Interview with Jonathan Blow

Some people want to build a company. That's not why I did that, but you have to do that if you want the freedom to make interesting things. If you are working for someone else's company, chances are those projects won’t be very interesting creatively.

There are other things about that. I not only run a game studio, but it's a self-funded independent studio. We don't do publisher funded projects, again because the things that publishers are willing to pay for are going to not be very creatively free.

I don't like what the “indie community” became after that. When Braid and some other games came out in 2008, it was a really good time. I think of 2008 as starting with N+. It was not as successful as the games that came out later in the year, but it was quite successful. I think it sold 70-100,000 copies, a lot, and it was also on Xbox Live Arcade. That was the signal that something was happening. Of course, later in the year there were Braid and Castle Crashers, and some other games.

That was interesting, but it wasn't exactly indie developers in the way it is thought of today. We were people who were independent. We were trying to make games, but we didn't self-identify as part of some indie game community. The people who came along to self-identify that way – I'm speaking of an aggregation – I feel their priorities were pretty wrong, or at least very different from my priorities.

When I sit down to make games, I'm interested in making the best thing. Both in terms of the idea, and in terms of making sure that my execution is good because we build everything technically from the ground. It means making sure I'm a good programmer. It means making sure that I'm good at scheduling my time, and making sure I'm working efficiently and all these things. And making sure that the things that I'm doing really are important to the vision that I'm trying to make happen.

I feel the indie community that arose since 2010 is based on a very different idea. It was based around good sounding philosophies that anybody can make games, and we can all participate, everybody can express themselves through their games. You're part of a community, and everybody is welcome in the community.

Which are all positive ideas. But the problem is they don't take very seriously that making games is hard. Making good games is really hard, and you have to have constant vigilance that you're doing the best thing.

It became very much a community that I didn't feel like I belonged in at all frankly.

Part of the reason I make independent games is because I want to be creatively free, and part of being creatively free is to not be bound by these commercial constraints that we all are aware of, because they pervade the whole environment of this conference [Game Developers Conference] that we're in. But the reason those commercial things are what they are is because they work. That is what gets people to buy your game.

When you make a decision to not do those things, then you're making the decision that it's quite likely that people will not buy your game. That's a decision I made when making Braid. That was a weird game for the time. The Witness was a really weird game – mechanically especially. It doesn't really have game mechanics in the traditional sense, and it's a 100-hour game.

I said, how can we encourage people at the conference to be more concrete about this complaint? How can we encourage doing more new things? But there were people on the internet doing interesting crazy things. How could we highlight that to people who haven't heard of these things?

And could we make it more deliberate? The reason it was called Experimental Gameplay is that when you're doing experiments, you have something specific that you're trying. Maybe you will find out what it feels like if it works, and what it feels like when it doesn't. This is little bit different just doing a wacky thing that is different from what exists, but with no real idea about what you are trying to do.

I felt that if you're going to progress as a movement of doing more creative things, it needs to know if it's succeeding or moving forward. That was the idea at the time, and I think it was a good time for it. Today, it's very easy to make games and a lot of people are making relatively unconventional games. It's still drowned in this giant sea of people who make derivative games.

There's a Twitter feed of trailers coming out on Steam today. Most of them you can immediately identify,  and it's not that interesting, but once in a while something interesting comes out. By percentage, it's probably the same percentage as it always was of things that are creatively different, but by volume there are more because a lot more is being made.

How did that happen? These game engines happened, and the internet made it easier to find audiences or to communicate. This certainly has very little to do with that Experimental Gameplay Workshop. I don't think that you could say that it won.

But, given those constraints, there's still a great deal of freedom and expression, and an ability to do interesting things. As opposed to, say, poetry where you're extremely free and have very few functional constraints that you have to meet. Unless you pick a very constrained form of writing. I detected back then that many people were doing undirected creativity. Random wacky ideas to me do not feel as good as well thought out things that work.

The way I design games often involves taking some foundational axioms of game play and then exploring the permutations in different forms. The Witness does that. It's an extrapolation of patterns that happens not at runtime, but at design time.

There's a lot of things to say about that. I wouldn't say it exactly misleads you in the beginning, but it only really directs your attention towards specific things, and you have to broaden it out yourself. That's one of the hopefully joyful things about that game is that you get to expand what you're paying attention to. That can be a pleasant surprise when you realize how you need to expand.

The play of form I was just talking about helps to generate those opportunities for realizing that you need to consider more than you had previously considered. Sometimes you're stumped trying to figure out a puzzle, but the reason you're stumped is because you're not paying attention to something that is very obvious in your visual or sensory field, and you're just filtering that out. Because you've decided it's not relevant, maybe even pre-consciously, and then suddenly you realize it's actually relevant and gives you the answer.

There's a delicate high-wire act to designing such things because all the information has to be there. There are one or two puzzles where you have to look in a specific spot, but usually that is telegraphed by other things as well. The design discipline was to make things as obvious as possible while also knowing that people wouldn't see them.

What guided what those surprises was this play of form where you say, "this is a game about conveying information about what the solutions to the puzzles could be, and the design is exploring all the ways in which we could convey that information given the fiction of this simplified world.” There are light and shadows, and sounds. The design part was about how to use these elements with each other as much as possible without getting too contrived and conflicted. The thing about reality is it's like a play of form, but things start interacting with each other in surprising ways that make things complicated very quickly. Which is why, for example, it's hard to make software like we were talking about before we started recording.

Because before you realize it, things that you didn't want to interact with each other are interacting with each other. That's a fundamental aspect of reality. For example, as an engineer you normally wouldn't think of your pixel shader that you're using to draw your surfaces as game design. It simulates the way light bounces off surfaces in reality, in an idealized version. Given that, there can be some understanding that you have of the angles at which things are happening and so forth. That can make interesting puzzles. That brings this aspect of the game, the shader, and it ties it to the core game idea. We do that with audio. We do that with color. The way that color is perceived by the human eye.

But it's designed intentionally so that's possible. If you have too much tunnel vision, you will not see any of these things. I didn't think of it specifically as punishing people for a certain attitude, but I guess it does that. Or maybe you just get the game that you expect in that case.

It's very out of the box the way you have to solve it. Maybe you just don't know right now. Maybe your subconscious has to work on it, and you have to come back later, but there's no recipe you can follow.

It goes back to Calvino’s Invisible Cities, which is the earlier version of that. At the time I was coming to GDC every year, and all these game people thought about trying to do better stories, but their model for stories was action movies. If you want to do good stories, your model should be something else.

With Braid, it was a little bit more, "I'm going to take the way that these books that I like are written and do the level design version of that, or the game-mechanic version of that.” But for the story part of the game, I was going to take more influence from books than from movies.

I thought of it as equivalent to an illustrated book. An illustrated book where it's mostly text with some pictures. I thought of the gameplay being the equivalent of the text, and the text being the equivalent of the pictures. There are occasional spurts of text that give you something new, and there's gameplay which is the main meat of the game.

Maybe there's a grain to each genre that lets a novel do that maybe a little bit better than a film. But I've certainly seen films, so my favorite film is David Lynch's Mulholland Drive, which wasn't intended to be a motion picture at the beginning.

When I watch that, by the time I'm done watching the movie, it has built a non-linear space of ideas in my mind. It's certainly possible to do that. With cinema, I think it's just a choice the commercial industry has made, to go to this linear model. But, also, that linear model is what these game designers are using as their model.

When you play a high budget commercial game, their idea of a story is, Lethal Weapon 2 or something newer.

I'm not sure that I'm a particularly well-read person. Maybe for a game programmer I'm very well-read, but compared to like somebody who reads a lot in life, I'm probably not. I could come up with many examples of well-regarded books that are part of the canon that I've not read and don't know anything about. I've read really good books, and those books have really good stories. Those stories look nothing like what video game people are trying to do. Nothing at all.

Am I trying to make games more literate?

The direct answer is no. not really, because what makes literature good belongs to literature. You certainly could make a game that's just text on the screen that you read, and then it's basically a book with a worse UI.

In that case, okay, then we can do everything literature can do. But, if you go toward the natural grain of what games are like, then you leave the area in which most good story-telling happens. There's been this thing for many years in academic circles about storytelling being important to the future of games. For some academics, the question is how to make AI generate stories that are worth playing. The industry version of that is more like Lethal Weapon 2.

Neither are in my opinion a clear view of the road ahead. Simply because games are just really different. Also, if you say, “wow, that was a good story” after a movie, you're thinking about very different things than when you’re saying it after you read a novel. The good story in the film version probably had a lot to do with cinematography, with performances, all these things that don’t exist in the novel. When we talk about these things, we're often not specific enough.

You would expect the game version to also be different, because our medium’s interactivity makes it more different than film and novels are from each other. So different that you shouldn't really say it's a story anymore.

For games, a lot of what makes a story has to go out the window. The things that we respond to about a story certainly can still exist, but it's hard because you have to re-invent it from almost nothing, and it's easy to just follow examples from movies and books. Those can get you a certain distance, but maybe it's a cul-de-sac, and you have to retreat on all your progress to get to the main road.

There's all sorts of things to say about that. Let's just say something about tight versus sprawling for a second. Braid is super tight in the sense that every world apart from the last one has the same structure. There's a same number of puzzle pieces. You solve exactly one thing to get a puzzle piece, and there are other structures, but it's a very tightly constructed thing.

Witness, I would characterize as more sprawling. There are 600 and something puzzles in it, and they're on different subjects. Sometimes what happens is that one subject has more interesting material than another subject. To force them to have the same number of puzzles in that case wasn't right. Some areas have four puzzles, and some have 37. That's just how it is.

It's sprawling in that way, but The Witness still is tightly constructed in the sense that almost always when we build something there's a very specific idea in it. We're not just doing some stuff for fun in the middle of the game, which Gravity's Rainbow has tons of. It might be entirely built of that. I don't think I meant copying the structure of the book in any away.

But, the density of ideas, and the depth and subtlety of ideas is something we should be paying attention to. Ideas in games today are usually very ham-fisted. They have no subtlety. Compare The Last of Us, which people think is a nice drama, to Gravity's Rainbow. Gravity's Rainbow has at least a thousand times as many ideas in it as The Last of Us has. Because it has so many more ideas, some of them are more subtle, and a large proportion of readers will never understand, including me, but you feel them there and you feel around the edges of them.

I was aiming more in that direction with The Witness, where we put a lot of ideas in, and people may not get everything, and that's okay. I don't think we can get anywhere near a book in terms of numbers of ideas. Simply because it's just a lot easier to write than it is to program and design a game that functions. I don't know if we'll forever be behind something in terms of idea density.

The other thing is that Gravity's Rainbow is a very free book in a certain way. It doesn't care if you're following along exactly. It has a certain freedom and joy to it while also being pretty dark a lot of the time. I can't think of a game that has managed this emotional pan-spectrality of being very joyful and playful while simultaneously being bleak.

With Witness, we have a large number of quotations in that game, but they all have citations on them. The videos don't, but you can usually figure them out where they came from.

What I'm trying to get at is, certainly when I made Braid, is that there's a fundamental difference between what some people think games are for, and what I think they're for. To make it a little more distant again, one thing I hear people say about movies is that movies are great because they're escapism and help us forget about our miserable lives. And that's not why I ever went to see a movie.

When I see a movie, mostly I want to explore the ideas that the movie gives me, and I want to go toward some vision that the makers of the movie have to show me.

With games like Braid or The Witness, there are a bunch of ideas in the games. The games are made for people who are interested in those ideas. They're not made for people who want to escape. Maybe playing a hard puzzle game that deals with existential issues is not the best way to escape from anything in the first place.

In those two games, the citations help tie the game to the real world. They say that this isn't just a toy. There's a continuum between playing this game and other things that I might think about during the course of the day when I'm not playing games.

With those two games specifically, the idea of what we were going to make came very quickly at the beginning of development. Braid is pretty far from Super Mario except in the fact that you jump, and we have little monster guys. The Witness is even further from Myst than Braid is from Mario. There's just something fun about using those templates. For both games, there's a specific functional reason. For Braid, the functional reason is that time manipulation could be really confusing.

For you to understand clearly what's happening, it helps if there's a very simple world with rules so simple that they would be boring if that was all there was. Braid is not even based on Super Mario. It's based on a super-simplified version of Super Mario that nobody would play, because the monsters are not that interesting, there are no action puzzles, the maps are not very big, and all these things.

But if we start with that simplicity, and do the time thing on top, the simplicity makes it tractable. When you're confused, or when you're trying to solve something, what you're trying to solve really has to do with time, and not with some other irrelevant thing in the game.

With The Witness, it was similar in that we wanted a mentally clean environment. In order to have subtle realizations about what you need to pay attention to, there has to be only few irrelevant noisy things grabbing your attention. That encompasses visual design first of all, but it also encompasses game design. If there's a guy running around trying to kill you, then you can't stop and clear your mind.

Myst was very appealing from that standpoint because you're completely alone in a place, and you proceed at your own pace. There are functional reasons for starting with those games, but usually it's not to comment on the specific games.

I just said something that wasn't true, because there actually is commentary in The Witness case. Games like Myst were exciting when I played them, but they were also pretty frustrating in a lot of ways. The Myst gameplay is just broken in ways that modern game designers would understand.

You go into a room, and you don't even know what on the screen is interactive, so you click on everything. You don't know what the things did. Maybe sometimes that's intentional. It's part of an interesting puzzle, but a lot of the times it's just confusing. Whenever I play games like that, I remember what I didn't like about them.

This is funny, because the indie people I complained about before often think I have a very negative personality. It's just a critical eye. I register and communicate truthfully about the criticisms that I have about things. Anytime I play a game, that stuff is in my head, so 20 years or 30 years later I remember what were the not-good parts of Myst. And then I seek to make a game design that doesn't have that problem.

That's why The Witness has explicit panel puzzles, because I didn't want people to be confused about what the puzzle was, just about how to solve it. There's an extra level of clarity that Myst doesn't have. I don't intend that to be exactly a commentary on a game like Myst. It's really a commentary on the whole genre of Myst clones, in the sense that I'm grumpy, and want to do it better. If I were to make a game the purpose of which was to comment on game design, I think it would look different.

That’s an interpretation that many people jump to. My frustration with a lot of commentary on Braid, for example concerning the reference to the nuclear bomb, is that people come up with a simplistic interpretation such as, “Braid is really a metaphor for the United States Nuclear program” because they found a specific reference in the game.

 The nuclear bomb is present as an idea, but then people decide it is the central focus and explanation for everything. The game is supposed to be more fuzzy than that.

This is a funny thing, because I'm in Indie Game: The Movie saying something like this, but too much context is edited out of the movie. People interpret that part of the movie as me being upset with the way the general audience doesn't understand the game, which is not really what that was.

I was upset with game critics specifically, for not understanding the game. I felt they really didn't get it, which it's their job to do. I was a little miffed at that. There are one or two things that make me mad. One of them is false teachers.

There is an echo here of the early idea that anybody could blog, so anybody could be a game critic and tell everybody that they were the person who could explain games. I just think if you're going to say that, you have a responsibility to really think deeply about the games.

I don't get to tell you what it meant to you, but if I see the sign that somebody didn't even notice blatant patterns in the game that anybody would notice, that starts to get upsetting when you're telling people what the game means. Not only that I think are wrong, but that this person has only thought about for 30 seconds. I got really upset about that. That's all that was.

Maybe that's true. I remember an interview I saw with David Foster Wallace, right when Infinite Jest was coming out. He was like, "All these people are writing reviews of this book, and I know they haven't really read it because they've had it for five days." It's a dense book, and how would you really have read it in that time? That's basically how I felt about the response to Braid.

If I can be very blunt, I felt like I was responding to people who were talking in order to promote themselves, and to use the game as a way of demonstrating that they were smart. That really made me mad. These days, I guess I got it out of my system. When The Witness came out, I really didn't pay attention to what critics had to say. I hardly read any game reviews. I read two reviews from people who I knew, that I respected. That was it.

The Witness is such a deeper game, and with so many more subtle things in it. If I was that upset about what people had to say about Braid, then The Witness was just a total loss. Don't even bother. Just don't even go there.

There's a common motivation and then there's a difference. The common motivation is when you're Indie and don't have a huge budget, you need to make something visually interesting somehow. You can't compete with Assassin's Creed 27 for budget because that's never going to happen.

You need a strategy for being visually interesting, and then the question is just what. Some indies have gone down the road of pixel art or vector art. Because games used to look like that. That was fine, or maybe it's some kind of fetishization of older things, which I don't share because I actually grew up in that time. A lot of those games were fine for the time, but they're not good if you go back and play them today.

The reason the games look like that is because people couldn't make them look any better, which is not to say that you can't do really impressive pixel art. The motivation in those cases is to look like something else. When we did Braid, I was very careful to not fall into that attitude of pretending to look like something else.

We did start with painting. We were trying to make an expressive game, in a way that video games usually aren't. What keeps video games from being that expressive? If you have a bitmap image and you're just splatting it down in many places, and it looks the same every time, that has a very different character from if you hand-drew something many times with different nuances. In the final game, there's a level where you have hunt down all the monsters, with bitmaps of monsters at the top. We actually have a different sprite for each one, even though it's the same monster. The monsters walking in the world we don't have different sprites for, because that would have been prohibitive.

The goal was not to pretend to look like a painting, so that you can impress people that it looks like a painting. Because I think that's a little vacuous. The goal was, instead, to take some of the things that are good about painting and use them to make our games better.

The texture of brush strokes is part of what makes images interesting sometimes. You don't exactly get the same thing on a game screen, because with an actual painting you have light bouncing off the object. To the extent of what we could with our small budget, we were able to do that. A lot of people say watercolors, but we didn't try to emulate watercolors either.

We knew that we were going to layer a lot of things on top of each other. When you blend things, especially on an older GPU, you have two choices. You can do a cross fade blend or an additive blend. We use both of those. But they have this property of layering semi-transparent things on top of another thing, which is what watercolors are like. It's not what oil paints are like usually. We just naturally arrived at what it was. It was using influence from those areas, but allowing it to become something different.

Yeah. In The Witness, the idea was to have visual clarity for the puzzles. There are two reasons. One is to prevent noisy information from distracting you. When you do discover some of the things in the game that are hidden in plain sight, the more obvious they are, or the more resoundingly visual they are when you see them, I think the better the impact of the discovery.

If something is huge, and in bright colors is on this wall right here, and you're like, "Why didn't I pay attention to that?" Then that's a stronger experience. The art style was designed to support that.

A lot of video games use texture maps that have a lot of details, because that's how they want to make the surfaces look interesting. We don't do that. We relied on lighting to give us most of the looks. A lot of our walls are something like that. You don't see that often, or perhaps nowadays you see it in low-budget games, where they didn't texture everything, but just turned on the light mapper.

We didn't have that back then we started The Witness. We rely on something like the light falling on the surface to provide visual interest. It's also low frequency visual interest, so it doesn't snap your eyes' attention to details very much. That gave us control over if you walk into a particular area, what seems important and what doesn't. We were able to focus that very clearly.

I certainly thought about different visual styles. We could have done this by painting again, but the emulations of painting in 3D that I have seen are all terrible. They don't capture the good part of painting. They just pretend to be painting.

As a technical person, I spend some time thinking about how I would do it, and make it be better. But even if you were sure that you wanted to do a painting style, why are you making it look painted in the first place?

It shouldn't be a pretentious thing. It should be that you want the subtlety and the expression that you get with painting. There shouldn't be technical limitations of your version of the technique messing those things up. If you walk around an object, it should look beautiful and painterly the whole time.

It shouldn't be like you see in a lot of these technical papers that are essentially fitting brush strokes to the 2D snapshot of the image. As you walk around, these techniques can't really be continuous. They have to invent and remove new brush strokes, and you would get flickering. I never have been very happy with the attempts that I've seen at that. It might be interesting someday to spend a year doing a very good version of that, but I've got too many things to do.

We decided for a more cartoony look. Differentiating myself from AAA was not really the primary goal. The goal was to serve the gameplay well, and then try to look good by any means necessary. I agree with you that it looks more like any 3D game than Braid looks like any 2D game. Part of that it's just it's hard. It's hard to do something authentic in 3D that stylized. There have been some things. Not things that would have been appropriate for The Witness.

There are attempts in 3D to create painting that are just pretending to be painting without the expressive part of painting.

What I mean by authentic is just that the things you do are really the things you meant to do. They didn't come from emulating something. Emulating is fine as long as you successfully emulate, but I don't think many of these attempts are successful.

When you say artist in the game industry people think you make the pictures. Auteur is a weird word. If only because a lot people use it and don't really know what it means. I just think that game design is an art and I do that. I respect it as an art, and work hard. Programming is also an art if you do it well. When you're a beginner it might be hard to see how.

I'm building a new programming language for example. A lot of the value that I I'm trying to create is the design aesthetic that's going into the programming language. There are all these levels at which art can happen.

All other things are being equal, there's not a difference for the person who sits down and plays the game. The difference is mostly quality of life for the people making the game. But, if I'm happier while making the game, then the game's probably going to come out better, or I'll have more time to think about things.

There is also a craftsmanship feeling of, "I built everything in this cabin." Though software is so big and complex at this point that I don't know that's actually achievable for any sizeable system.