July 4th, 2008

Licensed Games I’d Want to Work On – Part 2: Silence of the Lambs

I’m not quite sure if this is my number two choice, but Silence of the Lambs definitely up there. The reason I chose to write about this next is because my previous post was very game-y (because it already was a game). I could be wrong, but I don’t think Silence of the Lambs is an obvious choice for a video game. I think if done correctly, it could be an amazing game that would potentially expand the game playing audience and push the boundaries of interactive entertainment.

Let me preface this by saying I have not read the books by Thomas Harris. I am definitely interested in reading them, but I have promised myself that before I read any other books, I’m going to read everything Phillip K. Dick wrote, and I’m only a third done. I don’t have much time for books presently, but the way I see it, I’m reading them faster than he is writing them (since he is deceased unfortunately), so I’m bound to accomplish my goal unless I meet an untimely demise. The Thomas Harris novels may be better than the movies as novels often are (I think it’d be hard to get much worse than the movie for Hannibal), but Silence of the Lambs is an amazing movie regardless. I also really enjoyed Red Dragon, and I have not seen Manhunter (the original film adaptation of Red Dragon). As many licensees have found out, just because a movie is good, it doesn’t mean the game adaptation will be good. I think Silence of the Lambs has what it takes to be an amazing game. Lets first look at what’s great about the movie and then I’ll discuss how I would preserve and enhance that in a game adaptation.

The one reason I chose Silence of the Lambs over Red Dragon to adapt is because I think the character of Clarice Starling is more interesting than Will Graham from Red Dragon. Her character is explored in Silence of the Lambs and a bit in Hannibal (I imagine much more so in the books, which is why I’m so interested in reading them), and Jodi Foster did an amazing job of bringing Clarice to life. Julianne Moore did a pretty good job in Hannibal, but she’s not the acting powerhouse that Jodi Foster is, although I love her portrayal of Maude Lebowski in the Big Lebowski. I’ve seen that film more times than I can count, and she cracks me up every time. One of the reasons that Clarice is more interesting than Will Graham is that she is very out of place in the FBI, and a lot of the conflicts in the films revolve around that. Firstly, she is a woman in the FBI, which I imagine is a male dominated organization. Now, I’m not saying that the organization as a whole is sexist, but I imagine there are individuals who are, and that will make her job more difficult. On top of that, being one of the few women in an organization can be uncomfortable, and I think the movies did a good job of conveying that. In Silence of the Lambs, she is often the only woman in a room or elevator . Also, she is from a small town in West Virgina, and it’s clear that she has seen some prejudice based on that too.

Although she may seem like a fish out of water, she is driven to grow lungs. One of the great things about the original Silence of the Lambs is the opening credits. Rather than just showing some text over a black background, or a picture of a butterfly, it shows Clarice training by herself. Now, one might think that they are just trying to make the credits visually more interesting than most, which is true, but I assert that they are trying to start the film with your first glimpse into the character of Clarice. She’s driven. She’s compelled to succeed against overwhelming odds. She’s pushed herself, and she will continue to push herself until she accomplishes her goal. It’s possible that the Red Dragon movie adaptation wasn’t as good because it didn’t really delve into the protagonist as much, but I think Clarice is a very interesting character and have trouble imagining that Will is as interesting, but I could be wrong.

Of course she isn’t the real star – it’s Hannibal. I’m excited to see Hannibal Rising, which comes out next month. I’m a bit concerned because giving him a back story may ruin the mystique of the character. Hannibal is an absolutely terrifying character because he is somewhere in between a brilliant genius and a complete psychopath. To be so intelligent and to understand what makes people tick so well, but to also have absolutely no conscience whatsoever is frightening. His intelligence gives him power as we see in the films. He only uses his power to pursue his own goals although it seems that he gained some sort of respect for Clarice although from what I understand, the resolution to the Hannibal book was quite a bit different than the film.

Besides the suspenseful night vision climax, the best scenes in Silence of the Lambs are the interactions between Hannibal and Clarice. Being able to capture an interaction like that and make it interactive is possibly the holy grail of game design. (Maybe not the industry’s Holy Grail, but my personal Holy Grail) Being able to interact with a computer simulation of a terrifying character like that would be amazing achievement in the field of interactive entertainment. Rather than just fearing for Clarice, you could experience it from her eyes. You could try to fight Hannibal in his mind games and that would be truly terrifying. In order for this to be pulled off well, you would have to be able to say anything. A dialog tree just wouldn’t give you the same experience – that you are battling wits with a psychopath. In order to be able to create an interactive character that could react to anything that the player could say, you would need a sophisticated AI that actually simulates Hannibal. This AI would have to have a lot of knowledge because that’s an important aspect of the character of Hannibal. This couldn’t be like a chat bot that merely tells you weather and movie times. Hannibal has a vast knowledge of psychiatry and the way people think. He isn’t like an encyclopedia that recites facts, but rather a person who understands these things and can use them as tools to get what he wants. To create the illusion of a real person, some sort of text to speech system should be implemented, so Hannibal doesn’t say things in the same pre-canned way. Unfortunately, current text to speech technology is notoriously bad, and doesn’t convey any emotion, which really ruins the humanity that we would try to convey. For game purposes, it would have to have ways of steering the conversation to stay on topic. This may limit the knowledge necessary to code into the system, but accurately steering conversations to fulfill game objectives is no small task.

Unfortunately, this is a very difficult problem to solve in the field of artificial intelligence and is currently the subject of a lot of research. The first hurdle is converting speech to text. You would want the player to be able to interact fluidly with the character, and the best way to do that is with their voice. This is a difficult problem. If anyone’s ever tried playing Seaman or using a voice to speech program, you know that these things are far from perfect. Let’s say that is already solved. The much harder problem is a field of research called natural language processing. Natural language processing is an unsolved and very hard problem that is the active subject of a number of research projects. Let’s look at an example. I might want to say, “If you give me that pair, I’ll cut you a deal.” Now, the computer program would be unlikely to get the text in that format. Firstly, I have purposely used a homonym. The text to speech program cannot recognize whether I am using the word “pair” or “pear”. Let’s pretend I’m referring to a pair of drawings Hannibal has made of the murderer’s house. In order for the computer to properly parse this sentence, it would need to keep track of the context of what we are talking about, which is pretty tough because there are a multitude of ways I could refer to the drawings: sketches, those, pictures, them, papers, illustrations, duo, couple, etc. Next, “cut you a deal” is a hard phrase to parse. Cut normally means to use something sharp to split something. Cut you a deal has nothing to do with that. I’m not exactly sure what the computer would make of it with a literal interpretation, but I chose it because it’s an example of something that seems natural for us to say that would be difficult for a computer to figure out. Next, let’s further complicate things by saying “I’ll cut you a deal for that pair”. Now, it is unclear whether I am cutting a deal by offering that pair or the previous meaning of the phrase. It may seem obvious to the player  because Hannibal was just showing us his pictures, but it is difficult for the computer to discern. Any time the game doesn’t recognize something that we say that should be obvious, the illusion is ruined.

I was excited by Lifeline, a game where your only way of interacting is with your voice. You help a character get out of life threatening situations by telling her what to do. I purchased it and tried to play it, but I quickly became frustrated. I wanted the main character to investigate the green thing in the first room. I wasn’t sure what it was (which is why I wanted her to look at it), so I didn’t know what to call it. I tried all sorts of things, and eventually the computer somehow matched my desperate attempts to refer to the green object to “go to the next room”. That was it. The game was a complete failure. Now, I certainly didn’t expect that it would be perfect. I didn’t expect to have long meaningful conversations with the protagonist. I didn’t think I’d be able to interact with a  believable entity. But if the entire gameplay hook is that you command the character with your voice, and she doesn’t do what you say, then it is indeed a failure.

The best example of a game that tries to do something like what I am suggesting is Facade. Unfortunately, I haven’t had time to play the full version of Facade, but I played the demo of it at GDC almost 3 years ago, and although it had issues, it showed promise. Facade is an interactive drama where you create your own character and interact with a married couple that you are friends with as their marriage falls apart. You have a limited range of actions, and unfortunately, you have to type in your text responses. Now as I said, I only played the beta so the released product could have been quite a bit better, but I would say we are still pretty far from developing the level of interaction that would “work” or be a success by my standards. An example is that I was trying to play the character of someone who was romantically interested in the wife and trying to break up the marriage. That’s the beauty of the idea of the interactive drama – if done right, you can create your own character and story. Unfortunately, it didn’t work out as well as I planned. Grace, the wife, didn’t even pick up on my blatant attempts to throw myself at her or my suggestive or downright dirty remarks to her. That’s the fundamental flaw with creating digital characters – with current technology, it’s impossible to get behavior that isn’t programmed in. To create a real digital character that can have unique and new reactions and emotions is the holy grail of AI. Firstly, if you are working with pre-recorded dialog, you are limited there. All of a sudden Grace can’t give up her husband and tell you that she’s always had feelings for you because there would be no dialog for that. Even if they tried to create an emergent system that had love meters for every character ala Fable and had text to speech and a complicated algorithm to generate new English sentences, it wouldn’t be able to come up with the response that she “always had feelings for you”. That was a creative leap that involved making up a back story. Even with a sophisticated character AI that tries to model all aspects of a human (which in and of itself is a difficult problem to solve), you wouldn’t be able to get that creative spark that a real human would have when doing improvisation. The simulation would only simulate the ideas put in by the author. It couldn’t share authorship in the story and come up with new input.

Perhaps I spent a bit too long explaining why my dreams our currently outside the realm of plausibility. Since this is my blog, I can express my blue sky desires. Although, if I were to make the game, I think I’d want the interaction with Hannibal to be the central hook, but there are many other things the game could offer. Detective work has been tried in many games (particularly adventure games) and has had some success. I would think with today’s technology we could create a pretty compelling detective game. Many believable characters could be created to give you clues and to guide the protagonist to solve the mystery if he/she is stuck. Ideally, in this game, any time you were stuck, you could turn to Hannibal for assistance, although his assistance always comes at a price. I think it would be really cool to do a sound based final boss fight similar to the end of Silence of the Lambs where Buffalo Bill has night vision goggles. It would obviously have to be much more forgiving than in the film where Clarice had a fraction of a second to figure out where Bill was and shoot him before he could shoot her. For the purposes of the game, the player would be presented with a  black screen. Buffalo Bill might be a bit clumsier, so there would be small audio cues like the sound of Bill’s foot bumping into a chair, or clothing brushing against furniture, but the player could take shots and the screen would light up briefly with the muzzle flash to illustrate where the enemy was. This would work out even better if surround sound systems were more common. Sound can be one of the most terrifying things. I watched the movie “Pulse” last night, and all in all it was pretty dull, (although I enjoyed the line where they “discovered frequencies they didn’t think existed”. What? Like -1 Hz?) there were some moments that were pretty creepy courtesy of my Dolby Digital 5.1 setup. Creepy noises came from behind me. Imagine sitting in a dark room with a black TV screen and a noise every few seconds that shifts in position as a pyscho killer moves in for the kill. Now THAT would be a game that would terrify people.

4 Responses to 'Licensed Games I’d Want to Work On – Part 2: Silence of the Lambs'

  1. 1Kevin Hayden
    January 24th, 2007 at 9:08 am

    Ok, I’ll bite. I think that Silence of the Lambs would possibly be the worst licensed game ever created. I know little bits of it (talking to Lecter, stumbling around in the dark inches Buffalo Bill) sound cool in theory, but overall, reading that post made me think:

    “WHAT’S THE GAME???”

    I know I’m going to come off like a patronizing dick, but I don’t see much actual gameplay in that idea. I see no core nugget of activity/purpose/fun from which all experiences stem. What would be the core of the game? Talking to Lecter? Interesting, but if you’ve seen the movie, what would there be to learn from Hannibal? You already know who Buffalo Bill is. You already know where he lives, what he looks like, where the kidnapped girl is located. You’ve solved the case before you get past the Start screen.

    The same situation existed in the actual movie, but it was forgivable because it was a movie. It was fun watching cop and criminal on two slowly colliding paths, and knowing things that neither party knew about the other (yet) because you were a casual observer. Putting you in the (cheap) shoes of Clarice Starling runs the risk of being annoyingly frustrating because you already know everything. Examples:

    - Autopsy scene: You walk into the funeral home and have the urge to yell out “OK! There’s a moth in this bitch’s throat! (”How did you know that Starling?) That’s not important now! Let’s go to the bug geeks to find out what kind of moth it is! Oh wait! I already know what kind of moth it is! Screw those guys!” ^_^

    - Buffalo Bill’s house: You knock on his door, following a lead… but since you’ve seen the movie, you just kick the door down and put two bullets into his forehead because you’re not falling for his bullshit. Open and shut case. Girl’s in the basement. Game over.

    So yeah…Silence of the Lambs, worst licensed game ever…unless you change the storyline to make things surprising…and if you do that, you have bigger balls than King Kong, and you had better be sure that what you come up with is just as compelling as the novel’s version, or you’re in trouble.

    This article you’ve written though, it is a good meditation on future “player-to-AI interaction” in games. I’d love to see a system like what you’ve described in an *original* game. That would be awesome. …But not Silence of the Lambs, man. That’s like playing roulette with being labeled a literary rapist.

    -Kevin


  2. 2Paul Skowronek
    January 25th, 2007 at 6:33 pm

    I don’t have anything too substantial to say (I agree with the concept generally and with Philip K. Dick emphatically); but I must say that I do not want games to spontaneously invent backstory. Ever.


  3. 3Matt
    January 30th, 2007 at 7:54 pm

    I don’t see why Silence of the Lambs has to be the exact same thing as the movie. Many licensed games deviate from following the storyline exactly. This can be for many reasons. Sometimes, the movie and the game are developed simultaneously and decreasing dependencies make it easier for both parties to work. Other times, it’s to offer a unique experience – something more to fans of the original work. A good example is the Chronicles of Riddick: Escape from Butcher Bay. That was probably a good thing they didn’t base it on the movie, since the movie was quite ridiculous. In terms of a movie that is part mystery like Silence of the Lambs, rehashing the storyline would be a poor choice. The license offers good characters and a great dynamic between Hannibal and Clarice. That would be what we’d be licensing – not the plot to the Silence of the Lambs.

    The game play would be a detective game. Although there hasn’t been a really good detective game that comes to mind, I think it can be done without resorting to pixel hunting or adventure game style item collection. Having a character (Hannibal) who can offer clues when the player is stuck is another good feature if the player gets stuck on the mystery themselves.


  4. 4Matt
    January 30th, 2007 at 8:14 pm

    Paul - What’s wrong with the idea of having a computer spontaneously invent backstory? In improv theater, working off of other actors creative input only makes the story more interesting. If all the characters are pre-made, then you will get a similar experience every time you place. If in Facade, one time you play being mean to Grace, maybe she will mention something about how you previously dated, and your relationship went sour. If I were playing, and I didn’t think of that twist, I would really enjoy that all of a sudden my role in the interactive drama changed, and I’d have to work from that. If I were playing a character that was trying to seduce Grace and break off the marriage, it would be cool to find out that Grace’s character always had feelings for me. What’s bad about that? To me, I think it only makes the interactive drama feel more interactive. If you are playing through and there are only a few ranges of reactions the characters can have despite wildly differing characters that you choose to play, that makes it feel less interactive. When I was playing the demo of Facade and trying to seduce Grace, I felt like my input was largely ignored because the designers didn’t think of that. (Although this could be addressed in the full game) If the computer could somehow creatively come up with stories to explain the current situation, that would be an amazing leap forward in interactive entertainment. Sadly, I think that leap forward is far away. Computers are the least creative entities in existence. The only way to get them do do something truly random is by feeding them outside input. (For an interesting way to get a truly random number, check out random.org’s explanation http://www.random.org/essay.html) If a computer can’t come up with a random number, how can it creatively co-author a story? If that is possible, that technology seems quite distant. Another problem with it is what if the stories the computer came up with were uninteresting? One would have to have some metric to tell the computer what an “interesting” story is. That in and of itself is a difficult problem. Perhaps if the computer could just do more of the leg work to achieve the story ideas – that would be good enough. For example, if the developer put in some interesting story ideas (and they’d have to put in quite a lot of them) and the computer could effectively choose which one would best fit the situation, that would be a real breakthrough in artificial intelligence. Perhaps that’s what the Facade developers are going for, but from what I saw, they haven’t gotten there yet (and again, I haven’t gotten a chance to play the full game).

    I’m planning on writing some thoughts on a Philip K. Dick novel adaptation next.


Leave a Response

- Why ask? This confirms you are a human user!

Everything on Binary Creativity is © 2006-2008 Matt Gilgenbach. All rights reserved. | RSS | Comments RSS