santa Cruz ...

CP: Is that where you went to school?

TS: For two years, and then I went to UC Berkeley.

CP: That explains a lot.

TS: (Laughs) There were two pivotal classes that have influenced my game work. There was the psychology of dreams class that I use a lot, and there was Alan Dundee's folklore class at Berkeley, which really inspired Grim Fandango. Folklore is such a great place to mine for ideas and stories. There's such a rich body of work that people have, for no commercial reason, dragged from generation to generation, so it obviously resonates with people.

And in the dreams class, there were so many examples of people having problems in their lives, and expressing them in their dreams in a disguised way. It's like they want to think about it, but they can't. And you can use so much of that ... going into a dream, you can have these crazy monsters and stuff, but knowing that they represent the fears of these characters, because you're inside their head. I thought that had so much possibility.

Taking this example of Psychonauts, where you're a young psychic kid, and you're going into people's heads, it's that whole idea of making a fun sandbox to work in. Be cause you can go into anyone's head, so each level of the game can be as different as the characters in the world. We have a character who's obsessed with black velvet art, and you go into his world, and his entire world is made of black velvet. It's a radically different style than the conspiracy theory nut, whose whole mind is literally a web of streets, spreading out from his house in the center, like concentric circles of conspiracy all around him. His mental state is represented by the level design, so that the actual world that he's in expresses his personality.

CP: It sounds Jim Ludtke's second game that he did with the Residents, Bad Day on the Midway, done in the mid-nineties. It was theme park, a sort of nightmare carnival where every character's brain was a ride. When you met them, if you wanted, you could go inside them, and the first thing that would happen is that you'd be in the ride inside their head, and then you'd be in their persona interacting with the other characters.

TS: You know I haven't played that. It sounds like I really should. (Laughs.)

CP: I think you would like it.

You were talking about the fact that you studied creative writing and computer science. Did you have it in your mind that you could do games with this, or was that just something that you came to when you got the job at LucasArts.

TS: No, I did games as a hobby, and I thought I would eventually be a writer, and write short stories. For some reason, it never occurred to me that you could actually make money making games. I just figured a very elite group of people did that. Instead, I had this idea that I would get a database programming job, inspired by a young Kurt Vonnegut, who worked at PG&E or somewhere as he was writing his first stories. I thought - "That's what I'm going to do. I'm going to get the most mundane job possible." Maybe if I'd gotten that mundane job I would have had so much more motivation to get something published as a writer. But because my job was very satisfying from a writing point of view, it never happened.

CP: So you played a lot of games. Did you program games as well?

TS: Yes. Not extremely well. I always loved adventure games. I used to play the old Scott Adams adventures on my Atari ... Savage Island, Voodoo Castle. It's like Zork but you only could type in two-word sentences, like "get rock," "use gun." I was really into games like that, and Zork of course.

CP: What about movies?

TS: Right when we were starting Grim there was a huge film noir festival in town. It was a strange coincidence, and I saw every single film showing there. I think I used something from every single movie. Like in Gilda, there's this character who always bets on the number two and wins, and that's how he gets his payoff in the Casino, and that's in the game. I just looted those games. Casablanca was the biggest influence on Grim, and The Big Sleep, and Double Indemnity.

For Throttle, it was more Kurosawa films. Yojimbo was a big inspiration for me - this really stoic main character, didn't speak unless he had to, seemed really gruff and like trouble, but he was really the morally pure character. Which is also kind of a noir thing, too.

CP: Writers usually talk about getting inside the character's head. Do you do that with game characters?

TS: I think you kind of have to think from their perspective, otherwise they become the victims of the plot. Say you need a character to be in a certain place at a certain time, so they just go there. But they don't know why they're there, so you kind of have to think about what they're up to. Getting inside the head of the non-main characters is also super important. Like for Throttle, the female lead, Maureen, is out on the road and she thinks you killed her dad. So there's this plot where you're trying to catch her. But instead of her just showing up in town, I made this chart of different states of the game, and what she's thinking at every state—kind of what she's up to. Every time you saw her, she was trying to go somewhere herself, rather than just showing up to give you the plot device you needed.

CP: One of the big debates that traditional writers have about interactive narrative is based on the premise that the decisions a character makes are what makes the character. So they're uncomfortable with letting a player make character decisions because that changes the character. So if you have multiple player choices, how do you think about making them consistent with the character?

TS: The old LucasArts school of thought was that we never used the word "you" in design documents. We never said "you open the door," we always said "Manny opens the door." You were not the main character, and that was a big design philosophy we all shared. I still believe in that. I mean there are games like Quake where you really are the main character. And I think that's totally valid, but that's a whole different kind of game. I like to do the game where you create a character that is a wish fulfillment, is a fantasy, is interesting enough that people want to jump inside their head and run around in it. Because that's an interesting thing to provide. It's like a different vehicle, like a car with different arms and legs and you make it so compelling that people want to try it out. They start to ego-invest, they share the motivations of the character. In any good book or movie, or anything, hopefully the audience shares that motivation, so that you want what the main character wants. If it's written well, you want them to succeed. It's the same with games. So you line up player motivation and the character motivation, then you'll find them doing the same thing. And when they diverge, it's kind of interesting to the player.

CP: So you try to map both the player and the main character, and think about both of them and where they overlap in their motivations.

TS: Yes, because otherwise the player will be like "Manny won't do this thing. I want him to do it really bad." Or, "Manny is talking about wanting to get to that door, but I don't care about that door, I want to go down here, because it's more fun."

CP: I want to go back again to our discussion about the worlds, because it seems like a lot of what you're talking about has to do with exploring the world. In terms of environments, are there any theme parks, or specific places that you are inspired by that have influenced the worlds that you've created? What are your favorite places to explore? Like if you were a game character, where you would want to explore? You mentioned haunted houses. Are there any others?

TS: A lot of the games come from the idea of a theme parks after hours—you're loose in this fantasy world, but done much better. I think Monkey Island, Ron will always tell you, came from the Disneyland ride Pirates of the Caribbean. He wanted to jump off the boat and run around Pirates of the Caribbean. That's very close to a theme park fantasy world, especially Disneyland - more narrative-based rides, not just thrill rides, the more atmospheric, "transport-you-to=another-world" rides. I think those types of rides are very closely tied to games.

CP: In Monkey Island you can also see the Disney influence even in the animation. Scott McCloud, who wrote, Understanding Comics, talks about is the relationship between the aesthetics of how an environment is made and how a character is made is very similar in comics, animated films, and games. The environments are usually drawn in a much more realistic way than the characters. His rationale is that you want the character to have a little space for the player's or the reader's imagination. You see that in cartoons too. You want the character to be a little more abstracted.

TS: I think that may be why it works, but it's definitely not why it happens. In games, it's just a production issue.

CP: (Laughs)

TS: Because you have to draw the main character over and over and over again, so you try to make him with as few lines as possible, and the backgrounds only need to get drawn once, so you get it super detailed ...

CP:  That was one of the things that I thought was brilliant in Grim Fandango. That came out at a point when doing 3D animation on a PC was not easy to do. And you guys came up with this skeleton head with this really simple automatic mouth movement, so you didn't have to have lips or anything ...

TS: ... or lip-synching .... (Laughs)

CP: ... but it didn't take anything away from the aesthetics. It was a great creative premise for solving a technical problem.

TS: I mean that's actually the genesis of Grim Fandango. 3D art had just started, but a lot of us didn't want to use it because we thought 2D art was so much better-looking at the time. I thought 3D characters in real time looked like a nylon stocking stretched over a bunch of cardboard boxes duck-taped together.

But when I was looking at the Dios de Los Muertos, I noticed they built the skeletons as these solid tube-shaped bodies with the ribcage painted on. And I thought, what does that remind me of? It was like cheap texture mapping. It's made to be mass-produced and built really quickly and so it's perfect for rudimentary 3D engines and stuff.

CP: That was when people were just starting to do this hybrid 3D background with sprite animation on top of it, so you felt like you  were in a 3D space, even though it was really just layers of animation over a virtual set.

TS: There was a game called BioForge, this really interesting game by Ken Demarest that came out a couple of years earlier. That was one of those interesting games, where your space ship crashes, there are fifteen people, everybody's dead but you. You hit your head, you don't remember who you are, and your body has been mutated by some scientist, so you look in the mirror and you're this hideous thing. And you find out at the end of the game, who you were on the ship, but it's different every time you play based on how you acted. So if you got through the puzzles by killing, you would turn out to be the more military guy on the space ship. And if you solve the puzzles in a different way you're a different character.

CP: That's kind of the Blade Runner model too. Depending on what you do, you're character becomes that thing. Do you do that in your games, do you think about that?

TS: They're not as flexible as that. There definitely is a stronger character set up, but I like to think of it more like the player is the intuition of the character. The character is a fully-formed person, but they're hearing this voice in their head that's saying "walk to the right." And they're like, "okay, I think I want to walk to the right." And the character always exhibits this cognitive dissonance. They act like they wanted to. "Yeah, I think I want to go over here. I think I want to open this door." But it's really you, you're kind of like this voice in their head, this Tourette's-syndrome compulsion - "Open the door. Open it!" And they're like, yeah, uh, I want to open the door.

CP: That's kind of a nice way to think about it—that you're the intuition of the character.

TS: Because you're not the actual thought of the character, you're sort of the hunch: "Go there. Something good will happen." It's like in real life, we're getting these weird impulses that we don't really understand sometimes, like "I think I should go to this party, I don't know why."

CP: I'm curious if there is a cumulative effect like Guybush Threepfoot [sic] ...

TS: Guybush Threepwood. It's from a P.G. Wodehouse novel. One of the main characters in the Jeeves stories is named Threepwood.

CP: Hmmm ... all these literary references. (Laughs) I noticed that there's a range of reactions that he has, sometimes of being kind of smart-alecky, as you mentioned before. Does that affect later things in the game? I know it influences the scene, like how effective he is at getting what he wants.

TS: Once in a while we would do something like that, where you could make friends with a character, and they would come back and help you later.

Real choice in games is tricky. If you have real branching, it can get exponentially large really fast. And it gets into these cheesy production issues. Are you going to record all that dialog? It costs $45 a line. Are you going to make all these choices actually possible?

Especially here at the Game Developer's conference, there's a lot of discussion about emergent behavior, especially with the success of GTA3. I think it's valid to start with emergent behavior, and make it more and more aesthetically pleasing and more and more convincing. But I come from a background where you start with the convincing, artistically satisfying stuff, and try to make it more and more emergent, more and more procedural as you go.

People don't like that as much because it sounds more like a movie model. The emergent behavior model is more of a game model. But I think in some ways we're working on the same problem from different ends. The goal is really to create this total immersive fantasy experience, where you're sucked into a strange world, where you are the character, and you're having all this fun, and you get to do anything you want.

Hopefully someday we'll figure out, with real AI and good simulation. There's possibilities that will actually happen. But in the meantime, is someone like me going to be happy with a computer-generated response, which is what you'd have to have? Probably not, so until then, we will have to employ this hybrid approach that integrates scripted elements with interactivity in order to make the experience as fun and seamless as possible.

[To the top of the page]

© 2001 - 2004 Game Studies
Copyright for articles published in this journal is retained by the journal, except for the right to republish in printed paper publications, which belongs to the authors, but with first publication rights granted to the journal. By virtue of their appearance in this open access journal, articles are free to use, with proper attribution, in educational and other non-commercial settings.