 I want to thank everybody for coming tonight. I've looked forward to it, and I bet you have too. I want to start by observing that this is a typically fascinating day at Purdue. Just the afternoon, I've just come from a presentation sponsored by a major company of what's called the traditional economy, Rolls-Royce, an annual lecture here. And now here, to meet along with you, a leader of the front edge of the newest economy, this posed a problem for me, which is, what do I wear today? And I told Paul, don't let it go to your head. This wasn't for you. This was for the last event at which the speaker was coincidentally, General Maziello, who runs the $4 billion a year Air Force Research Organization. And he has cruise missiles, and you don't. So I dressed for him. That's the. But seriously, our guest tonight represents an intriguing bridge between those two worlds, just coincidentally. Both are, he worked at DARPA. I'm guessing this audience knows a lot about it. If not, we can talk about it. And of course, the Air Force is a principal, a partner, and customer of DARPA as well. So we are thrilled to have with us Paul Aramanco, one of the leading innovators in the most innovative business arguably that we have. We can claim him partially. We claim some patrimony, because he grew up in West Lafayette. And from there to MIT and Caltech and a bizarre trip through Georgetown Law School, until tonight, I didn't know anyone who ever went there came to anything. I did. That's what that's about. So let's, I'm going to try to draw Paul out on a few questions. And then we want the most of the hour to be led by your question. So please be formulating them. But let's get started. I, Paul's very self-effacing and wouldn't do this on his own. But I'm going to insist that he talk just for a minute or two about his own career. And I'll ask you, having done that, you're a visionary. Look forward. What would you like the rest of your career to look like? So past and future, please. Sure. Absolutely. Well, first of all, thank you. Thank you very much, President Daniels, for having me here. Thank you all for coming and for taking the time. I hope to make this, that we can make this interesting and worthwhile for you. This, by the way, represents Google formal attire. Just so you know, this is like Google black tie. Otherwise, it would be a hoodie. But to answer your question, both simultaneously sort of going back in time and going forward in time, because there is a certain circularity and symmetry to most of our lives, right, I want to build a starship. I've always wanted to build a starship. And for sort of as long as I can remember, after I stopped wanting to be a bunny rabbit, I wanted to build a starship. And I grew up, as President Daniels mentioned, in West Lafayette. I went to West Lafayette High School. Both of my parents are in the audience and both are at Purdue. And quite early in my high school career, I started taking flight lessons because that was the closest I could get to a starship at the time. And I trained at Eretz Airport. For those of you who remember it, sadly, the field, it was in Lafayette. And sadly, the field closed a few years ago. Purdue Flying Club used to be there. Moved over to the field here in West Lafayette. And then after getting my pilot's license and graduating high school, I would have loved to stay here at the wonderful aeronautics school that there is at Purdue. But as I said, both of my parents were on campus. So I fled. And after getting my undergraduate and master's, I built UAVs, what are, I think, somewhat uncharitably and a little bit unfairly are commonly known as drones. And so I was an aerospace engineer. I was a UAV designer for some years. Did a stint management consulting. And over that trajectory, I got really interested. So I never lost sight of the goal of ultimately wanting to build a starship. And one thing that really struck me, for those of you who've ever seen this, Norm Augustine, a former CEO of Lockheed Martin, is very fond of plotting the cost of airplanes over time, going back to the Wright brothers. And he plots it on a logarithmic scale. So it's an exponential curve. And if you look at it, it's been growing exponentially. And if you project it forward in the year 2054, the entire US defense budget buys one airplane. And the Malthusian fallacy sort of not withstanding, it appears that we're tracking that trend line and have been for some time. And so my view was, if I'm going to build a starship, it's going to be like year 2054. And if an airplane is going to take up the entire US defense budget, what's a starship going to cost? And so what I got really interested in is the limits of complexity that we, as a species, are capable of grappling with in engineering systems. And I think that a lot of the big defense systems, a lot of the aerospace and defense systems are pushing that edge of what's feasible, what we as a society are capable of managing. And I think that's a lot of why that escalation has taken place. And so one of my passions in animating things throughout my career to date, and I suspect going forward, has been, what are the ways of tackling complexity? And that complexity transcends just, I mean my particular interest isn't sort of going to the nearest star, but that complexity is all around us. It's in cities, it's in the energy grid, it's in the internet, frankly. So that's really been my interest. And then after spending some time at DARPA, for those of you who know DARPA, it's a tenure-limited organization. That's one of the ways of sort of applying pressure to the innovation process. ATAP, which is the group at Google that I sit in, the Advanced Technologies and Projects Group, is meant to be an attempt to replicate some of the DARPA DNA and the DARPA innovation model in the private sector. And so it is also a tenure-limited organization. So I know my expiration date and it creates very interesting sort of psychological incentives. And frankly, I don't know what I'll do after that, but the ultimate goal is to build a starship. Yeah. Well, the overlap between your presentation and General Marcielos is even bigger than I thought. There you go. He probably wishes he would have been here to hear that. He's grappling with those curves that you're talking about. That's exactly right. Well, our guest is not just a practitioner, but if you read about him, a thoughtful and insightful theoretician about innovation. So let's just spend a little bit of time on that. Are great innovators born or can they be made? Can it be taught as some universities hope they can? Can it be nurtured if they get the environment right? Can people become more imaginative and creative than they might otherwise be? Or is it pretty much a gift? It's an interesting question, right? And I don't have an empirical answer to it, but I have my own theory that's at least anecdotally supported. And I tend to subscribe to paraphrase Richard Feynman. The great physicist, I think he said that everything he ever, he sort of stopped learning new things right around grad school. And it's been always applying the toolset that he has learned to new domains and to new sets of problems. That's probably gross paraphrasing of what he actually said. And so I do think that a lot of innovation is taking a set of tools that we know and sort of patterns of how we know to solve various problems and applying them in new territories. I'm not saying that an aerospace engineer can build a cell phone, are you? Well, so you would be surprised, right? So a lot of the genesis of the ARA project actually comes from a satellite program and a manufacturing program, a design tools and manufacturing tools program that I ran at DARPA. And I joke to Regina, my boss, also my boss at DARPA that I'm a one-trick pony and she just needs to milk me for all the different tricks that I can do in different domains. But in all seriousness, I do think that the tools that we're all equipped with, now those tools may have some, obviously, you might have some biological genetic predisposition towards having a richer tool set than somebody else. But I think that most of innovation really is the ability to force yourself out of your domain of comfort and apply those tools somewhere new. So for those who haven't had a chance to read about the DARPA way or however you term it, the DARPA, now the Google ATAP modes, say a little bit about that approach to research innovation. Yeah, so just a real quick summary for those of you, I was giving a talk in a class earlier and I was going on about the DARPA model of innovation and then at the end, the very first question was what's DARPA? So I think I should just make sure that everybody's on the same page. So DARPA is the Defense Advanced Research Projects Agency created by President Eisenhower in the wake of the Soviet launch of Sputnik. And the purpose, the charter of the agency was the creation and prevention of strategic surprise. So quite explicitly, by that charter, it is not a requirements-based organization. So the ideas are supposed to be, so it's an inspiration-driven organization, if you will. And it has a couple of attributes and ATAP is our way of trying to replicate that DNA in the private sector. And actually, there have been a lot of attempts to sort of duplicate the DARPA DNA and I would say that they've been met with mixed success at best. And so one of the things that I would credit Regina, who was President Obama's, Regina Dugan who was President Obama's first appointee as DARPA director in 2009, was trying to distill that DNA to a core set of ideas. So first and foremost is the P in DARPA and the P in ATAP and that's projects. And projects is a very special term of art for us. It means that we operate at the intersection of a compelling use case with fundamental scientific or technological insight. And I think one or the other, by itself, Pasteur's Quadrant, that's exactly right. That's a term that Donald Stokes, I believe, coined. And it is in contrast to Bohr's Quadrant, which is sort of pure fundamental research and Edison's Quadrant, which is more applied, tinkering, right, in essence. And so Pasteur's Quadrant requires both. And there is an observation, which I think is empirically borne out if you look historically sort of down back to the primitive era of innovation when we were inventing sort of stone tools, all the way through the Manhattan Project and Apollo, where use-inspired research, use-inspired basic research produced some really exciting results. And I think a lot of the fundamental innovation model, both in the government and industry and in academia, I think over the course of the 20th century has drifted away from that, actually. And we think that it's helpful to all of those entities to try and bring it back to that. So that's one aspect, is the project focus. The second aspect is trying to overcome our predilection, or I think innate, I suspect, predilection towards being risk averse. And getting people to take risks is not just a matter of telling them, be more risk tolerant. You actually have to create incentives that appeal to them at a more fundamental level. And so we have two examples of those. One is applying time pressure to projects. And in some cases, that's an intrinsic time pressure. So at DARPA, for instance, there's frequently a critical warfighter need, unfilled need, and people are dying or being injured. And there's nothing quite like that to drive urgency and the ability to take risk. In the commercial sector, some of those timelines are perhaps more artificial, but if you create a specific timeline and you stick to it, you have to have sort of persistence of principle, then it creates, it sort of puts getting things done over making sure that you dot all the i's and cross all the t's. And in large bureaucratic organizations, as they tend to be, that's very important, is to get people to prize that. Now the second thing we do is we limit people's tenure. So at DARPA, it's typically two to four years. At ATAP right now, it's fixed at two years. And that sounds a little draconian. Not two years on a project, but two years in the organization. That's right, that's right. And it sounds a little draconian, but the reason is because we don't want people to make a career out of it, because as soon as they start thinking about sort of what's gonna happen 10 years from now, and am I gonna get good peer reviews, et cetera, they don't wanna take organization bureaucratic risks or organizational risks. They don't push us hard, they try to play nice. And we try to incentivize people not to play nice, but to get stuff done, quite frankly. And those can be intention. We're all lovely people, I assure you. But when the two are intention, we want people to do that. And then the third aspect is open innovation. And particularly at DARPA, that's a non-trivial thing. There is a tendency in the government to classify things, particularly in an aerospace and critical military system domain. And I'm not disputing the importance of keeping some crown jewels close to the chest in any way. And similarly in industry. But our belief is that open innovation, both in terms of the partner ecosystem that we access for innovation, and in terms of actually making the product known, ultimately make for better products. They create some commercial risk or some competitive risk in the case of defense systems. But ultimately that risk trade-off is worth it. And so for instance on Project ARA, which is a fairly significant undertaking for ATAP, there are five of us who are at Google on the Google payroll. That includes me. And there is some three, four, five hundred people outside in the partner ecosystem. It includes a team at Purdue. It includes teams in industry and startups and nonprofits sort of all over the place. And that lets us capture best in class, best in the world mind share. People I could never convince to sort of pack up and move to Mountain View. Yeah. Now it's a fascinating model. Road tested at DARPA. And it's interesting that businesses have been so, it's such difficulty either noticing or emulating it. But I did enjoy, when most of us think of Google or have visited Google, you think of this wide open, anything goes, environment. I did enjoy reading somewhere where either you or one of your colleagues were glad to move, you moved ATAP away. You didn't want to be stifled by the Google bureaucracy, which is not the way most of us think of that company. Google is a unique place. And one thing that Google has done, right? So the draconian measures about time pressure and sort of kicking people out after a couple of years. Apart, Google has more so than any other place that I've worked, organically created a culture that I think is more accepting, much more accepting of innovation than others. But it is fundamentally different and Google is a software company for the most part. ATAP is a hardware, is a mobile hardware organization for the most part. And we are trying to do sort of a fairly controlled experiment. So I think that that level of autonomy has been helpful to us rather than being assimilated into the broader culture. So let's hear about Aura. I asked you before we came out, is it reasonably accurate usage to think of this as the democratization of hardware, much as many have opened the world of software to the world of those who are interested? Is this something like the same animating? I think that's an excellent analogy that holds sort of not 100% because hardware in the end is a little bit different than software. But we're basically trying to create the equivalent of an Android type ecosystem, but for hardware. And so Android has been sort of phenomenally successful in taking software development, which was the province of a good number of competitors, but still multiplying that by several orders of magnitude in app development, right? Creating an app is almost trivial today. So anybody can do it, anybody can go home and write an Android app, publish it to the Play Store and it could very well catch on, right? So it's kind of a pure meritocracy, highly competitive and the innovation time scales are extraordinary, extraordinarily short. And so we wanted to see if that was doable in hardware. And so by following sort of the Android model, we are creating a free and open platform. So as Android is open source and freely available to anybody, the Aura developers kit, what we call the MDK is free and open and available to anybody. And so anybody can create a module per the specifications of the developers kit and put it in the Aura module marketplace which is analogous to the Google Play Store and sell directly to consumers and see if their module is interesting, if it flies. And also we're trying to create a new generation of design tools, something that I started at DARPA that makes the design of hardware much more like the software design experience. It raises the level of abstraction on the design process. And so we're hoping to shrink the ability to introduce new technology into the smartphone domain by factor five, that's our objective. So today it takes maybe 12, 18, 24 months to make a new smartphone and bring it to market. And we're hoping to get that down to a couple months. So on an app type, time scale. So that's kind of the developer-facing goal of the project. There's also consumer-facing goal of the project. In any platform, right, there's the two sides and we play as the arbiter or the facilitator of this two-sided market. And so the user-facing goal for us is to be able to create a platform that allows people to customize the device, both functionally in terms of what modules you wish, and also aesthetically, and that's not just a matter of sort of selecting the color of the device, but actually being able to tell some sort of story through the device. My goal, I like to say, is that you put your RFM down on the table at dinner and it becomes the topic of conversation for the first 10 or 15 minutes. So we wanted to tell your story to actually be expressive. And we believe that that level of flexibility for consumers to customize the device, and if we can span price points that range from sort of a feature phone crossover price point all the way to an aspirational, very high-end device that may be able to do portable medical diagnostics, that that could potentially help us on the way of delivering the mobile internet to the next five billion people who are not currently smartphone-enabled. And we often forget that living in the developed world. So those are the two goals. I have some pictures of sort of where we stand, so if you don't mind, I'll show those. So this is two different sizes of the frame. On the right is our medium variant, on the left is a mini. There's also a jumbo that's two thirds wider than the medium. You see, so this is the back of the device, obviously. There are three types of modules. There is a one-by-one, which is the little square one. There's a one-by-two, which is twice the size of a one-by-one. There's the larger two-by-two modules, and those are interchangeable between, the first two are interchangeable between the two variants there. The front of the device is also modular, so you can replace the display, you can upgrade it, or if it breaks, it becomes a relatively low-cost component to be able to replace. Here's our first functional prototype, which with a module disassembled, and so you see the module base populated by whatever the module developer creates. There's a shield on top of that that makes all modules look the same from an RF, from a radio frequency emissions perspective. An antenna, if this is a Wi-Fi module, so there's an antenna, and a shell. And the shell is consumer-removable. So you can put a picture of your cat on the shell, and then if your cat dies, you can get a picture of a new cat on the shell and swap it out. Couple of interesting technologies here, so, and in fact, maybe before I get to that, I should say one thing about the industrial design, and this was, by the way, the biggest change for me going from developing airplanes and spacecraft and robots to cell phones, airplanes are kind of, either we care less, we have a less intimate relationship with airplanes, which is probably true, or maybe they're just inherently beautiful, but you kind of don't have to worry about the industrial design, or the appearance, the outward appearance too much. Here industrial design is a really big deal, and modularity has negative connotations for most people. They think Legos, they think Blocky, they think Brick-like. And so we saw design inspiration for this from a variety of different sources, and actually the key inspiration, the key moment of inspiration, which unfortunately I can't take 100% credit for, John Mada from the Rhode Island School of Design gets some credit for it, was the Japanese bento box, which is, if you think about it, is something that brings together very disparate dishes that usually look very different, and in fact, they try to make them as colorful and interesting as possible. Very different tastes, so those are kind of analogs to aesthetic and functional customization for us, and brings it into a coherent meal for people, and lets them enjoy these very different things as one whole. So it kind of navigates the contrast between bringing together a bunch of separate pieces, but also giving the consumer holistic experience. So it was that as well as my sort of penchant for art deco aesthetics that led to the industrial design. But one of the things we wanted to do was we wanted to make the module smooth pebbles, so that there are, to try and overcome this notion of boxiness and Lego-like appearance. And so we use magnets to attach the modules, they're called electro permanent magnets, that's a picture of one right there, which is a unique type of magnet that you can turn on and off, but it doesn't consume power in either of those states. And the other thing that we did that you can't see very well, you can't see very well there, but use inductive coupling for the data transfer. So there's no connector, there's nothing sticking out of the module, and I don't know. So this is sort of a close-up of a module, you can see the two magnets, and you can see the data transfer pads there. So that was sort of the industrial design. I wanna ask you a consumer related question. Well, Henry Ford famously said, if I'd asked the consumer what he wanted, I'd have built a faster horse. And there have been many parallels, if I'm permitted to mention his name, that guy at Apple, Steve Jobs, was, I don't know if disdainful is the right word, but certainly didn't rate market research very highly. He held, I think, to the view, as I understood it, that really it's a job of people like you to imagine new products and services that people on their own wouldn't know that they wouldn't think to request demand or desire. Which is the are? Did you, was there some sense that the world was waiting for this or is this more intuitive? So I do think that it seems to tap into some unmet desire among smartphone users. At least that's the response that we've seen in the Twitter sphere, blogosphere, social media and in various venues. And I think that's because the industry seems to have converged towards a dominant design. And frankly, at least my view is that successive cell phones from whatever manufacturer are rather boring, because they all look the same. And it's a labor arbitrage play and number of megapixels, and that's about it. So I don't hold my breath for a new cell phone release. And I think people share that, and so there is some angst for something new. That said, I do share Mr. Jobs' view that it's very, for a category-defying device, it's very difficult to design it based on polling data. And so I think that there's been this big trend over the last several decades towards design for empathy or human-centric design. And I think that there is some merit to that, but I think it's better at doing incremental advances than it is at category-defying kinds of devices. So we're pleased to see very positive consumer sentiment after the fact, but it was not designed based on research data. So I'm gonna invite the audience, anybody, to make their way toward the, I think there are two, mics, and in hopes I've just incited somebody to do that. Just a couple other quick questions. You talked about virtues, maybe, of working under pressure and DARPA and model and all that. I love to quote, I read somewhere that you said that innovation under extreme time pressure is generally of higher quality, which I guess is that point. Are you gonna make the deadline of early next year for this one? So I do wanna set the record straight. One of the things about running a very open project is that there is a lot of, there's a lot of crosstalk in the social media and the tech blogs, and not everybody hires fact checkers. So the actual fact on this point is that we are planning to do a market pilot in the course of 2015, and we don't have a general market release date. There's a lot more that we need to do in terms of retiring technical risks on the project. We have probably two more iterations of spirals, as we call them, of prototypes to make, and then during the market pilot, we have a lot to learn about how people interact with this device. We have a hypothesis, for instance, that a whole set of secondary markets will emerge, and by which I mean the idea of selling used modules, sharing modules with friends, swapping modules, maybe having one module for a given type per family or per group of folks. And we're very interested in understanding all of those dynamics first before we take a step to unveil something like this globally. Well, it's one release. I will be breathless to see come out. So, well, let's start over here. We'll alternate sides. Hello? Yeah. Mr. Aramanko, at the beginning of your talk, you mentioned your interest in building a spaceship, and this is something that Elon Musk is also investigating. He's right now looking into ways to recycle rockets, make them cheaper for space exploration. He gave a talk last year where he made kind of an audacious claim that within 15 or 20 years, we might expect to see a human on Mars. And I was wondering what your thoughts on that timeline were, whether you think that's realistic or outrageous, or if you know of any progress being made that direction. Yeah, I mean, I think that, I think there's a two-part answer to that. I think there is the technology question of do we have the technology necessary to get there, to sustain a human, to either bring them back or to credibly colonize the planet. And I actually think most of the tech is there. There's obviously some, there's devils in the details, and even bigger devil is in the integration problem of a very large complex system. And, but I think that's all, I think that's tractable, and it's tractable probably on a timeline that's much shorter than the one you cited there. I do think there is a question of willpower and commitment, and I think that one of the challenges that we have, and actually, maybe not to put you on the spot, but maybe President Daniels would comment on it, given his experience in government, is sort of the vagaries of the fiscal cycle, is one of the aspects of democracy is the fact that administrations change, Congress has changed, and priorities as a consequence change. And so NASA, which ostensibly is our societies, or at least America's societies, sort of answer to the challenge of getting to Mars, has a tendency to sort of redirect rather wildly over time, and there does need to be a sustained commitment on the sort of order of magnitude of a decade or so, to be able to do that. And so I think that's the bigger challenge. I do think that the advent of the new sort of wealthy Silicon Valley class, of which Mr. Musk, I think, I suspect is a member, could offer a different, an alternative answer, an alternative answer to that question. As it happens, I just finished literally yesterday, I think we're done. Coaching a National Academy's look at this very question, and I guess my short answer, first of all, our principal finding was exactly the one Paul just talked about. We need a totally different way of governance, different mode of governance. Presidents have to quit changing the direction, Congress has to quit tinkering and micromanaging and so forth. They have to sustain this over time, which democracies aren't known for doing. But while I hope Musk and a few other enthusiasts are right, after spending a year and a half with some of the top technical people in the world, the short answer to the question is, in 15 or 20 years, you can go there and crash. You can fly by, but you won't go there and land and stay based on anything we know now. There's some just gigantic, I'm sorry to say, daunting hurdles to be crossed. We named entry, descent, landing. It's an unbelievably hard thing to imagine. In space propulsion and radiation protection for the crew was the three biggest, so let's go. Can I follow up on that comment? So I do wonder whether the question is also with what degree of reliability, because if you were to say, okay, you have a 70% chance of coming back, that might not be palatable to NASA, for instance, but might be palatable to a private sector. Very important, and I confess that having been the one to press this point, we also said in the report that the nation, to embody, if you're serious about going to Mars, the nation has to accept that there are risks, there will be setbacks, and there probably will be the loss of life, and our laws are not set up that way right now. In reaction to the shuttle disasters, they're very, I'd say, defensive about that, so. Let's go over here. Hi. This is about aura. So a lot of manufacturers, they'll eliminate like a battery cover, and they'll take like a micro SD card slot out just to make room for more technology in the phones, and with a modular design, you mentioned some cool things that probably save some space, but just how do you compete with people that take a clip out, like a battery cover out just to save that space? Sorry, I'm not sure I fully understood the question. Okay, so like with a lot of manufacturers like Apple, they take the back cover off, right, just to more space as much as possible in that phone to fit electronics, whereas you're designed to be modular and there's plastic in between everything, how do you compete on the specs and everything? Oh, I see, I see, I understand the question. Yeah, so, yeah, so I guess I should point out that we didn't invent modularity. There have been a number of attempts at modular devices. There was a modular PDA, some years back, there was an Israeli company called Modu that tried to make a modular phone a few years in the sort of early to mid 2000s. So we are by no means original to the concept. I think we have an original industrial design, but, and I think what led us to believe that this could be the time is the trend towards miniaturization is in essence a corollary to Moore's law, right, that the overhead associated with modularity gets smaller and smaller and smaller. And that includes both on the material side, the mechanical overhead, as well as the electrical overhead associated with all of the networking and signaling and the flexible power bus that has to go into the device. But there is still overhead. There's no free lunch. And our estimate, just speaking in general terms, across sort of the size, meaning PCB area, weight and power, assuming the outer shape of the outer dimensions of the device are fixed, is about 25% overhead, is that we can get it down to that level. And so that means 25% less battery volume or 25% less circuit board area for functionality, maybe 25% more mass perhaps. And so our assessment, which again, it's very difficult to get consumer data to support this sort of thing. But our assessment is that at that threshold, it would be acceptable to consumers in exchange for the diversity of the kinds of things that you could get in the device and for the customization. Thank you. Over here. In terms of career choices, how did you go from West Lafayette to Google? So I actually ended up at Google sort of accidentally. I, as I mentioned, I went to DARPA and then I followed Regina to ATAP. But ATAP was in Motorola. So I actually thought I was going to work for Motorola. Motorola was of course owned by Google at the time. And then just recently, just a few months ago, I think February was the date of the announcement, Google sold Motorola but kept the ATAP organization and moved it into core Google. So I'm a bit of an unwitting Google employee. That said, I do think that the Google sort of innovation culture is very consonant with the kinds of, the kinds of radical things that we're trying to do. Over here. First of all, thank you for joining us here. It's been a pleasure hearing you speak. You mentioned industrial design earlier. And as an industrial design student, that is very interesting to me. And given we're at an engineering school, do you think there's kind of going to be a kind of a hybrid field between engineering and design in the near future? Since the definition of design right now is kind of ambiguous, it's kind of stylistic, but at the same time it's engineering. What kind of hybrid environments have you encountered Google between engineering and design? So my personal view is that the best engineers and the best designers are polymaths, meaning that they can do a little bit of everything and they can synthesize these things in their head. I actually think that Mr. Jobs is an excellent example of somebody who was able to do all of those things and envision a holistic product, both in terms of its functionality and in terms of its aesthetics. And I think you have to have, I think the really successful folks in any discipline are those kinds of interdisciplinary synthesizers. I think that I don't know of any trend, but I also don't know sort of the industrial design, academic landscape that well, towards educating students that way. And so that may be something that you have to do on your own in your copious spare time. Thank you. Great, thanks. First, I'd like to congratulate you on your free trip home. So I guess my question is how do you balance the openness of the design of the phone with the reliability of the product as a whole? For instance, power supply requirements for each module, things like that. Yeah, so there are two thresholds to that for basic safety of modules. So making sure that a module doesn't melt, doesn't catch on fire, doesn't fry your brain, those kinds of things. We will do that safety testing or we will require module developers to self-certify to that kind of safety testing before being admitted into the module marketplace. In terms of the module meeting performance specifications or being nice to use, the intent is to rely on user feedback in the very much the same way that the Play Store does with I think reasonable success rely on reviews and user ratings. Thank you, Mr. Emanko, for being here. My question was about small and medium sized industry manufacturing units making these modules for as a business. What kind of setup and capital investment would that sort of a business need? So we've tried to lower the capital investment as close to zero as at all possible. So whereas you could read the MDK, the module developers kit, and make everything from scratch based on the MDK. So from a philosophical perspective, I view that as the gold standard and has to be there as a fallback position to keep us honest. But as a practical matter, we are striving to make it so that you can buy all of the key components. So the electro permanent magnets, there's a bunch of interface circuitry that's required in the module as a system and package, what's called the SIP. But you should be able to buy those from Digi-Key and that there will be design templates, reference designs of a wide assortment of modules that we will put out there that you can basically take and steal, if you will, and copy at will. And we have been trying to get contract manufacturers, so companies like Flextronics, companies like Quanta, to also offer fee-for-service, but they're relatively inexpensive fee-for-service kinds of things. To lay out a module, if you have a widget, say you have a new kind of sensor, and you are a savant at that kind of sensor, but you don't really care about everything else, you should be able to call them up and say, hey, lay out a module, lay out a PCB and manufacture it around the sensor for reasonably low investment. I can't quantify it yet for you. But that's the goal. And new tools. So the tools are kind of in alpha bordering on beta right now, but that should also make it easier for you to do all of the verification on the module on a desktop. Yes, sir. Yes, was this design exercise done with the idea of applying this methodology to other devices like automobiles or other technological things? And when you say this design exercise, you're referring to Project ARA? Yes. So there is sort of a natural tendency to veer, especially because I think it's invoked towards internet of things kinds of applications, and they are not lost on me. I do find them very high potential and extremely exciting. I have had to focus the team. So my belief is that if we try to be good at a wide assortment of things, we may end up being kind of mediocre at everything and not particularly good at anything, any one thing. So we have focused on the smartphone application sort of as a razor sharp focus. We want to make a great smartphone and it will probably be pretty good at other things. So you'll probably be able to take a module out and put it in your car, put it in your thermostat, put it in your television and perhaps do other things. We are not explicitly designing for any of those applications because first and foremost, I want to make a great smartphone. And I think that's a compelling enough and a difficult enough application that it should yield a platform that's very robust to a wide assortment of other things. Thanks. How do you make Project ARA popular so like cells and like huge volumes? Cause I worked as a cell phone salesman. It's very hard to sell people like something that's incredibly different. And I know Project R, you want to have like modules made by third parties so like to convince a Sony to build a camera like it needs that volume. What are you guys going to do to make it popular so more than us are buying it? Well, I'm glad you are all sold. So that's a good start. So what we have seen so far is sort of a fairly remarkable outpouring of interest from consumers. Now obviously conversion from interest and sentiment into sales is, you know, it's not there until it's there. So we'll see. And that's one of the reasons we want to do the market pilot before we commit to any sort of global launch or going beyond that is we want to see what it is about the product that appeals to people and what is the best way to market it to them. Because one of the things that you also probably appreciate having the experience that you've had is the paradox of choice is when people are presented with a large number of choices, they tend to seize up and even if they're capable of navigating the choice space, they tend to have buyer's remorse because they're worried that they made the wrong choice. And this is replete with choices. It'll probably have a higher dimensionality of the choice space or the degrees of freedom of customization than any mass market product. And so we have been thinking very carefully about how to help consumers try and navigate those decisions and curate them based on that particular consumer's appetite for choice because not everybody has an equal veracity for it. What I read about the sensor that's going to check my sweat and decide how much stress I'm under and then help me with the choices, didn't I? That is one of the threads that we're pursuing is adaptive commerce experiences that actually non-invasively monitor biometric signals about you. So that includes galvanic skin response, sweat levels, your pupil dilation, gaze direction, and a variety of other things that we can pick up non-invasively about you as you're interacting with the e-commerce system. And if you're stressed or impatient or bored with the experience, we could present a differently curated experience to you, to Kate or Morton. It's in that category where my daughters call too much information, I think. It's okay if you know, but I don't want to know. Go ahead. Well, apps are very easily distributable, you know, since they're all code and stuff like that. But I was wondering whether you think that maybe this with project error, it will be possible someday to maybe get a good distribution system that's going to be able to drive costs down and get assembly and all the, since this is physical, physical components and stuff, but like get it down to where it can be almost a streamline as app development is now where you develop once and then you can release it to the masses in, like, equates. So, the one thing I don't have a solution for is being able to download physical atoms yet, right? That would be cool. And that may be a worthwhile DARPA or ATAP project, but you still have to ship, I mean, pushing photons and electrons is still different than moving atoms around. So I don't have a magic solution to getting module, I mean, aside from a nice logistics and distribution system, which we fully intend to build for the market pilot, I don't know of a way of getting modules to you faster. I think DHL, FedEx and UPS have pretty much optimized the hell out of that problem. Go ahead. All right, Mr. Armaco, my question is about the endoskeletons, actually. So how did you, I know you mentioned you had three sizes. How did you decide upon those particular sizes and is there room for expansion in the future? Yeah, so you want the honest answer? So the medium variant, we sized based on anthropometric data about the human hand and the human pocket, and it's about the size of a Moto X or a variety of other sort of normal size, non-jumbo size cell phones. And so that's kind of the nominal variant. And then the mini is one-third narrower and the jumbo is one-third wider because of the partitioning scheme, basically. Now, it so happens that, so everybody tells me, by the way, so we're probably gonna take two of those sizes and I'm not sure which two to the market pilot. And everybody tells me, oh, the industry's trend is towards these ginormous devices, and in fact, if you travel to Asia, people walk around with talking on tablets, which I think is kind of an odd fashion trend. And so everybody is, the conventional wisdom is at least pushing us towards the larger device, either me being contrarian or wishful thinking, I don't know, I have a particular affinity for the mini. I'm not, it's not really clear to me why cell phones are getting bigger. Because to me, sort of Moore's Law and all the other things should make things smaller. And I understand that too small is, there is a too small for hand comfort, but the mini variant, when you hold it in your hand, is just really, really compelling. It's just, it's very sexy, I think, form factor. And so I'm actually inclined to go orthogonally to the rest of the industry and try that. But who knows. All right, thank you. What markets are for? That's right. Yep, thanks for the talk. And my question is, what is the, what is how to solve the compatibility issue in between the hardware and software? Do you have a plan for the, solving that issue? So I'm not sure exactly which compatibility issue you're referring to. Such as if there is a variety of different hardware and operating system or the other software, you need to allow some counter mode or some development. Yeah, so we are using Android on the device, which should come as no surprise. And we do have to make some changes to Android in order to be able to support this kind of ecosystem. One major change that we're making is that the Linux kernel level, so Android runs on top of the Linux kernel, is we're implementing a set of device class drivers. So rather than having for each camera having a unique device driver for that camera, which is the state of the world today, we will have a camera class and the device will need to conform to the camera class specification. And then if it has unique functionality, you know, say it's a hyperspectral camera or something like that, that functionality has to be pushed to user space so that it doesn't run at kernel level. And then that's a security scaling issue. If you now have thousands of modules and each one has its unique device driver, you can't link them all into the kernel and have them run at sort of the highest level of privilege. Please. Hello. Regarding this innovation, there is kind of a dilemma for the PhD students, especially because after spent like four years or five years studying one specific area, it's kind of surrounded by a barrier around him for long and probably you cannot jump through it. So I would ask, do you have any suggestion for this kind of PhD student? And another one is that when you're like, when you're meeting graduation, so you want to go to the job market. You want to convince the employers saying that you can apply your knowledge in this field to other field. So, but they always think that you always, you already studied like a long time in this field. Probably your knowledge is only in these small areas. So how can you broaden your horizon like you jump from the spaceship to the smartphone? That is a, those are provocative questions and at the risk of having fruits and vegetables thrown at me from the audience, I will say that I do think that there is a lot of opportunity for academia to work much more closely with industry and I think that's a missed opportunity and actually the same is true is between academia and government. I think DARPA was accused for many years for becoming kind of insular and scaring away academics by say publication review and other kinds of things. And while I was at DARPA under Regina's leadership, we tried to do away with a lot of that and become much more academic friendly because we do think that academic research, when crossed with a compelling use case as DARPA is prone to offer yields higher quality research and frankly yields better career opportunities for the students. We've done similarly at ATAP and so we have created something called the MURA, the Multieuniversity Research Agreement which allows us to contract with universities who subscribe to the agreement, who agree to the agreement and I think we have 15 or 16 universities now, Purdue is one of the schools and allow us to contract with those universities very, very quickly. So all of the intellectual property terms and all of the things that are usually sticking points and that usually take up to maybe six months to negotiate in a traditional corporate university agreement can now be done in a matter of days. They're all pre-negotiated and so there's basically a template and so we've demonstrated being able to get university research on contract in support of a particular project including Project ARA within a matter of days and I think that's remarkable and I think that's good for both of us. So it lets us tap talent such as yourselves but it also gives you a compelling use cases in the context of which to conduct your research and to help drive your research and I think it ends up being higher quality research. So I think my meta answer maybe to your very specific question is that by bringing corporate research and academic research closer together I think we'll ultimately benefit both. I think we'll call back stage that when I heard that pitch from one of his colleagues, this is about 16 months ago, partly for demonstration purposes, I insisted that we send our mural back in one day. That's right. Just to let them know we... And that's the master group, that was the master group that you sent back. That's right, that's right. I wanna thank everybody for the brevity of the questions. If we're gonna get through what we have in our point of time we're gonna need to continue that so please over here and as concisely as you can. Okay, I was wondering your points of views on Project ARA being used in the makerspace. How friendly you want to be to hackers using the guts and the body to do things and how much you think there's a potential for that. Yeah, we are extraordinarily friendly to that. You can do whatever you want with it, right? It's a completely open, it's an open design. We make developer hardware available to folks and we've had a lot of interest from the maker community. I do think that ARA is an opportunity to actually convert sort of hobbyist level projects into something that you can ultimately monetize and turn into a career if it's a really good idea. Which is not an avenue that easily exists today for makers and DIY enthusiasts. Great, yep. Yes, my question. So the modules are intended to be very open and anyone can create one but on the other side, the base hardware. Is that something that's intended to, you have these two options and obviously there may be more in the future but is it something like maybe Samsung makes a new smartphone and has a module slot in it because you can change out your camera or something like that? Is the base hardware something that will be able to be made by other third parties? So for the foreseeable future, our intent is to fairly tightly control the endos. And that's actually not a, that's not meant to be a money making kind of thing. It's a fairly low value item. It's just a frame and a network switch. The reason for that is until the ecosystem grows up, both on the developer side and on the consumer side, we're very, very worried about fragmentation or perturbations to the platform specification itself. So I wouldn't rule out licensing the endo potentially to third parties in the long run. But in the immediate future, by which I mean for the duration of the development program and for the market pilot. So probably through 2015, I think we're gonna keep fairly tight control to ensure consistency across the endos. Hi, I have a question similarly about the base system. Where would you put your project on a scale from, let's say, like a wall phone where you don't really have any control, it's non-modular to a box of certain, like, that was a bad way of, like how modular can you go? Because you can say we can replace the memory, we can replace the storage, we can replace the battery, but are you willing to go down to we can replace the CPU, we can replace the PCBs? That's right. So the frame itself is a dumb device. It doesn't do anything by itself. So the CPU is one of the modules. So everything is a module. And the partitioning of functionality across modules is for the developer ecosystem to sort out. So some things probably are naturally bundled together. Some things should be separated because consumers want to swap those things out more frequently. And we don't have a strong position on that. I mean, we put out some reference designs, but they're non-prescriptive, they're non-prescriptive in nature. Thank you. Hello, Paul. My name is Sergey. I have one question. I heard that are you using 3D printing for producing some parts? Why are you choosing that way? Because I heard that very expensive and slow way. Please comment that. So we did in fact look at making the shells using 3D printed materials that offers sort of an extra degree of customization in that allows you to alter the texture, creates sort of two and a half dimensional structures and sort of an interesting new direction for digital artists and consumers alike to customize. As it turned out, there is a little ways to go in the 3D printing space. So for now, we're using polycarbonate with disublimation as the aesthetic elements of the device. I do think the 3D printing offers some really exciting prospects and I think that cost in the long run will diminish as it gets to maturity and scale. That strikes me as an industrial-based domain that will follow something very close to Moore's law in terms of the pace of advances. So I'm not too concerned about the long-term costs of 3D printing and the capabilities are pretty staggering. So one of the things that we have looked at doing is embedding the antenna in the shell and creating the antenna custom from shell to shell using conductive ink layer. So you just have another inkjet head that deposits conductive ink as opposed to the accolade ink that's used to make the plastic. So it is a very exciting technology and one day I think it'd be pretty cool if you could have a machine. It's probably a combination of additive, so 3D printing and subtractive, like milling and machining processes and some pick-and-place, but it would be pretty cool if you could have a machine that you say, here's the phone I want, here are the features I want. It doesn't have to be modular and it just makes the phone. And I do think that day will come. We've got five would-be questioners and actually fewer than five minutes, but we can take five. So let's keep moving and briskly please. Hi, I was wondering about the process that like a DARPA or the ATAP uses to go fishing for innovative ideas and then choose which ones to pursue further. So we at the executive level, and this is true both at DARPA and at ATAP, is there has to be some overarching vision, right? So what are our technology areas that we're interested in getting into or technology areas that we're not interested in getting into? And that does require some higher level thinking about the strategic landscape and that's done by the leadership. So at ATAP it's done by Regina, at DARPA it was done by the agency director and the office directors. And then once we identify a domain, so in the case of for instance, me being recruited into ATAP, Regina was interested in the domain of advanced manufacturing and customization. And so you identify it at that sort of level of resolution and you say, that sounds like an interesting area. There may be some new possibility, exciting new technologies emerging, right? So let's see if that opens up new product possibilities and let's find the best person in the world in that particular area and let's bring them in as a, to formulate a project and run the project. And so that's kind of the recruitment model, both at DARPA and at ATAP, is you identify broad areas of interest and then you try to find the best talent that you can anywhere and convince them to come in and explore those areas. Great, thank you, over here. How you doing? You pretty much answered most of my questions as soon as I walked up, but I refuse to walk away. Make up another one. My comments, my questions are regarding the security and privacy of the device, especially since the device is very modular and we're heading towards this concept of internet, things where all of our devices are interconnected. How do you ensure, well one, how do you ensure security and privacy of the user? How do you, how does the modularity of the components affect the security model of the operating system architecture and how does each component affect another component's security, especially in terms of communicating with each other? So these are all very fair and very apropos questions and I wish I had sort of a pithy catch-all answer to them and I don't. So as any new innovation expands the, and changes I should say, not necessarily expands the attack surface from a security perspective, but good ones also introduce new opportunities to improve security and manage security. And so in this case, yes, you now have a network on device and that all of the different attack vectors associated with the network architecture are now present on the device. And so the same kinds of tools that we use to safeguard Ethernet type network and Wi-Fi type networks, we now have to miniaturize and in place in the device, we have to apply appropriate encryption standards and things like that. I should say though that our architecture does offer some interesting new opportunities for privacy and security. One example is those little one by one modules that you have, I think it'd be really cool. And mind you, Google doesn't make, it's not in the module business, so this is just an idea for somebody else to go make. Hopefully there's some entrepreneurs in the audience. It's an identity module that stores all of my sort of roots of trust, all of my certificates, all of my private data, for instance, that's on the phone. And I take it out and the phone is completely anonymized. And I can hand the phone to someone to Mitch and he would put in his own identity module and it becomes his phone completely. And everything about it becomes his phone. And then I carry that on my key chain and hopefully don't lose my key chain, right? So there's another, so that's an opportunity, right? But creates another set of vulnerabilities that you have to figure out. So there's no, I don't have a magic answer for you, but these are clearly very serious concerns and we have to address them before we put this in consumer hands. Great question, thank you. Hi Paul. So this is a more general question on innovation. So you mentioned a couple of times that we've been, as a society, pushing against the limits of complexity that we can deal with. Do you think there is like an ultimate limit before the system becomes too complex for humans to innovate? So I don't think that there is. And I think that there are many examples of us introducing better tools, smarter architectures. So I don't think that the only answer to complexity is make things simpler. I think that there are more sophisticated answers. I think integrated circuit design is one example of where we have been able to sustain some seven orders of magnitude growth in complexity, as in that case measured by transistor count, without compromising product development timelines or product quality. And in fact, product development timelines have gone down, if anything, slightly. And the way that's been done is through modularization of the product, which makes it a little bit less optimal, which as we discussed, and through tools that allow you to design at a higher level of abstraction. So designers no longer, as was the case in the 70s and early 80s, when you made an integrated circuit, and if you find a photograph of the 8008 or the 8080 from Intel, early 80s kind of processor, 70s, 80s type processors, you can see them, it's a jumble of wires, right? And it was very clear that a human did all the routing and placement of transistors on those chips. And as you move past the Intel 386, and I'm just using Intel as the common timeline here, this is true across the integrated circuit industrial base, you can, just looking at the chip, looking at a microscope photograph of the chip, you can see that it's clearly machine generated. And so today, in order to design an integrated circuit, you describe the logic functionality in a fairly high level language, like RTL is one example, and then you get the physical system. And the physical system is less optimal, both because of modularity and because it's auto designed and follow certain rules and patterns in order to simplify it. So you sacrifice optimality in exchange for complexity and keeping time and keeping the verifiability of the system. And I think that's possible across a broader range of cyber-electromechanical systems. I just think that the, I think one of the things is we tend to reward in, particularly in the corporate world, we tend to reward engineers for the what, but not so much for the how. And many fewer minds are thinking about the how, the tools and the manufacturing processes. The sexy stuff is this shiny new airplane and nobody cares about how that airplane was designed, but really a lot of unlocking, a lot of the innovation might be in looking at new ways of doing the how. So I don't think there's an inherent limit to complexity. I do think that adding brains, using the same tools and just adding brains and growing the team or growing the company, growing the organization is not a way to tackle complexity because humans are not composably additive in that sense. We have a very low bandwidth comms link between us. And so doing parallel processing between our two brains is not an efficient way to divvy up the problem. We need a new way. Hi, Paul. So I'm a father of two boys in the West Side School system, so you're a great inspiration to me. Thank you. I wonder if you can share with the parents out here and the students how hard you've worked and your approach to education. My approach to education? Yeah. Gosh. So I think I have a very simple answer to your question, which is you just have to find something that you or the object of your educational efforts is passionate about and something that excites you and inspires you. And then it ceases to become hard work and it becomes fun. And I have been extremely lucky, and I'm grateful to the West Side School system and to Purdue and the other entities that have enabled, people and entities that have enabled this for me. But I have been extremely lucky that every single job I've had has been the most fun I've had up to that point in my life. And I hope that progression can continue. So I don't even think of it as work. I think of it as fun. And so having hard fun doesn't sound nearly as bad as doing hard work, does it? Last word. So other than smartphones, what other areas does ATAP work on? And second, do you hire interns? Excellent. Those are appropriate notes to close on. So ATAP is still growing. There is a number of different projects. Not all of which are public. The public ones that you can read about are Atango is a sister project of ours. It's an augmented reality system. It's a sensor suite that goes into tablets and into smartphones that allows you to not just take visual images and video of the world around you, but three-dimensional mapping. So you get a depth map. And so that allows you to do all sorts of fun augmented reality things, from gaming to trying to potentially create a map of all indoor spaces in the world. So imagine Google Maps, but indoors everywhere also. And so that's pretty exciting. It's led by a very talented researcher originally from CMU who also worked on the Microsoft Connect system. His name is Johnny Lee. Another one that's completely different is called Spotlight Stories. And Spotlight Stories is about using your phone as a new storytelling canvas. And we've been very lucky to have some very famous animators, director, producers, Glenn Keen most recently, who was the creator of The Little Mermaid and a variety of other iconic Disney films, get excited about creating stories for this new medium. And the reason it's a new medium is because you can control the direction of the camera. You have gyros and accelerometers in your device. And so the phone can act as a window into the animator's world, as opposed to the animator fixing your point of view and sort of rendering all the scenes in advance and saying, this is exactly what you'll see. You can explore the story. And the story can be nonlinear and can evolve with what it is that you're interested in. And so you can follow different characters. And the story can fork and then hopefully converge in the end to make it a good story. So there are a couple of episodes of Spotlight Stories that I think right now is only for the Moto X. But there are a couple of different episodes for it. And they're pretty interesting. So I encourage you to have a look at it. It is a fundamentally different and really delightful experience. And we don't participate in the Google Internship Program. So if you want the experience from the movie The Internship, you should apply to the big Google Internship Program. We do, however, hire people for the summer and people in a variety of other capacities. Flexibility in the way we access talent is one of the staples of the operating model. And so if you're interested, it's my last name at Google.com. Thank you. I want to thank everyone for coming. I want to, in particular, the questioners for a great set of questions. I'll venture to say that this is exactly the sort of evening we had in mind in starting this series. And we have a very diverse and, I think, terrific lineup coming. But ours also is an open process. And I would very much welcome suggestions about other figures you would like to see. I've yet to approach anybody from any realm who wasn't interested in coming to Purdue, even those who didn't grow up in West Lafayette. But I will also venture to say that no matter who comes to this series in the future, we're not going to have a night any more interesting than the one we just enjoyed. Paul, thank you very, very much. And we're proud to be indirectly associated. Thanks. It was terrific.