|Interview: Jonathan Blow|
Jonathan Blow is the programmer and designer of the forthcoming time manipulation game Braid, which was recently previewed on Arthouse Games. Though not yet released for public consumption, Braid has already made it into festivals and won a major award. Jonathan has been working in the games industry since 1995. His mainstream industry work has included a port of Doom 2 as well as contract work on the Oddworld, Deus Ex, and Thief franchises. Along the way, he also co-founded a company to create the online sci-fi action/strategy game Wulfram, which still has players to this day. In 2002, Jonathan started the Experimental Gameplay Workshop at the Game Developers Conference, and he has been running it yearly ever since.
The following interview was conducted by email on February 9, 2007.
Jason Rohrer: Can you describe the circumstances that lead to your creation of Braid? Where did the initial idea come from?
Jonathan Blow: The earliest influence on Braid that I clearly remember was a game prototype called "Oracle Billiards" [you can download a working demo for Windows].
The idea for "Oracle Billiards" came about one day when I saw the film Matrix Reloaded. The film was terrible, for sure, but there were some very specific things about it that triggered me. One of the things I liked about the first Matrix movie was the character of the Oracle; the first movie seemed to have set up some interesting questions about her, but Reloaded dropped the ball completely, and I was really pissed off.
It seemed to me that the ability to see the future was fundamentally way, way more interesting than those guys were making it out to be. So I thought about how to explore that in a game, and pretty quickly I came up with the design behind "Oracle Billiards." It's a standard pool game---you knock the cue ball and it hits some other balls, and they roll for a while and stop. In all games like that there is a physics simulator that controls what the balls do---how they slow down, how they bounce off walls and each other. All that stuff is modelable by some basic math that you learn in high school or college.
The interesting thing, though, is that this math is deterministic---given a certain starting state of the balls (positions, velocities, sizes, coefficients of friction, and what have you), there is exactly one answer for any future point in time about where the balls will be, unless you decide to explicitly introduce a random element. Some pool games do add such a random element, but I decided to keep it out of the design here.
So for a particular shot direction and force, there's exactly one future about how the balls will come to rest. The premise behind "Oracle Billiards" is that you can see that future before you make the shot. (Nobody had really done this before, not only because it's a different idea but because it requires a lot of CPU power. In order to show you where the balls end up, every frame the game has to crank through all of the physics computations for the 30 seconds or so that it would take the balls to come to rest... in 1/60 of a second or so. But computers are fast now, so I figured, why not?)
This ability to see the results of the shot before you shoot fundamentally changes the nature of the game. It changes things from an aiming game into a strategy game; the aiming isn't a problem, you just need to decide strategically where you want the balls to end up, so that you sink a lot of yours, few of your opponents, and give him a poor position to make his own shot. I'm not an expert billiards player, but it seems that this is a lot more like the way the game is for expert players---they are pretty good at making the shots, so they are really planning ahead and making choices about which shots to take. So in addition to the basic mind-expanding nature of the game, it also gives you something like a glimpse into what the game is like for a small percentage of elite players, and that is interesting.
This was a little prototype and never reached a stage of serious development. I never got to the point of even adding pockets so that you can sink balls and win. The reason is that, whereas the game accomplished what I set out to do, the way it felt to play was not what I envisioned. I had hoped it could be an interesting strategy game, but it's not, really. The reason is that the physical interactions of the balls are far too chaotic for the visualization to help you make a plan. Small differences in angle are magnified hugely with each collision, so that slight changes in your aiming direction result in very fast changes to where the balls end up, so that it's very difficult to inspect the space and really make any kind of informed decision from the available choices.
My first attempt to deal with this problem was to add in the keys so that you can turn your shot very slowly and precisely. But it didn't help enough, so I decided that the billiard-balls-physics phenomenon was inherently too chaotic for this kind of game. (And this chaos also illustrates that the game isn't that close to what it's like for an expert to play, either---trying to capture that, to discover how those players really see the play field and separate some possibilities from all the others, would be a really interesting exploration in cognitive science).
Despite the fact that the strategy part didn't work out, there were some interesting side-effects from the game that I wouldn't expect. For example: watching balls rolling, you see two of them heading toward each other; normally you wouldn't be sure yet whether they are going to collide, but you see that the endpoints for the balls are off to the sides, not directly in their paths---so they must collide, and you can visualize the collision in your head, but what you envision doesn't necessarily put the balls where it says they will go. But then the collision happens, and they go right there. It feels pretty surreal.
So I wanted to pursue the same game idea in a setting that was much less chaotic. And the setting that jumped to mind was a 2D platformer like Super Mario Bros. If you take the Super Mario Bros. games and strip out all the extra stuff---mushrooms and raccoon tails and whatever---you can fundamentally get down to a very simple situation, a guy who walks and jumps, and bounces off monsters' heads. It seemed to me that that was a very simple template and could be used for exploring a lot of different things. (Because if the template itself is very complicated, (a) you run into a lot of complications and problems when you try to build sophisticated things on top of it, and (b) you're not seeing very clearly into the basic nature of the experiment you're trying out, because the template has a huge bearing on what it feels like to play).
So from this point I had in my mind to do a 2D platformer that contained weird stuff like being able to see the future. But I didn't go and do it right away, because it just wasn't clear to me exactly what the game design would be (for example---a platformer runs in continuous time, not turn-based, so how do you have enough time to stop and see coherent visions of the future and then act on them?)
I'm on a few mailing lists for game designers; some time later, on one of those, a discussion arose about Prince of Persia: Sands of Time and Blinx: The Time Sweeper. Both of these games provide the ability for the player to rewind events, like rewinding a movie in a VCR. One aspect of the discussion was that both these games used rewinding as a gimmick---your ability to use it was limited for when you collected power-ups or whatever. A very opinionated friend of mine, Casey Muratori, said that all games should give you the ability to rewind without limitation. In a lot of games we already effectively have that---the ability to save anywhere and then reload that save point -- it's just a much, much more inconvenient interface.
This was a controversial position, because if players can just rewind any time they want, then consequence and tension seem to go out the window. So there was this big argument with everyone taking a different position. Nobody actually tried implementing the unlimited-rewind in a game, which in retrospect seems kind of weird (but not too weird, because the people on the list tend to be pretty busy.)
At some point this integrated into my idea for the 2D platformer. It could be a game where you go from world to world, and each world has a different way that time works. In one world, you can see the future before it happens; in another world, you can rewind what you have done, etc.
This reminded me of Alan Lightman's book Einstein's Dreams, and that got me to thinking more about different ways that time can behave. And because I have a technical background it seemed obvious that the first world is the "no special rules" world---where you just get used to the basic rules of the platformer, and there is no special behavior of time at all. (These things all changed for the final game, of course; but this was the original conception).
Eventually I decided to start work on this game, and I started with the rewinding first, since it seemed like the clearest game design idea. I had never played much of Blinx, but Prince of Persia really annoyed me. Rather than making for a stronger game design, the rewind ability made it weaker. Because when you run out of sand (which allows you to rewind), you still get killed. So you have to have all the same kinds of save and restore inconvenience and interface problems as usual, but now you also have this extra complication of sand, which doesn't even work that well because the interface is super annoying (it's slow to take effect and hard to control, easy to waste, etc.). So it ended up being a bullet point to help sell the game, and to look cool in videos, but ultimately wasn't very good.
I think the Prince of Persia designers did things that way because they didn't know how to make a game any other way---how are you supposed to make a platform game where you can't punish/kill the player for failing jumps? How do you get around that lack of challenge/consequence? They didn't know, so they punted on the problem. Or maybe they didn't even think that deeply about it. Their preconceived notions about what it meant to be a platformer were so strong that the rewind couldn't blossom.
My main goal in working on the Rewind World was to do all these things right. The ability to rewind would be completely unlimited, both in terms of how often you could use it, and how far back you could rewind. It would affect everything in the world consistently (the sounds and animations and music would all play backwards---something Prince of Persia didn't do). But most importantly, it would be the foundation of the gameplay in that world. The rewind was fundamental, and the goal was to explore what that meant in terms of game design. How do I make the game interesting, given that the player has this ability? I was willing to throw out any and all video game conventions if they conflicted with rewinding.
So I did this, and it just worked great. Eventually the rewinding showed itself to be such a strong game mechanic that the entire game was based around it, every world had it instead of just one. The different behaviors of time were now just things that would interact with the rewinding, as opposed to replacing it. And as the design evolved, I tried out different things, discarding some different paradigms for time behavior, adding others, until you get to where the game is today.
JR: You game includes a text-based story that is revealed, paragraph by paragraph, as the player advances through the worlds. Why did you chose to present the story in that way?
JB: In the original version of the game, the story was lumped all together on an intro screen (just text, no gameplay), and you'd hit spacebar to go from there into the gameplay. First of all, that took the player out of the game, which felt bad. But also, a lot of people are not really interested in reading a lot of words on the screen, so they would skip it. When there is a screen that is only about presenting the story text, and you skip through that, there is a sense that you are fighting the game, that you're not doing what it wants you to do. I didn't like that.
I think that putting the story on the same screen where you choose which level to play reduces this feeling. You can still choose not to pay attention to the words, but it doesn't feel like you are ignoring the whole reason for that screen to exist. It also lets me break the story up into smaller pieces, which will be good when porting the game to handheld platforms with less screen real-estate (the text can be a readable size).
JR: What inspired the story?
JB: Earlier I mentioned the book Einstein's Dreams. I actually don't like the book that much---I think it's a good idea for a book, but I don't like the writing or the way it approaches the themes. But, Einstein's Dreams is clearly an homage to Italo Calvino's Invisible Cities, which has been one of my favorite books since I was forced to read it in college. Invisible Cities had a lot of themes that I felt were in the same mode of thought as my ideas for Braid: physical spaces as metaphors for mental spaces, metaphors about life accomplishment and a certain yearning. That's where the story pieces in Braid came from. But since a game isn't the same as a work of fiction, the way the story pieces came out, they don't appear to be so much like Invisible Cities. Which I think is the way it should be---if the objective were just to write a bunch of text that is like something Calvino would write, but shoehorn it into a game, well, I think that's not such a great objective.
Another large influence on the story is David Lynch's film Mulholland Dr. If you read about this film on the net, a lot of people are convinced that they've figured out the story, that there's this one specific interpretation that's clearly right. I think that interpretation is definitely hinted at, but I also think those people are fooling themselves; they are ignoring aspects of the film that conflict with the interpretation they want to believe. I think Mulholland Dr. is meant not to have any kind of linear, serializable story---it's more like a bigger space of things that connect up to each other, sometimes with interesting and weird topologies. And the movie is about communicating to you the vision of that rich space, not some flattened projection of the space onto 1D. It's also about making your mind explore the web of implications and possibilities to resolve a sort of mystery, but the mystery is so deep it doesn't have a solution in the 3D space and 1D time we inhabit.
It's like the following paradox that most people have seen before:
The next sentence is true.
The previous sentence is false.
If you just read these sentences as a sequence in time, and try to resolve them with each other, you can't. They contradict each other, and you can very easily just get stuck somewhere. Your mind can race along this path trying to resolve the mystery but there are not many places for your mind to go so you quickly realize that it's stuck. But if you approach this as a programmer or logician, then maybe instead of looking at it as a 1D sequence, you can break it out into a diagram with boxes and arrows that point at each other.
It's still the same paradox, but rather than trying to evaluate its truth value and getting stuck, you can step back and actually see and appreciate the structure of it. And from that elevated viewpoint where you look at that structure there's no paradox; it's just this simple thing. You can choose which way you want to view it. Mulholland Dr. is like that diagram, but in more dimensions and with many more boxes and arrows.
This kind of approach to story seemed to jive well with my vision for the gameplay; I have these worlds where time behaves in contradicting ways, and already a lot of the classical ideas from time manipulation are possible (alternate/conflicting realities, paradoxes where you travel into the past and change something that allowed you to exist, etc.). Mirroring that in the fiction seemed like the right thing to do.
JR: Some observers peg the peak of video games, as an art form, in the 1980s---during the so-called "Golden Era" of arcade games. Back then, lone creators, or perhaps very small teams, crafted some very novel games (Dona Bailey's Centipede and Toru Iwatani's Pac-Man come to mind). Do you feel that artistically-valuable games can be made by the huge development teams present in the industry today? Do they somehow hold an advantage, perhaps in terms of the sheer volume of high-quality content that they can generate? How can lone creators or small teams compete artistically?
JB: I think the way large-scale games get made today does make it difficult for deep artistic expression to happen. However, it is definitely possible---for example, I don't know many people who would argue with the idea that Shadow of the Colossus was made with a lot of artistic intent.
Part of the problem is just that there is so much chaos in the game development process. We see films that are artistic, that have huge budgets with lots of people working on them. But it's a lot harder to make a game than a film, and I think when it's so hard just to get something done at all---doing it with artistic intent often involves a lot of extra work, makes things a lot harder, and for many teams it would push the game over an event horizon, into something that is impossible to make.
As for small teams and lone creators, they have the advantage over big teams in terms of artistic expression. They can just do what they want, without it being caught in all these corporate mechanisms or the realities of compromise that happen whenever you've got a large team. As a bonus, it is much, much easier to make a game now than it was when I started; the barrier to entry is lower than ever, and you can actually distribute/sell games over the internet now.
However, many indies are not making games to be artistic; they are just making games, and that is the limit of their ambition. So you get a lot of indies that act like they want to be big companies (making clones of casual games, or making very traditional genre games). One definite power that indie developers have---their competitive advantage against the big guys---is the power to lose money, and to be okay with losing money. Most of the time, a big game company just can't lose money, and that controls what they can do, it even controls the developers' thought patterns from day to day. Indie developers don't have that problem so much---especially in the case of a lone developer who has a day job and spends his spare time making a game, he is free to not earn a monetary return on his time. Which may enable him to create something pretty great, without worry of pandering to an audience of a certain size.
JR: Can you give us a one-sentence definition of art? In other words, how do you differentiate works of entertainment from works of art?
JB: There are a lot of things that we call art right now, and if I provide a one-sentence definition it will exclude some things that we consider art. Some people use definitions like "anything that is not necessary for survival," and if you like that, that's great, but for me it seems too broad to be useful. For the purposes of what I'm doing, it's: the expression of something the artist feels important or interesting, with the hope that the expression resonates with an audience in a deep way.
JR: What electronic game do you consider to be the pinnacle of artistic achievement?
JB: I can't think of any games that I would describe that way. I don't think we're there yet, in terms of people exploiting the form in an effective way. There have been lots of games that achieved artistic things with graphics, or with story. But I feel that a game needs to express its message through gameplay---otherwise the author might as well have made a movie or written a book. Not very many games have expressed sophisticated things through gameplay.
At the Experimental Gameplay Sessions, we're going to show a game by Rod Humble, called The Marriage, that is 100% about expressing a complex theme through gameplay. The understanding conveyed by the game is very different from what you would get with words, or with pictures. It communicates to the audience in a way that is unique to games, and furthermore, that is different from what most game pundits have been describing as the communicative power of gameplay. The Marriage is a very simple game, but I think it's a promising start. I hope that after playing it, designers feel inspired to go out and build on that foundation.
|by Jonathan Blow||Tuesday, February 20, 2007 [12:13 pm]|
Re-reading this interview, I see that I came off as pretty harsh-sounding toward the Prince of Persia designers.
That wasn't my intention. Yes, I think they dropped the ball when it came to exploiting rewindability, but I think it is pretty cool that they put it in there in the first place; without Prince of Persia: Sands of Time as an example to start from, Braid might never have been made, or might have turned out very different.
So, thank you PoP: SoT design team, and sorry this interview came out sounding all negative about your choices.
|by Aubrey||Wednesday, February 21, 2007 [11:01 am]|
I felt the exact same way about that weird concession to convention with PoP's sand. I'm trying to think of reasons why they may have done it, but none are particularly convincing: It could be to do with how much of a "movement buffer" they had - they weren't able, perhaps, to store unlimited movement in memory. So, to explain that limitation, they provided the systems you saw (the radial bar which grows during regular play, and shrinks when you rewind time). That still doesn't explain why you'd bother to limit it to "sands", though.
Another possibility: there could have been some bizarre "game must last X hours" clause in the contract, which Ubi has been known to use in the past, and which has resulted in developers doing really devious shit just to reach that target - I have it on good authority that FarCry's monsters were jacked right up in the last few episodes to stop the game being anything close to a walk-through.
Still, this was a very insightful, very agreeable piece (It looks like you just needed an excuse to get it all out ;) ). Thanks for it, ArtHouse + JB. Really looking forward to Braid.
|by Z. Hiwiller||Thursday, February 22, 2007 [11:21 am]|
Can you provide a link to this "The Marriage" game?
|by Guy||Friday, February 23, 2007 [5:31 am]|
Thanks for saying that thing about art in gameplay! I've been making semi-articulate noises in that same direction for years but because of lots of things, probably the main one being my badness as a programmer, I've never been able to do anything about it. A game which is artistic in the sense of A Mind Forever Voyaging or King of Dragon Pass may have genuine artistic qualities, but it's essentially just the abduction of the artistry of other media. The artistry in a game, to be unique to the realm of games, has to be in the quality of the gameplay. I think to a certain extent there's a feeling of "art" in games which have very highly crafted standard-gameplay-designs - just because of the admiration that seeing something done so well provokes (I'm thinking of something like Settlers of Catan here, where it's a very traditional "build things and win" design but just very skillfully done) but it's not the same as having gameplay which actually has something artistic in it. If I could, I would make a game about feeding a skittish horse. :)
|by Joq||Wednesday, March 7, 2007 [3:55 am]|
The nesDS emulator (NES on Nintendo DS) actually features a nice rewind system: no matter which NES game you play, the DS's shoulder buttons are used to move time backward or forward. The rewind function is limited to something like ten seconds, though.
But I played through Megaman 2 with it, rewinding each failed jump...