 All right, good afternoon. Welcome to you all to this panel, which is going to talk about grand things, technology and policy grand challenges for this thing called the internet that we may have heard about and we have a very distinguished panel of speakers here. So I'm the moderator. My name is Saurabh Bakshi and we will have David Love, Mike Attala, Sanjay Rao, Duck Comer and our distinguished visitor Robert Khan as panelists in here. So I'm going to start out with some introductory remarks. So how many of the top 25 inventions of the year do you think are significantly reliant on the internet? And of course, the term significantly reliant is subjective. So I used my subjective liberty to come up with that definition. So if you look out over the last few years, last 15 or so years, we find that there is a number which is not huge but not insignificant either of these top 25 inventions that are reliant on the internet. And it's been going up some, but I don't think we can read too much into that trend yet. So what are some of these examples of internet-reliant technologies that you can see? I just picked representative samples. These are clothes which talk and which communicate wirelessly. There's a surf board which you can actually surf while you are surfing. This is something which we have all seen when we go to the farmers market and we don't have to carry cash with us anymore. And this is actually the interplanetary internet which has been in the works for a few years now. And in 2017, there was also an invention which is very thankfully not reliant on the internet at all. So that's just a quick run through. Now let's think about the reach of the internet. We all know it's phenomenal. In terms of quantitative numbers, the Pew Research Center has been tracking this for since the early 2000s and at that time about half of the US adult population was using the internet on a regular basis and today about 90% of the population is using the internet. Now you would see that these different lines correspond to the different income brackets and what is also, I think, a desirable, a heartening trend is that the gap among these different income brackets has been shrinking. Now let's look at the reach of the internet in terms of the percentage of US adults who have access to the broadband at the home. And beginning in the year 2000, it was practically nobody. And now we have about three quarters of the US adults have access to the broadband at home but you would notice that this curve which was increasing quite fast in the early days has slowed down some. And another point that gives us some food for thought is that the gaps that you would see across the income categories are quite a bit larger than you see with respect to the general internet. Another point which gives me some cause for thought is the state of the freedom of the internet. And this is a map which is brought out by the Freedom House. They've been doing this for about 15 years where they profile the state of freedom in the different countries across the world. How free is the internet? So you would see that the green countries where it is free or mostly free are actually a very small minority of the countries and also a very small minority of the global population that falls in that bracket. Then you have some pockets which are partly free and large pockets which are not free. So if you count up the numbers today, less than 25% of the internet users live in a country where it is free or mostly free. And what should also give us some cause for concern is that over the last seven years that the Freedom House has been doing this, the level of internet freedom has been on a downward slope. However, the Rebel Alliance has been striking back through technological marvels, I would say things like telegram, signal, or where you can exchange information in a secure manner. And at the same time, it is quite usable where you don't have to have a PhD in computer science to be able to use these privacy preserving technologies, I think is a very heartening trend. I looked at a recent survey done by the Pew Research Center where they went out and asked US adults, as to what level of intervention would you tolerate from the government or from technology companies in preserving internet freedom or in preserving sort of protecting you from false information. So when it comes to the government, about four in ten say the US government should take steps to restrict false info online, but the majority, 60%, says we should not. The freedom of information is much more sacrosanct. Interestingly, when you ask about, are technology companies allowed to make these kinds of interventions, the numbers are almost flipped, where about six in ten adults say that it's okay for technology companies to put in these kinds of protections to protect us. So I thought that's an interesting trend and this is from a very recent March 2018 survey. So I'll pause there and with a few open ended questions that have been marinating in my head, what are the bright spots that we see in technological innovation on the internet? Are we losing our capability for doing revolutionary changes? And are we forced to do only small evolutionary changes on the internet today? What are the big policy battles that we see coming down the pipe? And which are the ones which we as academicians, researchers, should take up cudgels for and fight these battles? And then, will the internet growth in different countries go hand in hand with the growth in GDP? And I couldn't resist the temptation of bringing in GDH as well. Does anyone know what GDH stands for? Yes, that's exactly right. So there is actually a reasonable, serious scientist do create indices for the gross domestic happiness. And I would like to see what the correlation between the two are. With that, I will hand over to David, who is going to lead with some introductory remarks. So the structure is each of the speakers has some introductory remarks followed immediately by any questions that the rest of the panel has to ask. And that should leave us with enough time at the end for Q&A and discussion. All right, so I'm mainly a single processing communications person. So my interpretation of internet future you'll see is highly skewed in that direction. So the interesting thing is that really since the introduction of the iPhone, there's been this major shift in how people can access the internet, the last top that's beating down their data. So this is worldwide data. There's this trend where in about 2016, mobile overtook desktop or laptop access. When you think about that, this mobile access is coming over the Wi-Fi. It's coming over a cellular link. And that's using Spectrum. That's consuming power. That's relying on all these standards for communications that all of you have heard of. So the question is how are service providers, equipment manufacturers going to keep up with this? Spectrum is limited. You have to rely on technological evolution. All right, so the current solution, at least for the next five or so years, five to 10 or 15 years, from the operators and equipment manufacturers is five years, which what that really means is all of your phones are using standards associated with 3GDP. That brought you WCMA, LTE, and the newest standard series is 5G NR, 5G New Radio. And they're looking at some very aggressive requirements for the 5G radio. It's just to highlight a few of these. 10,000 times more traffic, a latency of less than one millisecond, which is really, really low. If you think about that, something traveling at speed of light, you can set your radius for where that information has to emerge. Peak data rates in excess of 10 gigabits per second, and also average or low end user data rates of 100 gigabits per second. So what's driving all these demands are future applications, and they're really starting to become more and more diverse. You still have your traditional high throughput applications which have become more demanding, higher resolution video, virtual reality, gaming that consumes large amounts of data, but what you're starting to see are new applications like self-driving cars, vehicle to vehicle, vehicle to roadside. There's currently this 5G automotive association which is really trying to push out DSRC as the future solution for how vehicles are going to connect. Because there will be some sort of connection in the future, is it going to be this sort of DSRC standard people are talking about, which is pretty much a low rate type of solution, or is it going to be something that's driven more by the cellular properties. There's also these Internet of Things type applications, where you might have thousands of devices in your home in the future with some sort of Internet connectivity. So in terms of how the actual network will work, you're going to have all these different services, right? Your traditional services, the enhanced mobile broadband, just giving you better and better throughput to your smartphone or other device. You're going to start having these mission critical services. That would be some sort of telemedicine that could be vehicular services, and you're going to have these Internet of Things applications. And from a radio point of view, really what's giving you this new throughput, these new latency benefits, are new bands coming out of them. So in terms of throughput, if you give me more bandwidth, at least in the regimes we're looking at, bandwidth scales linearly with bandwidth. So people are looking at bringing new bands online in a 20 to 100 gigahertz regime. So right now, all broadband wireless is pretty much 6 gigahertz and below. That would be Wi-Fi, your LTE. These 20 to 100 gigahertz bands have been lightly used or in some cases unused, and that's been because they have some propagation issues in terms of path loss, and also because the hardware has been too expensive. That's been coming down in price. So what 3GDP, and really I believe to a point, has done is look at these new bands where you can get really large bandwidths you can get to users. The millimeter wave is really the key innovation for 5G. In terms of the deployments, you're still going to see your macro cell which is this really powerful, lots of antennas base station using some of the research ideas like massive MIMO things like this, and that would be pretty much operating with sub 6 gigahertz. But you're going to start seeing smaller cell millimeter wave panels, maybe 50 to 100 meter kind of radius, where they might have hundreds of antennas doing beamforming and giving you a very, very narrow beam. And on the other end of that, you'll have these small cells which are more like a Wi-Fi kind of system, and you're going to start seeing more and more interoperability between Wi-Fi and your cellular type of service. Now 5G, though, is really not going to solve all your problems. There's a lot of hype right now, but if you actually pull out the standard, go to 3GDP's website, 5G is very, very similar to LTE. The physical channels are the same, the control channels operate pretty much the same. All the ATNEC, it's all like an LTE evolution. The really only big change is the millimeter wave. But the key here, though, is that the spectrum is limited even though we have all these new bands coming in line. And really what you're going to have to do is start leveraging more of Cooper's Law, which is if you're an operator, you want to increase the bits per second per hertz per kilometer squared. And to do that, you really have to shrink cell sizes, which is where you get your radio closer and closer to a high-rate fiber optic lane for backhaul lanes, and you start increasing the spectral efficiency using not just point-to-point techniques, but multi-point-to-multi-point techniques. And because 5G is really more of an evolutionary rather than a revolutionary technology, I think very quickly you're going to see a lot of hyphers 6G in the audience. And that will be where you're looking at underutilized frequencies, still in the 20 to 100 gigahertz, but also in the terahertz, which is really 300 gigahertz and above, which also had those same issues. It's not cost-effective right now to design a radio. You're going to have leveraging of low-rate connectivity, and I could have low-rate quotes, because low-rate connectivity might be your WC e-mailing or your LTE, which today is your high-rate connection, and that can detect some sort of cloud service, and there's lots of talk about AI-inspired communication, where you can take your network, the radios are feeding data to some sort of cloud system, a database system, and they can go through and learn about how the network should operate, adapt the radio link in the network link. And then another thing that's interesting is large networks of smart devices. As I mentioned, there's going to be large networks of devices in your home, in your car, on your body. I think in the future they're going to work together even at the radio layer in a very, very tight way. It might do you that you don't have to worry about a specific device with 100 antennas, because you'll have 100 devices working together almost as a distributor, right? So that's everything I've had, and thanks. I'm Mike Tuchella, I'm a computer scientist at ECE, and my interests are in security and algorithms, so my talk may have such a slant. The challenges to me are obvious, that same technology, that same wonderful internet, and other technologies that have brought us such enrichment in our lives is also bringing a lot of mischief and a lot of misery. You've had your identity stolen, you know what I'm talking about, and I know people who want to do that. So how do you mitigate the drawbacks without attracting from the benefits? It's all too easy to propose things that will solve the problem, but are worse than the disease, because you're damaging that wonderful technology. So how do we do that? And I focus on three issues, and one of them is that technology gives too much destructive power to small groups, and what can we do to mitigate that? And the second one is that technology may be more difficult to enforce laws, and the third one is what I recall legislative capture that the laws are not doing what they're supposed to do. I'll give you examples of each, but first let me talk about the first one. I'll finish in less than five minutes. Never in the history of our species have we been in a situation like this, where such small groups of people can wreak such havoc on such large numbers of people. If you're someone looking at birth through the sweep of time and space, the one striking thing that jumps at you is that in the days of the Babylonians, 12 disgruntled people could do this much damage. Today, what can they do? They can just read the newspaper. So, and this is very unprecedented, very unique, and it really used to be very hard for disgruntled individuals to wreak global havoc, and it's all too easy now. The disruptive power of small groups has long, for a very long time, been an increasing function of time through the centuries. There has been nothing but increasing, and there are consequences to that. One consequence is you have many more countries than you have in the past. Have you noticed that? But that's one of the consequences of small groups being able to... The center cannot assert itself anymore. Now, in the past, to disrupt the lives of millions, it took large numbers of determined problem makers, invading armies, for example. I mean, Jantis Khan could not take a peek at your refrigerator. He couldn't look at your refrigerator. Now, anyone can actually improve their future. Now, why refrigerator today? But anyway, this is one of the issues I want to put forward. In fact, solutions to that and mitigations of that exist. If only the technologies and the proposals make in fact by Bob Khan himself a while ago, and others, if these technologies and ideas were deployed, they would actually go a long way towards mitigating these issues. The issue is that they're not deployed. So it's not entirely technological issues. It's incentives, economics. There's other things at play. So my... I list these not as necessary technological challenges, everybody challenges between technology, policy, law, and all that. The second issue is... So I have examples of that. So there are many examples, but nothing beats the stock market for wreaking havoc. So the two examples I mentioned are from the stock market. The flash crash of 2010, the market went down 10% for a few minutes and wiped out 1 trillion in value. But you might think, well, it went back up. Yeah, but the people who had placed they took everything down and never took it back up. So real money was lost. And in one person, one person, using an old EC from the parents who did that, accidentally did that. Accidentally bumped into what we call in systems theory, the positive feedback. You know what a positive feedback is when a change in one direction self-reinforces and you have instability and you didn't want to do that. Why? Because they had just made 40 million dollars, very undetected, spoofing trades and having multiple identities and manipulating markets. And they really didn't want to do that. Now, if someone who doesn't want to do that stumbles upon accidentally by a positive feedback loop that really does such havoc, think what a nation state could do who are very determined to actually on purpose do something like this. I would say something else. How many other positive feedback loops are there that we don't know about and have instability? I know a few that have never been exploited and they have to tell you about them in private, if you want, but not much. There's no time for that. The flash crash of 2013 is another example. You know Twitter account half, fake news released, market drops, someone makes money out of the drop. Anyway, let me move to the next issue which is what I call algorithmic malfunctions. I started calling them algorithmic malfeasance, but it's not. It's really algorithmic malfunctions. Well, this is malfunctions. The first one is about function. The Amazon is there. So this is why you have algorithmic software agents that were written by different people and couldn't predict the behavior of what they are. Because they didn't know what others were doing. And they interact with each other and so this is how you end up with soft pricing agents resulting in Amazon having a 24 million dollar biology textbook that was discovered, by the way, by a biology person who asked the research system why don't you buy us an extra copy of that book? From Amazon, he says, so how much should I pay? Whatever it costs, we charge it to my grant. And then the student goes and finds out that watches in this belief as the price goes up, over a few days it didn't happen instantly. You know, exceeds the million, two million, three million, 24 million for a biology textbook on the food floor. And you can see how it happened. These are some... Can you buy it? See, these are agents. So who's guilty of this? Nobody. Nobody's guilty. Because you can't sue me for stuff like this. I mean, when I'm writing the software, I didn't know what other people were doing and this is an emergent behavior of multiple software agents that is very, very hard to predict. And they have to pay the blame on someone and the outcome of it could be quite bad. Look at Uber's algorithmic search pricing. It results in a market-cleaving price that's much larger than if the drivers were acting independently. In the old days, we used to call that pollution. An antitrust clause and legislation would kick in. Today we shrug and pay and move on and we tolerate monopolistic behavior a lot more today. But this algorithm, the pricing algorithm, has turned competitors. There will have been 100 people working for Uber's, might have been taxi drivers, they're independent and they will be competing with each other and now, effectively, they're a cartel. Tell me what's different between a cartel and this? This is open on a micro scale, on a small scale. And the undesirable social outcomes don't have to be of that nature. They could be, for example, discrimination against people, unfair discrimination, hurting people. Who do you punish? Who do you sue? I didn't intend it, I didn't want to do that. It's not me, it's his software that caused it. No, it's not, it's not yours, it's not his, it's the combination of them. So, this is another one, and finally I want to talk about spam laws as an example of legislation and culture. I mentioned something about the legal system having failed us. It has, spectacularly. And that's a big problem. So, for years, the direct marketing association, DM&M, successfully blocked any federal law on spam. Absolutely any. Why? Because they're the spammers. Then, what happened is the states, California was a leader in that, passed effective laws on spam, really, really good laws. I can tell you about them, there's no time for that, you can ask me if you want. Fantastic laws on spam. And then they censored the heat, the DM&M. They woke up, they had to disable those laws. So, what did they do? They switched to lobbying for the passage of a federal spam that both preempts the state laws, makes them null and void, and replaces them with a very, very weak, ethnically neat, almost conically neat law. Why I think that? I can tell you more if you want. But it's weaker than any of the state laws it replaces, and it's weaker than any of the foreign laws on spam. And here we are, professors, trying to come up with technologies to filter spam when the real, real problem is not technology. The real problem is both those who are not doing what they're supposed to do. This is the form of capture. It's not regulatory capture, it's legislative capture. And after passing the federal law, what do you want the DM&M to do? They kept what they wanted, they supported the law, the law of the past. The first thing they do after that, they try to define it to weaken it as much as possible and make it, you know, again, you know, I can say more about this. But this is it for me and I, you know, get questions you can ask there. And then you need your questions for mine? All right, we'll go on to find them. Yeah, good afternoon everyone. It's an honor to be here, especially to be here, especially to be here, especially to be here, So where are we going and where are we headed? So this graph talks about how the video has become a very dominant part of the internet ecosystem. It's grown from about 20%, about 10 years ago to almost 80% today. Now from a user perspective, you could think of a Netflix style video where you're watching a two hour long video where you could buffer things in a few minutes in advance. And we've come a long way there, though you might still see these annoying buffering symbols and there's much work to be done in that space. But there are many new forms of video that are getting even more demanding. For example, this Facebook live where you could sort of broadcast small events such as this on Facebook live. And there are also interactive video where based on the user changing his or her orientation, you might change the video stream. In cases like this, there are especially latency sensitive and you don't have the luxury to buffer ahead to avoid variability in bandwidth. So what would the internet support for live and interactive video look like? Can we create a lightning fast internet? So there's a lot of talk with edge computing where you move services from cloud providers very close to users. Can that be part of the solution? And some of that might go hand in hand with what Dave talked about, 5G and some of that stuff. Now if you think about this from the networking perspective, we have this phenomenal growth in traffic over the last few years and these internet service providers who've done a lot of over provisioning the networks, that's perhaps not as viable an option anymore. And yet there is this great demand for multiple lines of service guarantees. So how do we go forward? Is this new technology that's coming out called software defined networking? Is that part of the answer? Do we need some kind of formal approaches to ensure networks behave and perform as desired? There's a lot of interest in this field currently just starting out in the networking community. A little bit on the policy front, there's been a lot of talk about net neutrality. So we don't want this situation where an internet service provider like AT&T kind of prioritizes Netflix traffic over say a low-end startup who's trying to innovate in video. On the other hand, to manage the internet for video, you definitely need to support some kind of prioritization. For example, I would rather have interactive video traffic prioritized over file transfer or even video on demand. And at least I think I would rather prioritize my traffic over my son's video game traffic. So what's a right framework to balance these technology and policy considerations going forward? Finally, I want to conclude with something we don't pay as much attention to as researchers, which is how internet videos revolutionizing education. And particularly notable is in the K to 12 arena where there is this national dearth or there's a feeling that we really have to improve computer programming education. Unfortunately, there just isn't enough expertise in terms of being able to hire teachers. It's hard to offer, right? So video is seen as a nice way to complement the classroom or school resources. Unfortunately, however, and despite some of what Sarah mentioned earlier, over 30% of Americans in rural areas lack access to broadband speeds or speeds needed for video, as per a recent FCC report. So I want to conclude with this question how might we bridge this digital divide and what could we do from a technical perspective? Imagining the future where though bandwidth speeds keep increasing, there's always going to be a wide heterogeneity in speeds. Yeah, thanks. Okay, the only thing you have to note here is the affiliation which is on there for Ragu. All right, so we were asked to talk about technology and policy grand challenges. So I've separated them out. Technology present. The internet really works pretty well. All of us use it every day. It all seems to work for most of us most of the time. So I don't think there's a present big challenge. I think it's all in the future. And by the way, if you want to know why it works so well, I can recommend an excellent textbook. So what are the future technical challenges? I think there is a big hurdle for higher bit rates, not just for wireless but for wired networks. When you get down to 100 gigabits per second, things start to be really funny and you have to call in the physicist to tell you why. And if you really want to go to a terabit, that's a big challenge. It's a much bigger challenge. DARPA had the gigabit challenge in the 90s. I think this is a much bigger challenge because it's more fundamental. So what else? Well, universal, high performance, wireless access. I'll defer to the experts on whether we can achieve that or how. But I think that's going to be a challenge. And the last one I put up is low power, everything. Power is critical. Go to Silicon Valley and talk to people. Everything from the IoT people, worrying about how many little tiny devices you'll have in a house to the data center people, worrying about how many gigawatts they can get from the power company. I think that's a big challenge. So what about the policy challenges? Well, I put at the top of the list laws in the legal system. I've done a lot of consulting. One of my consulting jobs was I was hired by these top lawyers in New York and I had to educate them about all the networking technologies, the internet protocols so they could take on more and more cases. They already had some lined up having to do with TCP, IP and the protocols. So I gave them an education and it lasted many days. And one of the days we were talking about stuff the topic was distributed down off service and I, not being a lawyer, they were all talking after I explained what it was and how it worked and what people could do to achieve this and how multiple computers were attacked. They were all mulling it over, oh, how are we going to present this? How are we going to take the case that has this? And I said to them very naively, I said, well, look, this is the moral equivalent of company A, hiring a bunch of people to stand in the doorway of company B so that nobody can get in. That's got to be illegal. And they looked at me like I was a child and they said, no, you don't understand. It's all based on case law and you cannot interpret an existing law that was made for blocking people in stores and they acted like I should have known this. Maybe I should have, I didn't know anything about law but so they explained we have to build a case for every little piece of digital technology. Now, let's go over to the other side, the guys that write the laws. I'm afraid that I have a very pessimistic view. They need an education. They don't not understand the digital world at all and so they can't begin to make a reasonable law because they just don't understand what it is that's out there. I put security and privacy on the policy side of things because I don't think all of that is a technical problem. So here I am, I'm a computer scientist and I'm saying we ought to throw this over the wall to the people that study public policy and social policy and sociology. But that's my view. I don't think we really need much more in the terms of technical expertise. We need a lot of policies. Society needs to decide what it wants. We are living in the days of the Wild West of the Internet. The trouble with the Wild West, if you go back and look at the history books, is it's really hard to tell the difference between the good guys and the bad guys. Are the security researchers these days being good guys by being black hat hackers and writing down instructions on how to break in the things and I have my view. My view is no, they're not. But they say, well, you know, it's the Wild West and anything goes so we better do something or maybe they'll do it first. So we have the Wild West, we need to think about how to come to an agreement on what we want and then how to put in some sort of regulations. Now, I point to this, there is some success. We have already regulated things like banking, we've regulated medical records, we've regulated educational records. There are very strict regulations that everybody has to adhere to and places like Purdue work very hard at that. They educate all the professors, they put in all of the technology to do it. Why can't we do that for other things? So I'm not in that field, I'm a computer scientist, I don't know about that stuff but it seems to me it ought to be possible. And if it's not possible to put in very strict regulations about letting data out, how about putting in laws that say if you use data in this particular way, if you try and blackmail somebody into paying you more money for your service because you've got data to release, then that's illegal. Alright, some more challenges. I think we need an increase in data filtering. Mike talked about spam. I have a plea on here for AI researchers to please come up with a way to get rid of all the spam I get. But I know it's hopeless and you probably know it's hopeless too. Do you know why it's hopeless? Because AI people can get paid money to figure out how to get the spam in. They don't get paid money to keep the spam out. So I don't know. Every day I keep thinking there's more and more and it's going to get worse. So another question is we used to have editorial processes and you know when Wikipedia came along people said we don't need editorial because everybody out there will make sure everything's right. And I have had situations where I found things that were wrong in technical specifications for protocols. My area of expertise. In one case I was working at Cisco and I asked some people there why don't you go out and correct it? They went out and corrected it. A month later they came back to me and they said no matter how fast we correct it it keeps getting changed back. Some company is making money as long as the specification says that their stuff is right. By the way there were 1,500 people in the group that I asked to go out and do it. So we need to increase some filtering. We need to decrease other data filtering. You know the story about Facebook and some of the experiments to make sure people get exactly what they want to read and see. It would be really nice if we had a way to do that and so that people could get always be guaranteed that they would get a balance. So there are my policy grand challenges. Unfortunately all of that has to do with society and society now crosses boundaries. We can't just do it in the US. So the last thing I'll say is we need to have international agreements. We need to internationally agree and I will point out that it's not hopeless. We have done this in the past with several things. So maybe the people that do international agreements once we understand technology can do international agreements for technology and the use of data. The other thing about understanding that protocols is that you have to understand a lot about what measurements about protocols mean. I remember in 1987, shortly after Gordon Bell had taken over running the brand new computer science directorate at NSF, he had commissioned some people to do performance testing of the TCPIP protocol. Well you know that the span of the technology has grown by what I estimate to be something like a factor of 10 million over the time. They were testing it on back then 50 kilobit lines and maybe they had a 1.5 megabit line somewhere although I don't think so. Anyway, he came back in and reported that TCPIP doesn't work and why not? He said, performance is terrible. I mean he couldn't hardly get any throughput. He had charts and so he said, really so let's go back and figure out what that was all about. Well it turns out what they were really measuring was the UNIX operating system and the way that particular protocol was embedded because today it's running on lines that operated at much higher speeds and it's not moving bits at two or three bits a second. It never has but that's what they were measuring. So you need to be very careful about that kind of thing. In the world of law that Doug mentioned very clearly I couldn't agree more that the importance of understanding relationship is incredibly important but the people that look at it from the point of view of law often look at things very differently than a technology person does and I think this is an area where we could start to teach in some of the schools kind of a legal perspective because often when we design things technically it's got to be compatible with the legal view. If we were just designing networks to move bits the IP lawyers don't typically get involved but if you're trying to do something with Disney's movies or the like the first line of resistance is the lawyers because they really control all the IP licensing. So if you were, I can speak with some authority here because my wife is a lawyer her field is in particular intellectual property and so she tends to look at things in a very different way than I would have before I knew her. So when you think about it over the radio transmission if you're a radio engineer you think about a signal in the modulation and how you're going to detect it and feed it and amplify it and all of that stuff and she's looking at it and saying there's actually intellectual property going over this line and who owns it and where the license is and it's a completely different analysis it's not like can you do it technically but who authorized you technically and where the authority is and all of that sort of stuff and that carries out into so many other aspects like we do sensor nets and you might think that's really simple you put a sensor connected to something and you can build it and then either do the job or it won't and she's looking at it and saying well where's the information flow and who owns it and did you get permission from the manufacturer of this device to build an interface to it and see it's a totally different view of the world and it's one that's really important I think they convey in the engineering context because the engineers they build these things but they've got to be compatible with all of that you know I'm reminded from your comment about SPAM in 2005 Peter Freeman who was then the head of the NSF size directorate that tried to put together a new program called CleanSlate Internet and you know I sat in on some of those briefings I think this was more an effort to kind of build up the funding base within the computer science directorate which had not really been able to compete with some of the other programs that got very large amounts of funding like the people building you know deep space observatories and Antarctic Earth Stations and the like and so the justification for their CleanSlate program which had a lot of money potentially associated with if they could get it basically was based on three problems with the current internet one was too much SPAM so Ergo CleanSlate Internet number two too many viruses Ergo CleanSlate Internet and number three not enough security Ergo CleanSlate Internet and so that led into a very interesting set of discussions which is even if you had a CleanSlate Internet which I didn't think was possible to do anyway any more than CleanSlate Earth what are you going to do to prevent SPAM on your CleanSlate Internet so if you've got a strategy for dealing with it there tell us what it is we can try it on the current internet and see if it works or figure out how to tweak it and if you're worried about viruses and worms and all of the attack kinds of things then you don't need a CleanSlate Internet you want a worse Internet because it's spreading these things around too well I mean it's working perfectly fine for this you need to be able to work on the applications and operating system which is where the real vulnerabilities are and that's a job for the industry that wants to tackle it and then finally if the issue is security I know a lot about this because I tried to deal with some of the original security efforts and we did some things back in those days to provide encrypted traffic end-to-end over the Internet and the Internet but the problem there was this tension between the need of the private sector for protection in its commercial dealings and the lack of any desire for the government to have key escrow or other back door ways of getting into their data otherwise it isn't real security for them as far as they're concerned it's a contact with the government versus the need for law enforcement and national security at large and that's not a problem that's a solvable problem because that problem will exist at any stage of society it's like can we find a solution for political problems forever in the future and the answer is it's not even a well-formed question because that's going to be a subject for discussion and debate going on as far as we can so those are just a couple of things I wanted to mention about that so a couple of issues that are really important to think about one is the question of since I've been focused on infrastructure a lot in my career although it's all technical and building things and making them work I mean I think your book helped but I think what we built actually was as important if I can put it that way how do you get infrastructure adopted? so we actually have a set of manuscripts that are fairly widely used in teaching mainly in the graduate schools how did the railroads get adopted in the country and you find that they were incentives and there are generally roles at government play and private sector play and it's all about funding and the like so for the railroads we created a number of institutions the state commerce commission I think came out of that a lot of the national accounting schemes because we now have larger distribution and other infrastructures came out the electrical infrastructure I mean when that was first introduced it didn't just show up everywhere instantaneously people were saying well you know what's the motivation for building it people weren't thinking about cooking in their homes or eating in their homes furnaces and the like and I think you can go back and look at various motivations but one of the big motivators was rural or outside outdoor electrical lighting people used to use gas lamps and having lighting that could be on persistently was good for public safety and the like so that was one of the things and they got public money to go in to help build it it was a smaller extension to get it into the homes because they had basic distribution systems worked out and the like I could talk about water and banking and other things like that but they're all very interesting questions and I think we still have it today in building out new infrastructure and what you find that even when something is brand new the only way you ever have relatively easy sledding you have the money and you're in some brand new idea that there are no existing businesses that have vested interests so that's why the internet was fairly easy to do early on I remember giving a lecture at the national academy building in washington to a group of I think it was an intel talent search or the westinghouse talent search they had these 15 to 17 year old high school kids with their science projects buying for first prize and this young lady got up at the end and she had a long list of questions she wasn't a million questions but it was a long list and first on her question list was how did we manage to convince all the governments of the world to let us build the internet now from the vantage point of a 15 year old or 17 whatever this is not an unreasonable question because they think about authority and the like and I said you're not going to like the answer to my question except for the handful of us that were actually working on this as a science experiment almost nobody else thought it was a good idea and so they paid no attention because they didn't think it was going to lead anywhere until it did, little by little one of the things that you have to realize is that whenever you try and build infrastructure that neighbor's vested interest of existing companies they're going to take a lot of interest in what you're doing because infrastructure tends to tilt the landscape in favor of the new infrastructure especially if it's got properties that older things don't have now one of the things I'm very interested in and which I just gave a lecture on in the other building was that we understand that the internet was really good and designed originally to move bits from A to B easily from computer to computer without the user having to figure out where the computer was and what network it was on and what protocols they used and how to route the data and we figured that once the bits got there then it was up to the user to figure out how to work with the other machine if you're looking for information you got to go find it and the like things have happened since then that have eased some things but the more general question of how you manage information such that an investment in information is good for the long haul is not a problem that's been solved in terms of deployment in the larger community and I think it's one of those things that we need to pay a lot of attention to if you're only interested in putting up information and having it around for a fairly short period of time we have strategies right now that people use for that all the time I do myself the web is great for that sort of thing but ask yourself if you wanted to get let's fast forward you're from a future generation and you want to say what were the rules on student behavior at Purdue University in 2020 where would you look I mean you're probably not going to Google anything and find it and in fact Google may not be around and you might want to find information that you have particular rights to access but it's not public information and you're not going to find that through any so you need ways to deal with some of these issues and I think once one learns that managing information is really important you're going to find we're making a lot of investment in that and you don't want to have to keep making that investment year after year after year so you know people will often say well we have databases today isn't that good enough put your information in database you don't want to have to keep reinventing it and recharacterizing and creating more metadata and the like every time you move it from a database to another in principle you could do that but that's really not a very good idea be very nice if you could just take the information codified in the appropriate form and I'm particularly in favor of this digital object architecture which we've been working on for quite a while which you can find references to on the net and go get the slides and the lecture I gave take the objects and move them from one system to another and they're going to take with it all of the associated information today the database knows about who can access it they know about Providence and all of that sort of stuff but it doesn't come with the information it comes with the system that's managing it a lot ought to be in the information itself and therefore wherever it shows up you ought to be able to access it one of the more a really good digital object architecture is one that lets you interact with the information itself and historically everything that we've ever done in the networking field has been about the technology you know so on ARPANET we had identifiers but they identified wires okay so if somebody in a hundred years say well it was on the computer attached to this wire on the ARPANET that wire is long gone the computer that was attached to it long gone so even when we went to IP addresses which kind of talked about individual machines that's not healthy either because that machine is probably not going to be here in a hundred years even if IP addresses are here in a hundred years and they might not be either and then finally when the web came along they were talking about files you know now they'll probably go along and get to more identifying the information that's in the files but this digital object architecture is all about the information itself and identifying the information regardless of what technology used to store it and that's like the internet was all about ways of interconnecting technologies that didn't exist necessarily now or in the future the final thing I just wanted to mention is that all of the efforts that we have spent on trying to deal with cyber security and all the other kinds of attacks have led to a kind of difficult race to look for bit patterns that can provide problems in operating systems and applications we don't deal with these threats at the right level of granularity so can you imagine being hired into border control where you're told look for photons and electrons crossing the border those are dangerous and you look in particular combinations of photons you want a way to deal with it in larger layers of aggregation so people that's a lot of photons and electrons potentially ships, planes, crates maybe even birds, pigeons things that have larger granularity where you can see the information and you have a chance of understanding what it is and so even if you ask people to describe their information say what type of digital object is this that you're sending then you have some choices about what to do with it you may want to know who it's coming from and therefore you may not want to deal with them at all or if it's somebody you knew they sent you something and they said okay this is a digital object that represents a song you might say okay it came from my friend he usually gives me good recommendations but to my song player now you better trust your song player isn't an attack item that's going to go back and attack you behind the scenes but let's assume you've got a cooperating song player what's the worst that can happen is you get a lousy song or no song at all maybe he sent you static but if he marked it as an operating system upgrade you're probably not going to feed it to your operating system and say upgrade you might so a lot of the cyber security concerns that we have today I think could be dealt with by dealing with this information exchange across the boundaries at a higher level of granularity and I think this is something that the research community should pay careful attention to how to describe it what strategies because if we can deal with it at a higher level most of us have no way of dealing with cyber security threats ourselves we don't even know how to think about it there are a few experts to do but they're fighting an uphill battle all the way so I think I'll stop there so we have the princely sum of minus 4 minutes for a panel discussion so I would lean on our panelists if he could extend by 10 more minutes so let's have the questions you can direct it to specific people in the panel and I would ask the panelists to be very direct and succinct in their answers too so we can go through a few of these questions please go ahead you were mentioning earlier sort of the question about the government needs to know for legitimate national services but this question changes when you think about the governments that are not necessarily democratic to protect the privacy of their citizens so more authoritarian how do you think the technology earlier in the day you mentioned how technology can have democratizing effects on lower levels in societies as well so do you see those democratizing effects of technology winning out in authoritarian regimes or do you see the consolidation of power I don't see technology trumping political vested interests the reason the internet works well today is because it's a grand cooperative collaboration every country in the world if they wanted could try and unplug themselves from the internet and there's nothing that would stop them and their citizens would face the consequences that's true you can't stop every radio wave from coming across your boundaries and you know who knows what could be done surreptitiously like that but politics is what it is and I don't think technology per se is going to change the decisions of some of the more repressive regimes that's my personal opinion now you could say as a pushback well we had all these social media networks that helped to facilitate the Arab Spring and you know it's true technology can help but for everyone that's helpful you can probably give a counterpoint where it can be used for nefarious purposes so it might help the bad guys and so I think there's no solution that's a technical solution that just fixes the problem sorry to say I wish they were so for Dr. Khan so you seem to suggest like these digital identifiers the architecture of that is sort of dynamic so you should be able to query them interact with them with some you know restraining in terms of weeds and rights so is that something that you have a matter where we have a dynamic relationship with these DOIs? the DOIs are the identifiers that and that term DOI is a trademark of the DOI foundation so those are all handles in the context of the way I think of identifiers which is a more generic term or digital object identifier which is the longer rather than the jargon form I mean I've always thought that the digital object architecture was about an infrastructure to structure the information and make it available but it was never an attempt to try and specify all the uses of it other than the basic ones like if I were to try all the uses of the electrical infrastructure I wouldn't even begin but I might say I'll tell you about a plug that can go into the wall plug it and get out of the wall and get to the application developers to figure it out so for example I always thought that one of the most interesting applications of it were if you built simulations that had rules for how the things should interoperate you could build objects of different kind, digital objects put them in different places the identifiers would be known or programs could find out about them through metadata and now they could start to interact with each other for purposes of carrying out interesting distributed simulations and you wouldn't have to figure out how do I talk to that one how do I know about it it starts to give you some primitives for doing it just the way in an operating system you have to figure out your own file system because it tells you how to store the stuff and how to access it but then tell you what to put in the files or how to use the files other than how to open it or close it that's the level in which I approach these things architecturally because the number of applications is essentially unlimited what you talked about is actually only one there are lots more that you can imagine even within the same genre of dynamically interacting with these digital objects next question I have one to mix things up I'll ask it to another panelist Mike I'm looking at you you mentioned all these security and policy challenges as well so what I was wondering is do you see solutions coming from top down government mandated ones or do you see more decentralized solutions but there isn't a central entity like Bitcoin being a popular soup de jour of this do you see that as being more success more likely to succeed I wouldn't bet on centralized clean power for best interests that were mentioned by Dr. Khan there's too many people too many people have too much at stake and they have legislators and regulators on their side through various mechanisms one being revolving door there's others like campaign contributions the point is that you see the reason I don't have much faith in that is these people have exempted themselves from the rules by which we abide here at Purdue if you subject to them the Purdue conflict of interest rules and ethics rules and they were at a university acting the way they act the university would roast them alive if I acted that way here at Purdue you know where I benefit indirectly no matter how indirectly from a decision I make Purdue has no sense of humor when it comes to that and nor should they have any sense of humor when it comes to that it's an agency problem it's a principal agent problem we are the principals we have voted for agents we have been betrayed by the agents it's well known in economics a principal agent problem and this is why what happens is to give us the appearance that they are serving us so they make a big fanfare about the ridiculous anti-spam law but the reality is otherwise so it's all a deception it's a con game I'm on your side I convince you I'm on your side while at the same time I act against your interests so I think it's going to have to be bottom up that would be my bet unless some change happens where next question is it probably time for one other question or if anybody else wants to weigh in on this how large and global will become we don't know I mean I think the fact of the matter is that this started out as a research endeavor a bunch of us trying to figure out how would you get different kind of machines to work together that problem was largely dealt with at the level of moving the bits on the ARPANET because we had a lot of different computers there and people have studied the protocols at some length the internet was just kind of an important step up but it had the effect of decentralizing control so it wasn't just one organization running the whole thing and we had to create all the social structures to enable that and those were interesting challenges as well in fact I think the creation of those social structures was really responsible for the ability of the system to evolve over time whether the IETF or whatever else but even there it was within the research community and we were not harboring thoughts that this was a megalomaniac activity to take over you know the world it was just you know it was a good idea just like I would articulate about this digital architecture it was a really good idea and one by one if we can get places to recognize it and do something about it maybe Purdue will be such a place maybe not but it will be in different places that you see pockets of expertise building up and then suddenly you get you know by the mid 1980s most of the experts in networking were involved with the internet related kinds of networking activities that were not on commercial nets because they weren't commercial nets connected in fact the first of them really started out a few years before very specialized kinds of things and then you know all of a sudden by the late 1980s when I participated in lots of meetings that were where did that come from I mean all of the people who wanted to do something new suddenly found that the real user community was on this other network there was a decision made to go with international protocols Doug you've probably written about that at some point and you know they never took off even though everybody said when they come to switch to them because there was no reason to switch I mean the internet stuff worked well enough that nobody saw any reason to switch to something else and it kind of kind of grew where it was personally the first time that I kind of realized it would go mainstream was probably in the early 90s I mean I saw all this other stuff happening but it seemed to be more localized more research oriented I didn't use these networks for commercial purposes but there was a bill passed in the U.S. Congress in 1993 it was called the Boucher bill or after a congressman from Virginia named Rick Boucher that opened up the use of the NSF net to more broad commercial use they recognized the value of that and it required an edict like that a law like that because these are government facilities or government supported facilities and so now suddenly other parties could join in and I came home one night and I saw on what was then I think the NBC Nightly News was probably Tom Brokaw's thing and now you can communicate with us through this new medium and guess what it was email and I said oh my goodness and suddenly thought of me this thing could be even bigger but you know NSF got out later and then the industry got in and who knew that you'd have this wireless revolution coming later because it wasn't 1991 and two we didn't have a big wireless activity I mean when did 2G really show up it was the 1990s maybe 1995-96 some of the early cellular stuff we had walkie-talkies before but that was sort of something else so sort of gradually kind of snuck up on me but if we had set out to create an internet that was sort of the global thing it is today I guarantee you we would not have accomplished it so we've run out of time in the 44 years since Bob Khan, Vin Sark, wrote their seminal PCP IP paper we have I think the internet has grown by as you mentioned a scale of 10 trillion I would say 10 million but a mic right away trillion so let's hope it grows by that scale in the next 10 years and we'll be back here in this panel discussing what we got right and what we got wrong you know what my hope is that somebody will figure out how to do things even better for some purpose and enable things that for whatever reason we never thought about to even go there I mean AT&T was probably not thinking about the internet back in the 50's and 60's they could have made it happen they had the capability I started my career at Bell Labs it was a great institution but it wasn't a business and it didn't become a business until people created new applications and those new applications really required something like a new capability that was in place and people reinvented over and over and over again so that new things are possible that we never thought about before that's when people ask me where do you see the internet in 10 years I say well you can do the linear extrapolation as well as I can you'll have more wireless devices in higher bandwidth and you'll have maybe we'll have some more digital object but tell me what we're going to have that nobody is thinking about today if nobody is thinking about it today there's nobody to ask somebody will come up with an idea and try and do something and they'll try to say can I piggyback on something that's already out there make use of it to make that happen in the web I mean the web didn't show up because Tim Berners-Lee built a brand new communication capability he piggybacked it on top of the internet which gave him this broad scale and more power to him because he figured out a very important thing very effectively who's going to make that leap in the future who's going to see the opportunity when we started to push the internet very few people thought there was a need for this kind of thing I'm talking about digital object stuff and even when people say it that's a really good idea unless they're prepared to do something about it it won't go nowhere and that's what I do I come up with things I think are really interesting ideas that have no counterpart in what we were doing before I don't want to reinvent the web I don't want to reinvent some other technology that somebody else has done as long as there's a community out there working on it people often ask me why don't you work on networking anymore because I'm working on extensions of it in other dimensions and when you've got 1,000 or 10,000 really smart young people on it my job is done I want to do something new that nobody's thinking about put that on the table see if people don't think about that so on that happy part of exponential growth let's hand the panelists