Drew Endy [2.17.08]

ED. NOTE: A theme appears to be evolving, beginning with the Edge event "Life: what A Concept!" in August, proceeding to Munich at DLD (Hubert Burda's Digital, Life, Design ) in January, where Craig Venter, and Richard Dawkins held an Edge conversation, "Life" A Gene-Centric View". Both events were important, and newsworthy. Next, the following conversation, Engineering Biology", with Drew Endy, a young researcher who is defining the cutting edge of synthetic biology.


The only thing that hasn't been engineered are the living things, ourselves. Again, what's the consequence of doing that at scale? Biotechnology is 30 years old; it's a young adult. Most of the work is still to come, but how do we actually do it? Let's not talk about it, let's actually go do it, and then let's deal with the consequences in terms of how this is going to change ourselves, how the biosecurity framework needs to recognize that it's not going to be nation-state driven work necessarily, how an ownership sharing and innovation framework needs to be developed that moves beyond patent-based intellectual property and recognizes that the information defining the genetic material's going to be more important than the stuff itself and so you might transition away from patents to copyright and so on and so forth. 

DREW ENDY, is Assistant Professor of Biological Engineering at MIT, where he is working to enable the design and construction of large scale integrated biological systems, and to develop and improve general methods for representing cellular behavior.

Drew Endy's Edge Bio Page


[DREW ENDY:] How can I make biology easy to engineer? I was born in 1970. Going back hundreds of years, people had imagined that you could always design and build or make life, but nobody could really do that much about it. As the 1970s rolled out, human beings invented a lot of technology, with recombinant DNA for cutting and pasting pre-existing fragments of genetic material, and polymerase chain reaction, which was invented in the '70s, but was not really figured out until the '80s. And then automatic sequencing right here in this town, with Fred Sanger in 1977.

Now, 30 years after the initial successes of biotechnology, it has only realized really one of the early promises. The early promises were, first making therapeutics via recombinant organisms, producing drugs like insulin via bacteria , which has worked. Gene therapy was the second promise, fixing genetic defects by patching our DNA programs, and this has not yet worked. And third, to develop crops that can fix nitrogen so that agriculture wouldn't have to rely on synthetic fertilizers. That hasn't worked either. So of the three great first promises I've been told rolled out with the beginning of genetic engineering, we have realized one of them.

Nevertheless, biotechnology exists, it's a huge positive contributor to our health and economy and the human condition generally, and now it's 2008. And so the question is, can we realize the initial promise of biotechnology? Or, forget that question, how do we make biology easy to engineer, so that anything we might want to manufacture out of the living world is something that we can pull off?

For reference, I'm in the Biological Engineering Department at MIT, which politically is a new department; it's only a couple of years old. A consequence of having a department is we have to teach undergraduates. Imagine that you are 15, you're an ambitious youngster, or 17 or 18 years old and you're showing up as a first year undergraduate at MIT, and you're choosing what to major in. Well, you could choose to major in biology or electrical engineering or computer science or ... oh, now you can major in biological engineering. What would you expect to learn? What would you expect of your faculty colleagues, your professors? What would they be able to teach you?

You look to your friends, who are going to study electrical engineering, and they can learn how to design and build computers, or write computer programs, and the objects that they make don't have emergent properties unless that's what's intended, instead they behave as expected. Then you look at biological engineering and you say, well, yes, I'd like to design and build living organisms, or program DNA to execute genetic programs that behave as expected. But, nobody can teach you how to do that.

Thirty years into biotechnology, despite all of the successes and attention and hype, we still are inept when it comes to engineering the living world. We haven't scratched the surface of it, and so the big question for me is, how do we make biology easy to engineer? For comparison, if you look at other examples of technology, there are many of them. Take modern electronics, during and following World War Two, people are building computers. Von Neumann is building a nice machine in the basement of the Institute for Advanced Study at Princeton. The official purpose of this machine is to design hydrogen bombs and compute the trajectories of munitions. And he of course is apparently running artificial life programs on it, because that's what he's more interested in Let's say it was 1950. The Apple One, the personal computer is only 25 years later.

We're now thirty years into biotechnology. Are we going to ever get to the point where it's not an exclusive technology, it's not a technology that requires experts? Are we ever going to get to the point where we can make many component integrated systems? Are we going to ever get to the point where we have separation of the types of work in biological engineering, so that one person might be an expert designer, another person might be an expert constructor, as we have expert architects and builders and what not?

I had an interesting day at the New School a couple of years ago, interacting with the students and faculty there, and it became apparent that that institution, which supports things like the Parsons School of Design, if I remember correctly, depends a lot on foundational technologies that make the work of design easier. And one of the questions that emerged from the day there was, should Parsons have a genetic design program? Should the New School have a genetic design lab?

The answer is, yes, absolutely, in the same way that graphic art depends on all sorts of tools, and even painting depends on ready-made paints, which are, in their modern form not more than a hundred years old, yet allows painters to start painting without first becoming expert in mixing up all the different colors themselves, you get the paint pre-positioned in a foil tube.

The big question, to come back to it, is, how do we make biology easy to engineer, and then the parallel question that comes along with that is, what are the consequences of success? If you look around the room that we're in, everything in the room is a synthetic or engineered artifact, right? From this stuff, to the wood itself, the materials here, even the air that we're breathing, has been engineered for temperature and humidity, so that it is easier for us to deal with.

The only thing that hasn't been engineered are the living things, ourselves. Again, what's the consequence of doing that at scale? Biotechnology is 30 years old; , it's a young adult. Most of the work is still to come, but how do we actually do it? Let's not talk about it, let's actually go do it, and then let's deal with the consequences in terms of how this is going to change ourselves, how the biosecurity framework needs to recognize that it's not going to be nation-state driven work necessarily, how an ownership sharing and innovation framework needs to be developed that moves beyond patent-based intellectual property and recognizes that the information defining the genetic material's going to be more important than the stuff itself and so you might transition away from patents to copyright and so on and so forth.

So to zoom out, how to make biology easy to engineer? I don't want to talk about it, I want to do it. And, how do we do this in a way that leads to constructive culture around the technologies that's overwhelmingly positive in terms of the consequences of its being rolled out?

What happens when the technology in support of engineering biology is sufficiently advanced that somebody like Stefan Sagmeister, the graphic designer, could sit down and design a life form that he would consider interesting or beautiful? How do we get from what we've got today, where we're basically celebrating a bunch of stunts, we've only delivered on 30 percent of the initial set of promises of biotechnology, and there are so many other things we can imagine that are fantastical because they're just too complicated given the current state of affairs, how do we get from that to ... yeah, the graphic designers are making beautiful living objects? Working through that, from a human perspective as well as a technology perspective, is pretty interesting.

There is also the issue of addressing the energy needs. A lot of people drive investments in biotechnology from the application side, and that's good. There are lots of pressing human needs and problems. Food, which is an energy of sorts for people and animals. Liquid fuels for cars and jets, and then you've got health and medicine, and then you've got environmental issues, and then you've got materials construction, and ta-da-da. What's interesting about biotech is that the applications have always been so unbelievably pressing.

So, let's wind the clock back 30 years… to a first approximation there's been an under-investment in tools, right? If you're running a team that's trying to figure out how to make insulin in bacteria, or how to make artemisinic acid for treating malaria in bacteria or yeast, and somebody says, hey, why don't you take five percent of your project's budget and instead of spending it on delivering your product as fast as you possibly can, leave behind a little bit of legacy engineering infrastructure so that the next time you go to do a project like this, it doesn't cost you $40 million? So that the next time you go do a project like this, it's much, much easier. The arguments then that take place in response to those sorts of suggestions are, if we delay shipping our product by a day, we will lose to our competitor, or 10,000 additional people will die, or something. On a short time scale it's impossible to argue against these positions, right?

But if you take a longer view on it, in the absence of making such foundational investments into technologies that support the engineering of biology, the engineering of biology's always going to be hard. We have to figure out how to solve that problem. This comes into play when you think about energy. What do I think about biological production of energy? Terrific, right? Yeah; we'd rather not be burning dinosaur juice. It seems like important work, and the lab next to mine when I was earning my Ph.D. was working on cellulosic ethanol. If the price of oil would go up by a factor of two, cellulosic ethanol would be cost competitive. That was 1994.

And I hope bioenergy succeeds. But consider that there was a trap for John von Neumann when he was building those early computers to compute the trajectories of munitions… it turns out that the actual utility of computers is, you know, much more than we could have imagined, much more than the military applications and accounting databases. So that was just stupid, right? But, with maybe a few exceptions, nobody back then had a clue about what these applications would be.

Thus, I'm not interested in pursuing any one application in biotechnology right now, because I want them all to come true, and I want them all to come true on a time scale that's relevant to me. I can be very direct and selfish about this. And the only way that's ever going to happen is if I don’t go work on bio-energy—there are enough people who are going to go work on that because it's a problem everybody could understand; you'll be able to raise resources around it and go do the work—there's the complementary problem, the meta-level problem, which is, let's go make all of biotechnology easier for everybody.

The underlying goal of synthetic biology is to make biology easy to engineer. What does that mean? It means that when I want to go build some new biotechnology, whether it makes a food that I can eat or a bio-fuel that I can use in my vehicle, or I have some disease I want to try and cure, I don't want that project to be a research project. I want it to be an engineering project. In the science of biology, the people you're talking to are scientists, they're not engineers, and—not to be arrogant, just to be an observationalist—the question is, if you're an engineer looking at biotechnology, what do you need to do in order to make it easy to engineer? That's what synthetic biology is about.

As soon as you start talking about engineering, and you could start talking about very real historical examples of what engineers do when faced with situations of this sort—America, 1860, machinists are building objects, steam engines, what have you. Nuts and bolts that hold together machines are specific to the particular machine shop that manufactures them. What that means is, if you buy some machine from a machine shop in Newark, New Jersey, and it breaks down in Chicago, you have to send it back probably to that specific machine shop, with the machines set to tool things on a particular set of designs in order to get the replacement part or to get the thing fixed.

In April of 1864, somebody says, enough; William Sellers of the Franklin Institute in Philadelphia gives a paper on a system for nuts and bolts. And he proposes the Sellers Screw Thread Standard, which is a 60 degree angle squared off at the top screw thread design, easier to manufacture than the English Whitworth Standard of a 55 degree angle, rounded screw thread. As a result, eventually everybody in the U.S. retools their machine shops to produce screws, nuts and bolts, in accordance with the Sellers Standard.

The consequence of this today is, when I go to the hardware store and get a nut and a bolt, so long as they don't screw up the English / metric thing, I can take those two objects, and I can put them together. I don't have to do an experiment. I don't need to go talk to some Harvard professor to figure this out. I don't need to do a controlled experiment to see if my first experiment worked. I just get the two objects and put them together.

That's an example of what an engineer would call reliable physical composition. Take two objects and put them together. The other thing that happens is when you have the nut and the bolt together as a composite object, when you pull on the nut, it stays put. It doesn't come flying off. The composite object has the expected behavior, it doesn't have some emergent property. That's reliable functional composition. The function of the two things when you put them together is what you expect. What's amazing is that I've lived taking this standard for granted my whole life. Even though I have three engineering degrees, I didn't know about this until a couple of years ago when Tom Knight of MIT pointed out to me that it would be nice if we had standard biological parts that could snap together and then behave as expected when we snapped them together.

George Church has been exposed to this, but it's not of his mother culture. He's a geneticist; he's reverse engineering natural biological complexity. That's a great thing to be doing. Engineers hate complexity. I hate emergent properties. I like simplicity. I don't want the plane I take tomorrow to have some emergent property while it's flying. If you look at the science of genetics, which has been in the business of trying to figure out the relevant information in coded DNA, the most important thing that has impacted the science of genetics has been technology. If you look at the science of genetics, you can look at it before DNA sequencing existed. People would find mutations and they'd map them, in the process, to different loci or regions on the DNA. And the mathematics that were being used was ) based on simple logic. Then, a number of great people drove forward with the sequencing of DNA. And as a result of that technology, we can now read DNA, and that technology continues to get better.

It is important to put the impact of advances in DNA sequencing technology in context. In 1990, nobody had sequenced anything, except for a couple of bacterial viruses, and maybe some other viruses. In 1995, the first bacterial genome, haemophilus influenza is sequenced. In 2001, there is a draft of the human genome sequence. How did we, in the 1990s, go from, we stink at sequencing DNA to, yeah, we just sequenced human beings, and now, only seven years later, the personal genome projects are  coming online? It's not because George Church and Craig Venter and Eric Lander and Francis Collins got ten billion times smarter during the Clinton years. It's because the technology for sequencing DNA got automated and scaled up sufficiently to do it.

The impact of underlying technologies on what's possible is an incredibly important thing to recognize, and it's something that, if successful, you want to be able to ignore in the same way that I want to be able to ignore nuts and bolts and screw thread standards. Genetics changed in response to sequencing technology. You could read DNA. We have no idea what it says. The mathematics is now pattern recognition to try and look at many sequences and find the conserved patterns that might have relevant functions. Synthesis technology is coming online next.

2008 is our 1995, if you will; this is the year where a bacterial genome's been synthesized from scratch. Ahead of that work, chloroplast genomes, mitochondrial genomes, have been constructed ; in fact a project from Japan a couple of years ago made a ten million base pair fragment of DNA from existing fragments, which is 15 times larger than anything getting attention these days.

So, what happens to the science of genetics as a new set of tools come online that let us build whatever DNA molecule we want, you get to make changes and see what happens. Instead of being called genetics, this is called reverse genetics, and the mathematics driving this is probably going to be perturbation design. What changes do you want to make, and how do you choose what to make?. So, first genetics goes from pre-sequencing technology and it's based on logic. Then it's post-sequencing and it's pattern recognition. And next there's going to be post-synthesis genetics, and it's going to be, make whatever you want. Perturbation design becomes the mathematics. And the whole field's going to change.

 When sequencing technology was developed, the scientific community, not to mention the rest of the world, did an incredibly poor job of anticipating that there would be this resulting challenge of, what the heck does all this DNA sequence information mean? How big is the pattern recognition problem? Fields of science like bio-informatics are purely reactionary, and have poorly planned responses to technology advances, and we're going to get the same thing again with synthesis.

For example, how do you manage the information going into a DNA synthesizer so that you can construct some useful object that'll help you do genetics? This is the reverse bio-informatics problem. I think George Church and Craig Venter have a lot to contribute to it, which will be terrific. It will be part of synthetic biology, but it will be synthetic biology impacting science, which is the worst case scenario for synthetic biology. We fail to actually deliver any useful artifacts that people want, but at least we'll fail, and we'll de-bug our failures, which will prioritize our misunderstandings of biology much more ruthlessly than anything else, and which is much better than a NIH study section.

What else might happen?  I've got a invitation to give a talk at the Chaos Communication Congress, which is the largest hacker meeting in Europe, about 4,000 people. People who like to make stuff, people who like to understand how things work. And they're very interested in learning how to program DNA, and how DNA works. One consequence of actually making biology easier to engineer, whether you're standardizing the components or figuring out how to develop higher level programming language is that other people besides the usual suspects are going to have access to the technology.

If you think about what happened between 1950 and 1975, when you went from  von Neumann's machine to the Apple One, a key parts of this transition was that folks were so stoked about computing, and were so fed up with limited access to centralized computing resources, that they went out and built their own computers—by definition, the personal computer. As a result, today we have a worldwide community of folks who are excited about building electronics and writing software, which includes school kids, professionals, big companies, small companies, governments, you name it, a very diverse ecology around that technology.

Programming DNA is more cool, it's more appealing, it's more powerful than silicon. You have an actual living, reproducing machine; it's nanotechnology that works. It's not some Drexlarian (Eric Drexler) fantasy. And we get to program it. And it's actually a pretty cheap technology. You don't need a FAB Lab like you need for silicon wafers. You grow some stuff up in sugar water with a little bit of nutrients. My read on the world is that there is tremendous pressure that's just started to be revealed around what heretofore has been extraordinarily limited access to biotechnology.

Take some of the writings of Freeman Dyson, he's imagining genetic engineers of the future winning the Philadelphia Flower Show, the San Diego Reptile Show or whatever it is. The question is, how do you get there? And as you start working through that path of getting there, and of trying to solve the problems, not for yourself, or not for bio-energy, but for everything, what you find is that there are vast communities of people who want to be doing this.

Perhaps the reactions being stimulated today around this technology are a direct result of the fact that the people who are promoting the technology in this way tend to favor exclusive ownership, limited access, and present themselves as God-like creators; as opposed to, we're constructing things, we could use your help, anything we do today is going to pale in comparison to what's coming, so let's figure out how to work together on this.

As a different example, in 2003 I taught a course at MIT, the Synthetic Biology Lab with some colleagues, and we had 16 students. For the last four years this course has been doubling every year, and it's now taught independently at about 60 schools in 30 or 40 countries worldwide, it's called IGEM, the International Genetically Engineered Machines competition. There are teams of teenagers from Germany programming DNA happily there, as well as Australia, Russia, Japan, China. The competition was won by the team from Peking University this year, and six or seven hundred students participated.

How do you recognize this exponential and serve it and bring more people to participate in it? The rewards of doing this are greater than any one group’s project.  For instance, the team from Melbourne, Australia showed up with a 6,000 base pair fragment of DNA that they found, which somehow, I don't know how this actually works, folds up ... the proteins get made and the proteins self-assemble into a 50 nanometer, very tiny, sphere that is filled up with gas. The protein shell is somehow gas-impermeable, and these little balloons, these protein balloons, get booted up inside the cytoplasm, the insides of cells, and you can control how many different balloons there are. Depending on the number of balloons, the cells will either float or sink or be neutral.

Who knew? I didn't know anything about this biology, and they showed up, they made this standard biological part, such that we can now snap it together with the 2,000 other parts that we've got in our collection so far, which is a free collection. We shipped over a hundred thousand parts around the world last year, for free, and the collection's doubling in size every year.

If you make biology easy to engineer, and you make it accessible, by definition people will learn about it, and write comic strips about it. You can talk to computer programming conferences about it, and it's a different world. It's a very different world from going around claiming that you've created life. It's a very different world from going around filing patent applications that say you have invented the idea of a synthetic genome. It's a very different world from going around and spending $40 billion on a classified biological defense facility, at the site of the past U.S. Offensive Biological Weapons program. And so there's a cultural mismatch.

The mismatch is largely generational, and it's also largely perspective driven. By that I mean that the previous generation of people working in biotechnology are scientists, and the ones coming up now are engineers.   We're going to have to invent our new world of biotechnology and I suspect we'll learn lessons around biological safety from the past generation, but all the other lessons are up for grabs. The bio-security framework is going to collapse. The IT framework based on patents isn't going to scale, and the questions of playing God or not are so superficial and embarrassingly simple that they're not going to be useful in discussion.

The more serious situation is that these issues of human practice don't get resolved in a six month conversation; it's not like what happened in Cambridge, Massachusetts in the '70s, where recombinant DNA work got shut down for a little bit, and then became okay. The technologies are being developed and distributed so quickly, yet there's still so much more to do in improving the work of biological engineering. The conversations we need to set up are conversations that need to persist in ways that are constructive for decades.

The Open Source world is one thing; if you're trying to invent a language for programming DNA, having a proprietary language seems stupid. If Oxford University had supported privatization of the English language hundreds of years ago, the dictionary they made wouldn't have been so useful. And so to a first approximation, there's going to be a core collection of standardized genetic objects that can define families of languages people can use to program DNA. And those have to be made a public resource.

This will be a big transition from today. Biotechnology today derives investments from business models that support the exclusive application of different biological functions for a very small number of problems. For example, there are wonderful companies that have locked up most of the relevant intellectual property around how to engineer proteins to bind DNA. The products that they can deliver are going to be measured in small positive integer numbers, a few diseases.

But, the real value associated with being able to engineer proteins that bind DNA are in the uncountable applications people could use the proteins for. It's like a programming language where it would be a big downstream economic cost if you owned "if/then" and you were the only person who could use it. We need to be able to reuse this stuff in combination. Note that the ownership of biotechnology will play out in a landscape that is surfing along a technology transition where, as automatic construction of DNA gets better and better and better, you're going to care less about the specific material you have, you're going to care more about the information on a computer data base and the computer design tool that lets you organize that information, compile it down to a DNA sequence, and print it. As soon as you start to manage information, all sorts of new ownership, sharing and innovation schemes become allowable.

Where will we be 30 years from now? 1995: haemophilus influenza sequence. 2001: draft of human genome available. 2007: multiple chromosomes assembled from scratch, bacterial virus, or organelle. 2012: design of eukaryotic chromosomes should be routine. You want to build baker's yeast from scratch? Also, five years from now, we may have just begun to make some good progress on reliable functional composition of standard biological parts. Nobody knows how expensive solving that problem will be, but because biology works there's plenty of existence proofs. There is where I can't tell you if it's going to be five, ten or 30 years. If I had to guess, I'd say we'll have a collection of tens of thousands of genetic objects that support reliable functional composition between ten and 15 years from now.

Okay, let's cut it a different way. I'd estimate the cost of synthesizing the DNA of every human being on the planet that'll be born in the next year at $10 trillion dollars. That's 20 percent of the world's economy. That number is dropping by about a factor of two every 12 to 18 months. On what time scale does it become worth considering whether or not we can actually afford to construct every new human genome that's going to come into existence, and we can de-couple the designs of human beings from the natural constraints of direct descent and replication with error? My sense is that technology will support this well in advance of our ability to have any conversation around the consequences of using the technology. It's not a 50 years off thing, it's not a 30 years off thing, it's probably not even 20 years off in terms of where the technology needs to get.


I grew up in Pennsylvania. I liked to build stuff with Legos, Lincoln Logs, this and that. My first exposure to biology was in a public high school, and it included some great stuff. It also included a semester where you had to memorize the Latin names of 200 insects. That didn't go so well. I went to college at Lehigh in Bethlehem, Pennsylvania—next to Bethlehem Steel, which was barely in operation, but still there—and studied structural engineering. There's one way to tell the story where I jump around from structural engineering to environmental engineering to chemical engineering, to genetics to cell biology, to biology to biological engineering, in this big loop. That's not it. If you do a frame of reference shift, the simpler way to say it is, I like to build stuff, and biology is the best technology we have for making stuff—trees, people, computing devices, food, chemicals, you name it. I somehow found my way to biology and, along with the ambition of getting better at engineering biology, there's this wonderful complementary puzzle of, how the hell does this stuff work? All these living systems we inherit from evolution. That's the condensed story.

I was very fortunate in the early '90s to find an engineer, John Yin, now at Wisconsin, who knew something about DNA. He had just come back from working with Manfred Eigen in Germany, and was studying virus evolution. He was at Dartmouth College, so I did my Ph.D. up there and had an interesting experience where I was an engineer and I was trying to develop computer models to help biologists understand the architecture of the natural genetic systems they were changing. My first experience of any interest was, I had some hypotheses coming from my work, and I tried to get some biologists to do experiments for me, and I was not successful in doing that. In hindsight, I recognize that that's because any good biologist who does experiments has multiple lifetimes of work to do. They're never going to do your experiment, so you need get in the lab and do it yourself.

That took me to Austin, Texas, at UT, I worked with Ian Molineux, who had done the early PCR work at MIT, he was running one of the last bacterial virus labs in the country, and still is, and he taught me how to map and clone DNA and do my experiments. I then spent a summer in Madison, Wisconsin and went to Berkeley, California, where I ended up working with Sydney Brenner, and Roger Brent, two good biologists, in an independent not for profit, where our mission was to go do the next generation of biology, whatever makes sense.

Part of my work there included taking a look at my results from Texas, and I noticed that all of the predictions I had made using my computer models about how these natural biological systems would behave when we changed them turned out to be wrong, especially the interesting predictions. I would want one behavior, and when I went to make the change, exactly the opposite would happen. In this situation, engineers do what's called a failure analysis. So I made some predictions. I'm going to make these changes to the architecture of this virus and as a result, the virus will grow faster. I would go into the lab, make those changes, it then would grow slower. My modeling tools weren’t good enough to support purposeful determinative changes that result in the behavior that I expect.

This was a sufficiently painful process to have given me a lot of time to think about why things weren't working out. And the conclusion that I came to, working at Berkeley, was that evolution is not selecting for designs of natural biological systems that we can understand, the things we inherit from the living world have not been selected for ease of understanding, let alone ease of manipulation. It's not part of evolution's objective function.

If I wanted to be able to model biological systems, if I wanted to be able to predict their behavior when the environment or I make a change to them, I should be building the biological systems myself. And so that for me was the transition into what is now called synthetic biology. I started broadcasting that idea in the 1990s, the only person who returned a coherent signal intellectually was Tom Knight, in the electrical engineering department at MIT.

Tom had self started in biology five years earlier and is now, in addition to being one of the best engineers I've ever met, one of the best microbiologists I've ever met. Tom was interested in it from his own perspective, having mostly to do with building computers . We need to use biology not to be a computer, but rather to build our computers, because we're going to need to put atoms exactly where we want; as semiconductor devices get smaller and smaller and smaller, you can't rely on random distribution of the dopant atoms in the devices, the statistics go to heck, and you have to figure out ways of putting exactly one or two or some small number of dopant atoms in every little gate you've got.

So that brought me to MIT in January 2002. And along with Tom, there's the opportunity to be responsive to this bigger opportunity, let's go engineer biology. We're going to pull off a new department and a new venture in biological engineering. In the context of MIT, it's not the first time that's been tried. There's a paper from 1939, written by Karl Compton, then the President of MIT, and the title of the paper is, "On the Genesis of a Curriculum in Biological Engineering", and it describes the very ambitious and impressive five year major, where you get a dual degree in biophysics and biological engineering.

Somehow that earlier effort crashed and burned. I haven’t completed  my failure analysis yet , and I don't know if it's simply World War Two and the redirection of interests, or other stuff, but what's interesting to consider is that 1939 is exactly the time that the Rockefeller Foundation's making investments in the science of biology, arguing, correctly, that the relevant physical level of resolution at which to understand the living world is atoms and molecules. That becomes molecular biology. And so biological engineering could have gotten started at that same time, but didn't. So anyway, that's sort of how I get to where I am.

I've resigned my position MIT; I'll be moving to Stanford next summer. It's a place that will support the scale of foundational research that needs to happen in biological engineering, on a time constant that's relevant to what needs to happen. There's a tremendous engineering community in the Bay Area, in electronics and software, and these are the folks who have the most relevant skills. If you look forward to what the challenges are in biological engineering, the main challenges are how to manage complexity, meaning how to produce simplicity in a many component integrated system, and how to develop a theory that supports the programming of evolution.

And there's a way to think about this that maps directly into communications theory, where you think about a sender and a receiver and a message being transmitted along a channel. In evolution I think you can make a mapping where you have a parent generation, which is the sender, the transmitter, and you have the progeny, the children, which are the receiver. The message that's being transmitted is the design of the living organism, and the channel that the signal's being propagated along is the process of replication of the machine. So at any case, the work force and the knowledge base that is most relevant to the future of biological engineering is now on the SF Peninsula.

When we organized the First Synthetic Biology Conference at MIT in 2004, we were expecting about 150 people, so we booked a room for 297. And 500 people wanted to come given six weeks of notice. Now it's going on four years later, the fourth meeting will be at the Hong Kong University of Science and Technology, which I think will blow doors off of most places in the world 20 years from now. The University is in Clearwater Bay in Kowloon, Hong Kong. I got off a flight there to meet with the local organizers, and the fellow who's championing our organizational visit is a retired banker, who was responsible for helping take Apple public all the way up to building up capital markets in China recently. He looks at what's going on and he says, explaining to a real estate tycoon in Hong Kong, 'I remember the personal computer and I remember recombinant DNA, and what's happening here is both. And this is important, and Hong Kong needs to support this.'  And he looks at me, and asks, 'Did I get that right?'  This is the first time I've ever met this fellow. 'Yeah, that's pretty much right.'  He is a very impressive fellow and a very, very good person.

There are some people who understand what's going on, and who are in a position, or who have comfort acting on time scales that are relevant. It is interesting for me to learn how difficult it is for folks to appreciate what an exponential technology really implies. The fact that sequencing goes from approximately zero to human genomes in ten years. The same thing is happening with construction of genomes. And with the parts collection—the standard biological parts doubling every year. And the same thing is happening with the number of teenagers who would like to do genetic engineering; it's doubling every year. How do you actually live in a world where you're surfing that exponential in a way that's constructive and responsible? Very few people get that.