DeconstructSeattle, WA - Thu & Fri, Apr 23-24 2020

← Back to 2017 talks

Transcript

(Editor's note: transcripts don't do talks justice. This transcript is useful for searching and reference, but we recommend watching the video rather than reading the transcript alone! For a reader of typical speed, reading this will take 15% less time than watching the video, but you'll miss out on body language and the speaker's slides!)

OK, so my talk is called Idee Fixe. It's the French for the fixed idea. And by fixed idea, I mean this notion that an idea that you have, but the idea is a kind of blockage. It prevents you from making progress. In fact, it's an idea that you have that's like a wall. It's in front of you, and it actually obscures your ability to see alternatives.

This is something as software developers you're probably intimately aware of. You've probably sat down to fix a bug at some point. And what happened is that you sat down and hours later, you still haven't solved it. And you sat down, and it's because you sat down with an idea of what you thought was wrong, and that idea prevents you from seeing what's actually wrong.

It's also the case that you've probably had the experience of you've wasted six hours on the bug, and you still can't see the answer. And you find a colleague, and say, look, I'm just going to describe this to you. Just listen for a second. And you spend a minute talking about what you think the problem is, and the person maybe asks you an interesting question, and then, like a flash of light, you're like, oh my God. I can't believe it. This whole time I had this assumption about what I thought was wrong, and it prevented me from seeing what was actually wrong.

And there's something real interesting about how you see through the fixed idea. This discursive element is extremely important, whether that's talking to a colleague, or sometimes you just say it out loud, you talk to yourself. Sometimes that's all you need. There's a sort of stepping outside of yourself, whether it's talking to somebody else, or verbalizing it to yourself, that's important in working through the fixed idea.

But fixed ideas aren't just about our individual experiences. I mean, fixed ideas scale. They scale all the way up. And certainly I would say, our tech industry is filled with many, many fixed ideas. And not all fixed ideas are bad, but I would say the tech industry tends have a preponderance of negative fixed ideas. It's a rat's nest. We're going to talk a bit about that.

So why I'm interested in this topic. I work for a software consultancy called Cognitech, and what we do is we use non-mainstream technologies. We work with clients. We believe that our non-mainstream approach is better. We maintain closure. Closure is a functional dynamic list for the Java virtual machine. And I'm the lead maintainer of Closure script, which is a compiler for closure to JavaScript. And so, I would say the last four or five years of my life has been convincing people that they have fixed notions about how software can be constructed. And a lot of my talks are about pushing people through, and getting them to see around all this wall that they put in front of them, in front of themselves.

My talk is not a happy talk. There's been lots of cool talks and you laugh, and have good feelings. My talk is actually a negative talk. It's a pessimistic talk. I would say my talk is an alternate or an updated version of this talk. CP Snow give a talk in 1959 at Cambridge called The Two Cultures in the Scientific Revolution, and he was lamenting the fact that in the 20th century, you were seeing a rapidly increasing divide between two important aspects of the humanities, which one is science and one is art. In the 20th century, art was becoming more rarefied and specialized, and science was becoming more rarefied and specialized.

And really, it was getting to the point where there was if not just disinterest, outright suspicion or antagonism between the two disciplines. CP Snow felt this very keenly because he was a writer of fiction as well as being a scientist. So, of course he gave this lecture, and a lot of scientists hated it. Because they either didn't believe him, or they were like, why does this matter? And I think it matters a lot, and I'll try to make the case for that. But I would argue that it basically happened. I mean, it's over.

My experience is that I went to high school in a typical American suburban city, Arlington, Texas. And I entered high school with a deep love of drawing. At one point I wanted to be an animator. So I loved visual art, and I also excelled at math and science. I was there for four years in high school, doing advanced math and science and art classes. And not once did I ever have anybody from my advanced math and science classes in my art classes. Everybody that was my peer in my art classes were quote unquote regular, or quote unquote, remedial.

And to me there's a real problem here, because when you do art, the cool thing about art, the power of art is that it's largely a subjective experience. I mean, what score are you going to put on an artwork. It's either something that people feel something about, or they don't. And everybody has a different kind of response to different artworks. Also, when you do art, when you're making a drawing, a lot of artists have to deal with failure in a very intimate way. Most of the drawings that I would make when I was young, I'd be like, ah, most of these aren't good. Occasionally, you'd make a good drawing, and that would be enough to keep you going. So you had a sort of intimate relationship with failure.

But if you were a math and science student, there's an absolute metric. You're either passing your test and maintaining your 4.0 so that you can get into the schools you want, or you're not. So you start from a very young age, if you excel in the math and sciences, you end up being trained to have a very different relationship to failure. And I actually think that is what leads to both misguided Utopian visions, and uncritical positivism. And really what happens is you see this sort of regressive take on failure. So when somebody says something like 10x programmer or ninja programmer, or rock star programmer, you can only say that if your idea of failure is regressive. It's the regressive notion of failure that creates regressive notions of success.

So how do we move past that? How do we get around this, as it's one thing to fix a bug, it's another thing to move tech culture forward. And there's an obvious way, and it's been known for a very long time, which is simply that you have to have a different perspective. You have to have a conversation. You have to have a conversation with people with different opinions, with different viewpoints, with different backgrounds, with different training. Alan Kay knew this quite well, and he has lots of interesting quips, and I really liked this one. This is one that gets often quoted. "A change of perspective is worth 80 IQ points." And I would argue these days, it's worth way more than that.

And certainly, one path to avoiding these kinds of situations is to have a true effort towards diversifying your community. But there's also another inspiring way to work through fixed ideas, and dismantle them, and it's not specific to art or science. I'm picking this quote because I think Georges Perec said it quite nicely. Georges Perec was an author of fiction. He wrote this really fantastic book which I read called, [SPEAKING FRENCH] Life: A User's Manual. He also wrote an interesting book called A Disappearance, which is a book without the letter E, which is pretty challenging to do in French.

And he had a take on his work which I think is relevant, regardless of whether your perspective is artistic or scientific. A theme throughout his work was this notion of the infra-ordinary. And he talks about the way that he would work, he would say, force yourself to write down what is of no interest. What is most obvious, most common, most colorless, make an inventory of your pockets, of your bag, question your teaspoons. And this sounds absurd, but we just said that it's often the thing that you can't see anymore that may point to some fixed idea that you might have.

There's a really beautiful scientific version of this precisely happening, and that's this story about Richard Feynman, the physicist. And I really liked this story, because it was about him experiencing a burnout with physics, because he was a very talented physicist, but all of his talents and efforts were spent on building the nuclear bomb. More effective ways to kill people. That's probably going to damage your enjoyment of something that you used to love. And in the story he talks about how he came back to physics. He learned how to love physics again, and it was by looking at the most ordinary of things. He was at Cornell, and he saw a student throw a plate in the air, and he noticed that the rate of the wobble of the plate was different from the rate at which the insignia of Cornell was spinning around, and he wanted to understand this relationship. And he started trying to figure out the math of that. And if a friend or colleague of his saw this and said, why are you doing this? There's no point. This seems so ordinary.

Of course, this research ended up getting Richard Feynman the Nobel Prize in physics. But this applies just as much to art. So I actually have no formal training in computer science, and never studied computer science. Computers were just a fun thing I did as a kid. My formal training when I went to college was in film. And I really enjoyed film, and film is great. When you go to school, you pick a field. And what they do is they pour a bunch of fixed ideas into your mind, and it's no different for film. Narrative cinema has a very specific structure. I went to school and you learn about lighting, directing, acting, blocking. All this stuff so that you too can construct a film that everybody else will say, that's just like a movie. A movie you would see in Hollywood. It's identifiably a movie.

But the cool thing about education is when a really great education says, OK, there's this main line of thought, but there are alternatives. And in fact, knowing about the alternatives will only enrich your understanding of the main line of thought. So I actually ended up really getting into and excited about experimental cinema. This is a film still from a film by Michael Snow called "Back and Forth." It was made in 1969, and he used the most ordinary and simplest of techniques. It's just a camera in a room, and he simply panned the camera like this for 30 minutes. That's all it is. And then for the last 20 minutes of the film-- it's a 50 minute long film-- It's the camera panning up like this. That's all it is.

But as a film student, it's shocking. Everything that I thought movies were, I here see a counterexample. Movies can be whatever you want. You can film completely ordinary things, and you can still have an incredible cinematic experience. Michael Snow, I once saw him give a talk. And he was talking about what would happen at some of his screenings. And I can't remember if it was this particular film, but in one of his films, someone in the audience got up, broke into the projection room, got into a wrestling match with the projectionists, because he wanted to rip the film off the projector. And I mean, it's funny, but actually, it shows something. It shows that fixed ideas often have an undue amount of power. And in fact, there's benefit. Somebody is often benefiting from the propagation of fixed ideas. And it's often the case it's not you or I, it's somebody else, whose intentions may not be very nice.

You've probably never heard of this person. This person is an ordinary person. I don't believe she was involved in any direct way in civil rights. This is LBJ's cook, Zephyr Wright. And it's hard to find now, but there was a really great interview with her that was done, I believe, in the '70s. And you can probably find it in a library. But the ordinary experience for her growing up in the United States was that she would go to high school, and then she would just go get a job. It turned out that she was actually really good at high school, and one of her teachers were like, well you're really good, so why don't you decide to go to college?

And so this was a stunning moment for her, because in her reality, completely normal for her is nobody does that. Nobody goes to college. And so she went to college, she was stunned by what she had learned. So many of the things that she took for granted, this is the way things are. Her ordinary experience was actually horrifying. That everything that she thought was normal was this edifice of evil that the United States had constructed.

So tech culture is just a reflection of everything that's bad. These things are already bad. But I find it funny because for a community that prides itself on originality of thought, it's fascinating to me how much technology is just a mirror of everything that's already wrong. And this problem is recursive. Certainly the diversity issue must be addressed, but even from the perspective of an engineer, down for engineering decisions you see this problem. I spent most of my time in my programming career being really excited about UI work, user interface work. And because there was a synergy for me, or a connection between doing cinema, which was creating an experience for somebody, and doing UI work, you're creating experience. There's some system, and this is your contact with the audience, the user interface.

And when I started out, everybody was doing NBC. NBC was the thing. That's just how it was done. It was the right way. Later, at some point, I built enough of these things where I was like, OK, NBC is kind of cool, but there's definitely some issues. And at some point, I got curious and I went back and looked at the history of NBC. And it turns out NBC was a memo. It was just a memo that they passed around at Xerox PARC. This was not the tablets coming down from the mountain. Somebody was like, this seems like a reasonable idea, if probably flawed, and we'll probably have better ones. And somehow, a memo got picked up, and it was just like, this is how it's done, and this is how will always be done. For 30 years that's how people did UI work, even with all of its flaws.

And sometimes you're lucky in technology to see a transition. And there's an interesting transitioning happening right now thanks to React. React was created by a very clever developer who is familiar with functional techniques, and they just tried it. They just tried it, and it turned out to be pretty cool. But even when I first saw this, I was also? Resistant. When it came out-- you know we all suffer from fixed ideas. I was like, come on, x-amount literals in a JavaScript file? No way. Not going to do it.

But fortunately, I had a friend who had a different opinion. My friend Brandon Bloom. He was like, look, I'm going to arrange a lunch with you and Peter Hunt-- Peter Hunt at the time was working on React-- and we'll talk about this. And we had a conversation. And at the end the lunch, I was like, OK, this is worth thinking about. This does seem interesting. And I ended up writing a blog post about it which was immensely popular, and probably helped out a little bit. But what's funny to me is not that this changed the way that people thought of a UI, for me it was interesting to see how quickly people went from NBC as like, this is how you do it, to being like React, this is how you do it. The path to cargo-culting is so short. And who knows how long we're going to get stuck with this, but I do believe, having done NBC, yeah, there were things that are wrong with it, but there are definitely properties that NBC had that were nice. And React has some interesting properties, and probably there's a thing beyond both these two things that could capture the things we liked about both systems. We shall see.

I'm actually almost done. So I really enjoyed Zach's talk, and there's a little bit of a connection here with this talk, which is that Google Deep Mind, they did this thing called AlphaGo, you've probably heard of it. Fascinating project because they accomplished something that people thought was still about 20 years away, using a lot of innovative techniques around deep learning. I played Go, I enjoy it. I'm not that good, but I enjoy it. But I knew it well enough to follow along with the game. And the thing that I really was surprised by was how many professional players-- professional meaning lives in China, Japan, and Korea-- but their commentary, they were shocked. Go has been developed for at least 3,000 years, if not 4,000 years. And there are all these proverbs, all these ways of playing that they believed to be the correct way. And AlphaGo showed them that there was other ways to think about the game.

That these fixed ideas that they developed for thousands of years, there was other ways. Other ways, other alternatives. And that was really cool. This is the one game that Lee Sedol won. Lee Sedol is one of the top players in the world. And this was a shocking move. This move 78, the commentators said, this is a divine move. So in Go culture, the divine move is something that only one person maybe gets to play once in 100 years. One player gets lucky. It's like the godly move, the move that changes the game entirely. And this move actually did completely change the game. It turns out that there was a refutation. AlphaGo couldn't find it. Why? Because AlphaGo gives each move that is going to consider a probability. What is the probability that my opponent will play that move? And this move, AlphaGo said, there's only a one in 10,000 chance, so I'm not going to spend any time examining it. And it turned out that because it didn't examine it, it actually made a bunch of mistakes and quickly lost. This was the winning move.

And Deep Mind learned a lot from this. And they're building and they will present next month in Wuzhen in China, playing the top Chinese players. They're going to present a new version of AlphaGo, which is trained by having an anti-AlphaGo. An AlphaGo f who actually considers these low probability moves to make sure it doesn't miss something. But this goes right back to the very first thing I said, the necessity to have somebody with a different perspective, a different viewpoint. And I think Deep Mind's really cool. I think what they work on is really fantastic. They seem to demonstrate quite a bit of rigor.

However, I'm pessimistic. This kicked off the AI gold rush. I work for Cognitech, but I'm now working for an AI startup, and there's so many of these AI startups. There's an AI gold rush happening. People think they're going to be able to convert this stuff into money. OK, maybe. But the problem is, we as software people, come on. The likelihood that we'll have the insight and the ethical compass to do the right thing consistently, it's highly unlikely. I mean, there's so many stories that are coming out right now about how unethical so many companies in the software industry are. Chances are many, many, many mistakes will be made, and it's actually something we should be afraid of. Because we already know as humans, we have this issue with fixed ideas. Are we simply going to build systems that again, are just a mirror? They're just a mirror of our own mistakes.

So there is no positive note to end on in my talk. So nothing happy, but I hope that I have armed you with some pessimism about the work that you do. And I really truly think that it is important, the sort of artistic stance, which is that to be critical of human activity, to look at things that on the surface appear normal, but maybe there is a deeper truth there, and it's not a good thing. And maybe we should dismantle it, and try to make progress. Anyways, that's it. Thank you.

[APPLAUSE]