Toward a science of simplicity - George Whitesides

[Music] [Music] [Applause] Most of the talks that you've heard in the last several fabulous days have been from people who have the characteristic that they have thought about something. They are experts; they know what's going on. All of you know about the topic that I'm supposed to talk about: that is, you know what simplicity is; you know what complexity is. The trouble is I don't, and what I'm going to do is share with you my ignorance on this subject.

I want you to read this because we're going to come back to it in a moment. The quote is from the fabled Potter Stewart opinion on pornography. Let me just read it: "The important details here shorthand description hardcore pornography and perhaps I could never succeed in intelligibly defining it but I know it when I see it." I'm going to come back to that in a moment.

So what is simplicity? It's good to start with some examples. A coffee cup—we don't think about coffee cups—but it's much more interesting than one might think. A coffee cup is a device, yes, which has a container, yes, and a handle, yes. The handle enables you to hold it when the container is filled with hot liquid. Yeah, why is that important? Well, it enables you to drink coffee, but also, by the way, the coffee is hot; the liquid is sterile. You're not likely to get colds that way.

So the coffee cup, or the cup with a handle, is one of the tools used by society to maintain public health. Scissors are your clothes; glasses enable you to see things and keep you from being eaten by cheetahs or run down by automobiles. And books are, after all, your education.

But there's another class of simple things which are also very important: simple in function but not at all simple in how they're constructed. The two here are just examples. One is the cell phone, which we use every day, and it rests on a complexity which has some characteristics very different from those that my friend W and BR discuss but are very interesting. The other, of course, is a birth control pill, which in a very simple way fundamentally changed the structure of society by changing the role of women in it by providing to them the opportunity to make reproductive choices.

So there are two ways of thinking about this word, I think, and here I've corrupted the Potter Stewart quotation by saying that we can think about something which spans all the way from scissors to the cell phone, the internet, and birth control pills by saying that they're simple. The functions are simple, and we recognize what that simplicity is when we see it. Or there may be another way of doing it, which is to think about the problem in terms of what if you associate with moral philosophers is called the teapot problem.

The teapot problem I'll pose this way: Suppose you see a teapot, and the teapot is filled with hot water, and you can then ask the question, "Why is the water hot?" That's a simple question. It's like what is simplicity. One answer would be because the kinetic energy of the water molecules is high and they bounce against things rapidly. That's a kind of physical science argument.

A second argument would be because it was sitting on a stove with the flame on; that's a historical argument. A third is that I wanted hot water for tea; that's an intentional argument. And since this is coming from a moral philosopher, the fourth would be that it's part of God's plan for the universe. All of these are possibilities. The point is that you get into trouble when you ask a single question with a single box for an answer, in which that single question actually is many questions with quite different meanings but with the same words.

Asking "What is simplicity?" I think falls in that category. What is the state of science? Interestingly, complexity is very highly evolved. We have a lot of interesting information about what complexity is. Simplicity, for reasons that are a little bit obscure, is almost not pursued, at least in the academic world. We academics—I am an academic—we love complexity. You can write papers about complexity, and the nice thing about complexity is it's fundamentally intractable in many ways, so you're not responsible for outcomes.

Simplicity, all of you really would like your wearing blender in the morning to make whatever a wearing blender does but not explode or play Beethoven. You're not interested in the limits of these things, so what one is interested in has a lot to do with the rewards of the system, and there's a lot of rewards in thinking about complexity and emergence—not so much in thinking about simplicity.

One of the things I want to do is to help you with a very important task, which you may not know that you have very often, which is to understand how to sit next to a physicist at a dinner party and have a conversation. The words that I would like you to focus on are complexity and emergence because these will enable you to start the conversation and then daydream about other things.

All right, what is complexity in this view of things, and what is emergence? We have actually a pretty good working definition of complexity. It is a system, like traffic, which has components. The components interact with one another; these are cars and drivers; they dissipate energy. It turns out that whenever you have that system, weird stuff happens. You in Los Angeles probably know this better than anyone.

Here's another example, which I put up because it's an example of really important current science—you can't possibly read that; it's not intended that you read it—but that's a tiny part of the chemical reactions going on in each of your cells at any given moment, and it's like the traffic that you see. The amazing thing about the cell is that that actually does maintain a fairly stable working relationship with other cells, but we don't know why. Anyone who tells you that we understand life, walk away.

Let me reduce this to the simplest level. We've heard from Bill Gates recently: all of us, to some extent, study this thing called Bill Gates. Terrific; you learn everything you can about that. Then there's another kind of thing that you might study, and you study that hard. That's Bono. This is Bono. But then, if you know everything you can know about those two things and you put them together, what can you say about this combination? The answer is, not a lot. And that's complexity.

Now imagine building that up to a city or to a society, and you've got, obviously, an interesting problem. All right, so let me give you an example of simplicity of a particular kind. I want to introduce a word that I think is very useful, which is stacking. I'm going to use stacking for a kind of simplicity that has the characteristic that it is so simple and so reliable that I can build things with it. I'm going to use simple to mean reliable, predictable, repeatable, and I'm going to use as an example the internet because it's a particularly good example of stacked simplicity.

We call it a complex system, which it is, but it's also something else. The internet starts with mathematics; it starts with binary. If you look at the list of things on the bottom, we are familiar with the Arabic numbers 1 to 10, and so on, in binary. One is 00001; 7 is 0111. The question is, why is binary simpler than Arabic? The answer is simply that if I hold up three fingers, you can count that easily, but if I hold up this—sort of hard to say that I just did seven. The virtue of binary is that it's the simplest possible way of representing numbers; anything else is more complicated. You can catch errors with it; it's unambiguous in its reading. There are lots of good things about binary, so it is very, very simple once you learn how to read it.

Now, if you'd like to represent this zero and one of binary, you need a device. Think of things in your life that are binary. One of them is light switches; they can be on and off—that's binary. Now, wall switches, we all know, fail. But our friends who are condensed matter physicists managed to come up with, some 50 years ago, a very nice device shown under that bell jar, which is a transistor. A transistor is nothing more than a wall switch; it turns things on and off, but it does so without moving parts, and it doesn't fail basically for a very long period of time.

So the second layer of simplicity was the transistor in the internet. Since the transistor is so simple, you can put lots of them together, and you put lots of them together, and you come up with something called integrated circuits. A current integrated circuit might have, in each one of these chips, something like a billion transistors, all of which have to work perfectly every time. So that's the next layer of simplicity, and in fact, integrated circuits are really simple in the sense that they, in general, work really well.

With integrated circuits, you can build cell phones. You all are accustomed to having your cell phones work the large majority of the time. In Boston, Boston is a little bit like Namibia in its cell phone coverage, so that we're not accustomed to that all the time, but some of the time. But in fact, if you have cell phones, you can now go to this nice lady who's somewhere like Namibia and who is extremely happy with the fact that although she does not have a master's degree in electrical engineering from MIT, she's nonetheless able to hack her cell phone to get power in some funny way. And from that comes the internet.

This is a map of bit flows across the continent. The two blobs that are light in the middle there are the United States and Europe. And then back to simplicity again. So here we have what I think is one of the great ideas, which is Google, which in this simple portal makes the claim that it makes accessible all of the world's information. But the point is that that extraordinarily simple idea rests on layers of simplicity, each compounded into a complexity that is itself simple in the sense that it's completely reliable.

All right, let me then finish off with four general statements, an example, and two aphorisms. The characteristics which I think are useful to think about for simple things: first, they are predictable; their behavior is predictable. Now one of the nice characteristics of simple things is you know what it's going to do in general. So simplicity and predictability are characteristics of simple things.

The second is—and this is a real-world statement—they're cheap. If you have things that are cheap enough, people will find uses for them even if they seem very primitive. So, for example, stones—you can build cathedrals out of stones; you just have to know what it does. You carve them in blocks, and then you pile them on top of one another, and they support weight. So there has to be fun—the function has to be predictable, and the cost has to be low. What that means is that you have to have a high performance or value for cost.

And then I would propose as this last component that they serve, or they have the potential to serve, as building blocks. That is, you can stack them, and stacking can mean this way, or it can mean this way, or it can mean in some arbitrary n-dimensional space. But if you have something that has a function and it's really cheap, people will find new ways of putting it together to make new things. Cheap, functional, reliable things unleash the creativity of people, who then build stuff that you could not imagine. There is no way of predicting the internet based on the first transistor; it just is not possible.

So these are the components. Now, the example is something that I want to give you from the work that we ourselves do. We are very interested in delivering health care in the developing world, and one of the things that we wish to do in this particular business is to find a way of doing medical diagnosis at as close to zero cost as we can manage.

So how does one do that? This is a world in which there's no electricity, there's no money, there's no medical competence, and I don't want to spend your time in going through the details. But in the lower right hand corner, you see an example of the kind of thing that we have. It's a little paper chip; it has a few things printed on it using the same technology that you use for making comic books, which was the inspiration for this particular idea.

You put a drop, in this case of urine, at the bottom; it wicks its way up into these little branches—you know, no power required. It turns colors; this particular case you're reading kidney function. And since the healthcare worker of much of this part of the world is an 18-year-old with an AK-47 who happens to be out of work and is willing to go around and do this sort of thing, he can take a picture of it with his cell phone, send the picture back to where there is a doctor, and the doctor can look at it.

So what you've done is to take a technology which is available everywhere, make a device which is extremely cheap, and make it in such a fashion that it is very, very reliable. If we can pull this off, if we can build more function, it will be stackable. That is to say, if we can make the basic technology of one or two things work, it will be applicable to a very, very large variety of human conditions and hence extendable in both vertical and horizontal directions.

Part of my interest in this, I have to say, is that I would like to—how do I put this politely?—change the way, or maybe eviscerate the capital structure of the US healthcare system, which I think is fundamentally broken. So let me close. Let me close with my two aphorisms. One of them is from Mr. Einstein, and he says, "Everything should be made as simple as possible, but not simpler." I think that's a very good way of thinking about the problem. If you take too much out of something that's simple, you lose function. You have to have low cost, but you also have to have a function, so you can't make it too simple.

The second is a design issue, and it's not directly relevant, but it's a nice statement. This is by S. Zupe, and he says, "You know you've achieved perfection in design not when you have nothing more to add, but when you have nothing more to take away." And that certainly is going in the right direction.

So what I think one can begin to do with this kind of cut at the word simplicity, which doesn't cover Branzi, it doesn't answer the question of why it's better or worse or simpler or less simple than van, and certainly doesn't address the question of whether Mozart is simpler than B, but it does make a point which is one which, in a sense, differentiates the real world of people who make things and the world of people who think about things.

Which is there is an intellectual merit to asking how do we make things as simple as we can, as cheap as we can, as functional as we can, and as freely interconnectable as we can. If we make that kind of simplicity in our technology and then give it to you guys, you can go off and do all kinds of fabulous things with it. Thank you very much.

A quick question: So can you picture that a science of simplicity might get to the point where you could look out at various systems, say a financial system or a legal system, health system, and say that has got to the point of danger or dysfunctionality for the following reasons, and this is how we might simplify it?

Yes, I think you could because if you look at the components from which the system is made and examine their fragility or their stability, you can probably build a kind of risk assessment based on that basis.

Have you started to do that? I mean, with the health system, you've got a sort of radical solution on the cost side, but in terms of the system itself—

Well, no. How do I put that simply? No. That was a simple, powerful answer.

Yes. In terms of that diagnostic technology that you've got, where is that, and when do you see that maybe getting rolled out to scale?

That's coming out soon. I mean, the systems work, and we have to find out how to manufacture them and do things of this kind, but the basic technology works.

You've got a company set up?

To a foundation.

A foundation, not for profit?

All right. Well, thank you so much for your talk. Thank you.

What does a machine know about itself? Can it know when it needs to be repaired and when it doesn't? In industries like manufacturing and energy, they're using predictive analytics to detect signs of trouble, helping some companies save millions on maintenance because machines seek help before they're broken and don't when they're not.

That's what I'm working on. I'm an IBM-er. Let's build a smarter planet.