1 00:00:02,590 --> 00:00:04,590 Good afternoon, everyone. 2 00:00:04,590 --> 00:00:10,080 I'm Chris Shipp, the acting chief of staff at the USPTO, it is my pleasure to welcome 3 00:00:10,080 --> 00:00:17,199 you to the Clara Barton auditorium, and thank you to those of you joining us on line. 4 00:00:17,199 --> 00:00:25,410 If you are doing other time technical training and in the room, sign up in the back to make 5 00:00:25,410 --> 00:00:29,170 sure that you get the proper credit. 6 00:00:29,170 --> 00:00:34,320 It is my pleasure to welcome you to today's speaker series, a collection of special talks 7 00:00:34,320 --> 00:00:38,820 featuring some of the world's leading minds in invention and innovation. 8 00:00:38,820 --> 00:00:44,330 The speaker series was started by our director Iancu and created wonderful opportunities 9 00:00:44,330 --> 00:00:49,000 for us to hear from some of the best and brightest people alive today. 10 00:00:49,000 --> 00:00:53,989 Without further ado, director Iancu, I'll turn it over to you. 11 00:00:53,989 --> 00:00:55,499 >> Thanks. 12 00:00:55,499 --> 00:00:57,180 Thank you, Chris. 13 00:00:57,180 --> 00:00:58,870 Welcome, everybody. 14 00:00:58,870 --> 00:00:59,880 Welcome Dr. Cerf. 15 00:00:59,880 --> 00:01:06,829 I have to say folks don't form lines to take pictures with many people here at the US PTO 16 00:01:06,829 --> 00:01:08,829 but it was good to see that. 17 00:01:08,829 --> 00:01:15,020 It says something. 18 00:01:15,020 --> 00:01:17,499 This is our fourth installment of the series. 19 00:01:17,499 --> 00:01:19,240 We started it last year. 20 00:01:19,240 --> 00:01:25,689 Dr. Cerf was one of the first people I called actually, I don't know if you remember, but 21 00:01:25,689 --> 00:01:28,299 I called you directly and we spoke a little bit. 22 00:01:28,299 --> 00:01:32,950 He immediately said he would like to come and we had all sorts of scheduling issues, 23 00:01:32,950 --> 00:01:39,579 but here we are better late than never, very happy to have him here. 24 00:01:39,579 --> 00:01:46,929 I say often that through the doors of the PTO walks our future. 25 00:01:46,929 --> 00:01:54,579 Every day, as employees of the PTO, examiners of the PTO, we get to come across our desks 26 00:01:54,579 --> 00:02:01,239 some of the most interesting, most advanced ideas and innovations and created by some 27 00:02:01,239 --> 00:02:06,950 of the most interesting people in the world, folks who literally change the world and the 28 00:02:06,950 --> 00:02:08,690 way we live. 29 00:02:08,690 --> 00:02:14,549 So it's so good to have an opportunity every now and then to meet face to face with these 30 00:02:14,549 --> 00:02:23,610 folks and hear from them directly, to remind us why we do what we do every single day. 31 00:02:23,610 --> 00:02:27,349 Dr. Cerf is certainly one of those people. 32 00:02:27,349 --> 00:02:33,129 Vint Cerf is widely known as one of the fathers of the Internet, we are going to come to that 33 00:02:33,129 --> 00:02:37,939 in a few minutes and get a lot of details from him on that. 34 00:02:37,939 --> 00:02:43,500 But it's without exaggeration to say that his work has literally changed the world in 35 00:02:43,500 --> 00:02:46,819 more ways than we can imagine. 36 00:02:46,819 --> 00:02:55,420 He began with a bachelors of science at Stanford University, that's okay. 37 00:02:55,420 --> 00:02:56,769 (chuckles). 38 00:02:56,769 --> 00:03:06,909 Not so bad, but then he peaked and culminated with a masters and PhD at UCLA. 39 00:03:06,909 --> 00:03:16,930 That is also where he met Bob Kahn with whom together they codesigned the elegant TCP/IP 40 00:03:16,930 --> 00:03:23,400 protocols, which allow data to be broken into packets, addressed, transmitted, routed and 41 00:03:23,400 --> 00:03:29,610 received, and as they are ushered through the Internet and have brought us what we know 42 00:03:29,610 --> 00:03:34,730 today as the modern Internet system. 43 00:03:34,730 --> 00:03:39,510 Entire new fields of scientific inquiry and technological advancements owe their existence 44 00:03:39,510 --> 00:03:41,349 to him. 45 00:03:41,349 --> 00:03:44,069 His achievements have been recognized worldwide. 46 00:03:44,069 --> 00:03:47,890 Let me touch on a couple. 47 00:03:47,890 --> 00:03:56,129 Bill Clinton in 1997 bestowed upon him the U.S. National Medal of Technology. 48 00:03:56,129 --> 00:04:04,290 In 2004, Vint Cerf received the Alan Turing award, considered to be the Nobel Prize for 49 00:04:04,290 --> 00:04:05,650 computer science. 50 00:04:05,650 --> 00:04:15,049 The Presidential Medal of Freedom was awarded to him in 2005 by President George Bush. 51 00:04:15,049 --> 00:04:19,160 Near and dear to our hearts here at the PTO he was inducted into the National Inventors 52 00:04:19,160 --> 00:04:22,300 Hall of Fame in 2006. 53 00:04:22,300 --> 00:04:29,860 He says that you, quote, you don't have to be young to learn about technology. 54 00:04:29,860 --> 00:04:33,150 You have to feel young. 55 00:04:33,150 --> 00:04:38,750 Let us now hear from somebody who feels very young. 56 00:04:38,750 --> 00:04:41,860 (applause). 57 00:04:41,860 --> 00:04:49,530 >> Before you even ask a question, I hope everybody in the room recognizes how terrific 58 00:04:49,530 --> 00:04:54,070 it is to have somebody like this with all this enthusiasm running this organization. 59 00:04:54,070 --> 00:05:01,900 It is a real joy to have you here. 60 00:05:01,900 --> 00:05:02,900 >> Thank you. 61 00:05:02,900 --> 00:05:03,900 (applause). 62 00:05:03,900 --> 00:05:07,900 It is easy to have the enthusiasm when we have folks like you come through here. 63 00:05:07,900 --> 00:05:16,080 Let me begin first by asking you, is it accurate to have a moniker such as father of the Internet? 64 00:05:16,080 --> 00:05:23,039 >> VINT CERF: Bob Kahn and I really did do the original design of what is today called 65 00:05:23,039 --> 00:05:24,389 the Internet, together. 66 00:05:24,389 --> 00:05:28,930 It was two hands on one pencil, in 1973. 67 00:05:28,930 --> 00:05:34,639 But it was in a context, just as all other inventions are in a context, so there was 68 00:05:34,639 --> 00:05:41,039 a predecessor network called the ARPANET which experimented on a wide area scale with packets 69 00:05:41,039 --> 00:05:48,110 which was funded by the defense advanced research projects agency, partly to figure out whether 70 00:05:48,110 --> 00:05:52,120 packet switching would actually serve well in the way it was intended, and second, it 71 00:05:52,120 --> 00:05:58,159 was to allow sharing of computing resources among a dozen universities that ARPA was funding 72 00:05:58,159 --> 00:06:01,400 to do research in artificial intelligence and computer science. 73 00:06:01,400 --> 00:06:04,360 This is way back around 1969. 74 00:06:04,360 --> 00:06:11,810 The success of the ARPA net project which had a lot of participants in it led directly 75 00:06:11,810 --> 00:06:17,760 to the idea of designing and building an Internet that allowed multiple networks to be interconnected 76 00:06:17,760 --> 00:06:22,680 and different kinds of computers to be interlinked. 77 00:06:22,680 --> 00:06:27,289 The answer is, for the specific Internet design, yes, I think it's fair to say that Bob and 78 00:06:27,289 --> 00:06:30,440 I are the fathers of that design. 79 00:06:30,440 --> 00:06:36,110 But I think everyone in this room must understand that any time you do something on this scale, 80 00:06:36,110 --> 00:06:40,430 a lot of people have to be involved, have to be committed to it, a lot of money was 81 00:06:40,430 --> 00:06:47,180 spent building the infrastructure, across a wide range of participants, whether it was 82 00:06:47,180 --> 00:06:53,110 the government initially that paid for the backbones at NSF, Department of Energy, NASA 83 00:06:53,110 --> 00:06:58,650 team as well as ARPA and then private sector. 84 00:06:58,650 --> 00:07:04,770 We may be the fathers of the Internet, but there was a wonderful sign that was put up 85 00:07:04,770 --> 00:07:10,740 in an operations center at MCI when I was the senior VP of engineering there, and it 86 00:07:10,740 --> 00:07:15,400 said Vint Cerf may be father of the Internet but we are the mothers who have to make it 87 00:07:15,400 --> 00:07:18,410 work. 88 00:07:18,410 --> 00:07:21,429 (laughter). 89 00:07:21,429 --> 00:07:25,770 >> By the way, you mention the people in this room, this room and the Clara Barton auditorium 90 00:07:25,770 --> 00:07:30,199 of the PTO is to capacity but I understand that we have more than 1,000 folks watching 91 00:07:30,199 --> 00:07:37,780 this on line as well, enabled by some of your technology obviously. 92 00:07:37,780 --> 00:07:41,430 Talk a little, let's go back to basics first of all. 93 00:07:41,430 --> 00:07:49,330 Describe what you actually came up with, what is TCP/IP, why is it that this is actually 94 00:07:49,330 --> 00:07:56,240 what has really created the Internet vis a vis what was there before, such as the ARPANET. 95 00:07:56,240 --> 00:08:03,789 >> VINT CERF: The idea of packet switching which is for all practical purposes electronic 96 00:08:03,789 --> 00:08:08,180 postcards, if everything you know about a postcard applies to Internet packets, they 97 00:08:08,180 --> 00:08:12,520 don't always make it through the network just like postcards are not guaranteed, if you 98 00:08:12,520 --> 00:08:16,780 put two postcards in the post box aimed at the same destination they don't necessarily 99 00:08:16,780 --> 00:08:20,010 come out in the same order you put them in. 100 00:08:20,010 --> 00:08:24,150 The Internet, packet switching can replicate packets. 101 00:08:24,150 --> 00:08:28,030 That is not a service the post office does as far as I know. 102 00:08:28,030 --> 00:08:31,719 But if you do retransmissions and they are routed in different directions, the recipient 103 00:08:31,719 --> 00:08:34,740 might end up with more than one copy. 104 00:08:34,740 --> 00:08:41,970 The packet switching idea did not necessarily assure any kind of guarantees of delivery. 105 00:08:41,970 --> 00:08:47,640 You had to add procedures in order to filter duplicates, retransmit when things were lost. 106 00:08:47,640 --> 00:08:55,130 In the original ARPANET design, we relied on a homogeneous collection of packet switches, 107 00:08:55,130 --> 00:09:00,330 to deliver reliable service, so that was the network interface was supposed to be reliable, 108 00:09:00,330 --> 00:09:07,890 the computers on the edges used a protocol we called the host to host protocol to deliver 109 00:09:07,890 --> 00:09:12,320 their packets into the network and the network would do its best to deliver everything on 110 00:09:12,320 --> 00:09:17,820 the other end, and there was end to end retransmission inside the network and filtering of duplicates 111 00:09:17,820 --> 00:09:19,960 and so on. 112 00:09:19,960 --> 00:09:21,360 That worked very well. 113 00:09:21,360 --> 00:09:27,950 But then, it worked so well that ARPA said I wonder what would happen if we decided to 114 00:09:27,950 --> 00:09:32,540 use computers in command and control. 115 00:09:32,540 --> 00:09:37,250 The implication of that, remember computers are getting smaller and smaller during this 116 00:09:37,250 --> 00:09:42,640 period, so they started out being big machines that fill rooms and they are in air conditioned 117 00:09:42,640 --> 00:09:45,190 facilities and in fixed locations. 118 00:09:45,190 --> 00:09:50,520 Then they come down to departmental computers for only $100,000 instead of $6 million, and 119 00:09:50,520 --> 00:09:55,370 there is work stations and desktops and laptops and so on. 120 00:09:55,370 --> 00:09:56,370 They are getting smaller and smaller. 121 00:09:56,370 --> 00:10:05,520 It gets to the point where in the early 1970s, it was feasible to imagine putting computers 122 00:10:05,520 --> 00:10:11,010 in ships at sea, in aircraft, and in mobile vehicles, and not just in fixed installations. 123 00:10:11,010 --> 00:10:15,900 But the ARPANET was designed out of packet switches that were connected to each other 124 00:10:15,900 --> 00:10:18,330 with dedicated telephone lines. 125 00:10:18,330 --> 00:10:22,600 You can't use telephone lines to connect the ships together because they get all tangled 126 00:10:22,600 --> 00:10:27,280 up, and the tanks run over the wires and they break and airplanes never make it off the 127 00:10:27,280 --> 00:10:29,280 Tarmac. 128 00:10:29,280 --> 00:10:35,330 Bob who had worked on the ARPANET, one of the architects, had gone from his company 129 00:10:35,330 --> 00:10:40,710 in Cambridge to ARPA in 1972. 130 00:10:40,710 --> 00:10:47,500 I had gone from UCLA to Stanford about that same time to join the faculty. 131 00:10:47,500 --> 00:10:53,960 He showed up in my office in the spring of 1973 and said we have a problem. 132 00:10:53,960 --> 00:10:56,660 My first reaction was, what do you mean, we have a problem? 133 00:10:56,660 --> 00:11:02,810 He said, well, we are serious about using computing and command and control, but now 134 00:11:02,810 --> 00:11:06,600 we are going to have to have radio based communication. 135 00:11:06,600 --> 00:11:11,910 He had already started working on a packet satellite system synchronous satellite with 136 00:11:11,910 --> 00:11:16,740 multiple ground stations, like an Ethernet in the sky, and these multiple ground stations 137 00:11:16,740 --> 00:11:20,020 would share a common packet satellite channel. 138 00:11:20,020 --> 00:11:27,990 Then he had a mobile radio system, the radios were designed by Collins radio in Texas, but 139 00:11:27,990 --> 00:11:36,160 the system was deployed in the bay area under the control or management of SRI International. 140 00:11:36,160 --> 00:11:39,980 He's already part of the way into a multi network environment. 141 00:11:39,980 --> 00:11:45,460 And by the way, a mile and a half away from my lab at Stanford was Xerox Palo Alto research 142 00:11:45,460 --> 00:11:53,140 center and in May of '73, Bob Metcalf and David Bogs invented Ethernet. 143 00:11:53,140 --> 00:12:00,020 That was a fourth packet switching technology to be lumped into this possible zoo of packet 144 00:12:00,020 --> 00:12:01,550 switching mechanisms. 145 00:12:01,550 --> 00:12:06,460 Bob says our problem is how to connect all these different networks together and make 146 00:12:06,460 --> 00:12:11,080 them look uniform to all of the computers that are at the edge. 147 00:12:11,080 --> 00:12:14,200 That was the Internet problem. 148 00:12:14,200 --> 00:12:17,760 Over a period of six months, we figured out how to do that. 149 00:12:17,760 --> 00:12:22,440 We designed a protocol for that, we called the transmission control protocol. 150 00:12:22,440 --> 00:12:27,411 It did all of the packet recovering and everything else that we had been relying on the network 151 00:12:27,411 --> 00:12:28,411 to do. 152 00:12:28,411 --> 00:12:33,070 It was important to do that, because there was no guarantee in the mobile radio environment 153 00:12:33,070 --> 00:12:37,800 that there would be reliability within the radio net, reliability of the satellite system. 154 00:12:37,800 --> 00:12:40,660 Ethernet was just a dumb piece of coaxial cable. 155 00:12:40,660 --> 00:12:43,300 It didn't have ability to retransmit anything. 156 00:12:43,300 --> 00:12:45,610 You had to do it all at the end. 157 00:12:45,610 --> 00:12:51,750 This end to end recovery retransmission filtering of duplicates and so on had to be integral 158 00:12:51,750 --> 00:12:52,780 to the design. 159 00:12:52,780 --> 00:13:00,270 We designed TCP, we wrote a paper in September, and that paper was published in May of 1974 160 00:13:00,270 --> 00:13:03,480 in the IEEE transactions on communications. 161 00:13:03,480 --> 00:13:08,990 But by that time, by the time it was published I was already busy doing the detailed design 162 00:13:08,990 --> 00:13:12,570 of TCP with a small team of graduate students at Stanford. 163 00:13:12,570 --> 00:13:19,320 We had some people coming from Xerox park who were part of the seminar. 164 00:13:19,320 --> 00:13:26,300 By the end of 1974, we had a very detailed design of the TCP protocol. 165 00:13:26,300 --> 00:13:32,800 Then we started implementing it in 1975, and we immediately discovered we made mistakes. 166 00:13:32,800 --> 00:13:39,950 We went through four iterations of the TCP design and on the third iteration, around 167 00:13:39,950 --> 00:13:48,370 1977, we realized that if we really wanted to support realtime communication, that all 168 00:13:48,370 --> 00:13:53,440 the hard work that this transmission control protocol is doing to retransmit lost packets 169 00:13:53,440 --> 00:14:00,550 and filter them out and duplicates out and manage control would not help us much if the 170 00:14:00,550 --> 00:14:06,620 application involved realtime communication, for example, radar tracking. 171 00:14:06,620 --> 00:14:11,190 You don't want to know where the missile was, you want to know where it is, and you don't 172 00:14:11,190 --> 00:14:13,360 care about where it was. 173 00:14:13,360 --> 00:14:16,770 If you didn't get that piece of information, don't worry about it. 174 00:14:16,770 --> 00:14:17,770 Forget it. 175 00:14:17,770 --> 00:14:25,010 So we split the TCP into two parts, the end to end part that worried about reliable delivery, 176 00:14:25,010 --> 00:14:29,900 and the low latency part that got the packets there as fast as they could and too bad if 177 00:14:29,900 --> 00:14:32,300 some of them got lost. 178 00:14:32,300 --> 00:14:39,190 IP was the unreliable portion, but it had the global addressing that was needed, that 179 00:14:39,190 --> 00:14:40,550 raises one other interesting point. 180 00:14:40,550 --> 00:14:44,480 >> Before you go to the other interesting point, on that question, you don't care where 181 00:14:44,480 --> 00:14:49,521 the missile was, you want to know where it is, you need to know right away, but don't 182 00:14:49,521 --> 00:14:55,190 you really need to know where it is accurately, because it wouldn't help you if you know right 183 00:14:55,190 --> 00:14:57,630 away where it is but that turns out to be wrong. 184 00:14:57,630 --> 00:15:03,460 >> VINT CERF: This reminds me of my early days at a company called Rocketdyne, I worked 185 00:15:03,460 --> 00:15:08,820 on the F1 engines for the Apollo program doing testing to make sure they would survive until 186 00:15:08,820 --> 00:15:10,030 they ran out of fuel. 187 00:15:10,030 --> 00:15:12,740 After that we didn't care what happened. 188 00:15:12,740 --> 00:15:18,450 To this position, well, let's be practical. 189 00:15:18,450 --> 00:15:19,450 (laughter). 190 00:15:19,450 --> 00:15:23,570 There was this question about positioning, position location and everything else. 191 00:15:23,570 --> 00:15:30,380 I remember somebody writing a explanation for this, the missile knows where it is because 192 00:15:30,380 --> 00:15:32,990 it knows where it isn't. 193 00:15:32,990 --> 00:15:39,200 The difference between where it is and isn't, we call error. 194 00:15:39,200 --> 00:15:41,970 They went on and on. 195 00:15:41,970 --> 00:15:45,731 The honest answer is that yes, you needed to have accurate information and for that, 196 00:15:45,731 --> 00:15:48,320 you needed timely information. 197 00:15:48,320 --> 00:15:55,190 The realtime aspects of IP were manifest in something called user data grand protocol 198 00:15:55,190 --> 00:16:01,250 which is parallel to TCP in the protocol architecture which was hierarchical and layered. 199 00:16:01,250 --> 00:16:06,610 Now we have learned from the ARPANET experiment, we carried that over into the Internet. 200 00:16:06,610 --> 00:16:13,920 The thing that Bob and I contributed in addition to the basic end to end protocols, was an 201 00:16:13,920 --> 00:16:18,470 addressing structure that would allow all the networks anything on any network to refer 202 00:16:18,470 --> 00:16:19,860 to anything else. 203 00:16:19,860 --> 00:16:25,320 We didn't know exactly when applications were going to be developed, and that was important. 204 00:16:25,320 --> 00:16:29,100 We did not specialize the network to do any particular thing. 205 00:16:29,100 --> 00:16:34,350 We just did it to switch packets around as conveniently and quickly as it could. 206 00:16:34,350 --> 00:16:39,010 We had a address structure, and the other thing which I think is worth mentioning if 207 00:16:39,010 --> 00:16:40,790 you don't mind my going on a bit. 208 00:16:40,790 --> 00:16:41,790 >> Please. 209 00:16:41,790 --> 00:16:47,220 >> VINT CERF: Is that we decided not to use an international numbering plan, and the telephone 210 00:16:47,220 --> 00:16:53,290 network assigned numbering plans at each country has a country code. 211 00:16:53,290 --> 00:16:58,330 Remember we are doing this for the defense department in the context of command and control. 212 00:16:58,330 --> 00:17:04,299 Imagine if we had decided to use country code numbering plans. 213 00:17:04,299 --> 00:17:12,649 Let's suppose hypothetically that we are faced with invading country X in a couple weeks 214 00:17:12,649 --> 00:17:18,550 and we have used a country code numbering planning for our command and control system. 215 00:17:18,550 --> 00:17:24,889 Can you imagine going over to country X and saying, hi, I wonder if we can get some address 216 00:17:24,889 --> 00:17:29,800 space from you, we are planning to invade you in a couple weeks, but our command and 217 00:17:29,800 --> 00:17:34,019 control system won't work unless we have addresses in your country to build the thing. 218 00:17:34,019 --> 00:17:35,330 That is not going to work. 219 00:17:35,330 --> 00:17:42,730 We said why don't we have a topological numbering plan, each network got its own number. 220 00:17:42,730 --> 00:17:47,320 It didn't matter where the pieces of the network were so any network could span any number 221 00:17:47,320 --> 00:17:48,320 of countries. 222 00:17:48,320 --> 00:17:54,210 The side effect is that the Internet doesn't know when its packets have crossed a international 223 00:17:54,210 --> 00:17:55,210 boundary. 224 00:17:55,210 --> 00:17:59,841 All of you know about the domain name system which came later in the early 1980s and there 225 00:17:59,841 --> 00:18:04,629 is provision for country code top level domains in the domain name system. 226 00:18:04,629 --> 00:18:10,750 But you notice there are noncountry codes, these are generic top level domains like dot 227 00:18:10,750 --> 00:18:15,960 com and dot net and dot org which span the entire globe. 228 00:18:15,960 --> 00:18:20,440 The system is global in its design and spirit. 229 00:18:20,440 --> 00:18:24,750 That gives both its power which says it doesn't matter where anything else, anywhere in the 230 00:18:24,750 --> 00:18:26,820 world it should be able to communicate. 231 00:18:26,820 --> 00:18:31,440 I creates problems for us also because if there is a person who does a bad thing in 232 00:18:31,440 --> 00:18:36,530 one country that hurts somebody in another country, now we have a international problem. 233 00:18:36,530 --> 00:18:43,330 Today when you see headlines, you realize that this globe girdling design introduces 234 00:18:43,330 --> 00:18:47,639 some of the problems that we have to solve. 235 00:18:47,639 --> 00:18:54,509 >> What you came up with is truly a protocol, but I suspect many other people can come up 236 00:18:54,509 --> 00:18:55,509 with other protocols. 237 00:18:55,509 --> 00:18:56,509 >> VINT CERF: And they did. 238 00:18:56,509 --> 00:18:57,550 >> They did. 239 00:18:57,550 --> 00:19:02,679 What is it about your protocol that you came up with, what, 40 years ago, 45 years ago 240 00:19:02,679 --> 00:19:11,460 now, that has with stood the test of time in the face of international work here, why 241 00:19:11,460 --> 00:19:15,330 didn't folks in Europe come up with their own protocol that stuck? 242 00:19:15,330 --> 00:19:20,639 >> VINT CERF: I didn't ask him to ask the question but it's a wonderful lead in. 243 00:19:20,639 --> 00:19:22,789 Thank you very much. 244 00:19:22,789 --> 00:19:25,769 In fact, they did. 245 00:19:25,769 --> 00:19:35,110 First of all, around 1975 or 6, there were a set of packet switches protocols developed, 246 00:19:35,110 --> 00:19:42,460 X25 and X75. 247 00:19:42,460 --> 00:19:49,520 Larry Roberts had run the ARPANET project at ARPA and left to run a company called telenet. 248 00:19:49,520 --> 00:19:57,220 He is just beginning his commercial telenet service and he says, what protocol should 249 00:19:57,220 --> 00:19:58,220 I use. 250 00:19:58,220 --> 00:19:59,220 I said why don't you use TCP/IP. 251 00:19:59,220 --> 00:20:01,830 He says I can't do that. 252 00:20:01,830 --> 00:20:03,429 Why not? 253 00:20:03,429 --> 00:20:04,600 Those are data grams. 254 00:20:04,600 --> 00:20:06,610 Yeah, that is the whole idea. 255 00:20:06,610 --> 00:20:08,220 I can't sell data grams. 256 00:20:08,220 --> 00:20:12,299 Nobody knows what they are but I can sell virtual circuits, because I can sell them 257 00:20:12,299 --> 00:20:15,659 like they were real circuits and they won't know the difference. 258 00:20:15,659 --> 00:20:22,950 He developed along with several colleagues in Canada and France and UK the X25 protocols 259 00:20:22,950 --> 00:20:23,950 for packet switching. 260 00:20:23,950 --> 00:20:30,649 They were standardized around 1976 and X75 for interconnecting X25 networks but it was 261 00:20:30,649 --> 00:20:36,620 all country code oriented, and virtual circuit oriented. 262 00:20:36,620 --> 00:20:41,539 That is going on its way, it's commercialized, it was a successful system. 263 00:20:41,539 --> 00:20:45,919 Meanwhile I'm still going along saying I need to have this other thing because I need to 264 00:20:45,919 --> 00:20:50,320 have the realtime capability for the defense department. 265 00:20:50,320 --> 00:20:56,710 Around 1978 or so, I'm here in Washington running the program for ARPA, I am, we had 266 00:20:56,710 --> 00:21:02,559 done a three network demonstration in November of '77 to show that the TCP/IP protocols will 267 00:21:02,559 --> 00:21:09,019 work across these packet radio, packet satellite Ethernet and ARPANET systems. 268 00:21:09,019 --> 00:21:13,929 I would like to freeze everything, so we can start implementing it on as many operating 269 00:21:13,929 --> 00:21:15,950 systems as possible. 270 00:21:15,950 --> 00:21:23,929 Right about 1978, our European colleagues said, paraphrasing, why would we ever want 271 00:21:23,929 --> 00:21:30,499 to take a protocol from the U.S. defense department, we should develop our own. 272 00:21:30,499 --> 00:21:32,100 We know what we want to do. 273 00:21:32,100 --> 00:21:40,370 They developed a 7 layer open systems interconnection protocol stack, which began in 1978, and for 274 00:21:40,370 --> 00:21:47,320 the next 15 years, there was competition between OSI and TCP/IP. 275 00:21:47,320 --> 00:21:54,280 The difference between the two unbiased, remember, because my thing worked and theirs, it is 276 00:21:54,280 --> 00:21:59,890 not that it didn't work, it is that they spent a enormous amount of time doing detailed design 277 00:21:59,890 --> 00:22:04,290 and documentation, so they had stacks and stacks of documentation, they introduced a 278 00:22:04,290 --> 00:22:10,499 whole vocabulary for talking about networking which in fact we use today. 279 00:22:10,499 --> 00:22:16,320 If there is anything that we should give them credit for, it is at least the vocabulary 280 00:22:16,320 --> 00:22:18,220 for talking about networking. 281 00:22:18,220 --> 00:22:22,049 However, we spent most of our time trying to make it work. 282 00:22:22,049 --> 00:22:25,990 We went through these four iterations, we implemented it across as many operating systems 283 00:22:25,990 --> 00:22:27,000 as we could. 284 00:22:27,000 --> 00:22:28,200 We put it in the field. 285 00:22:28,200 --> 00:22:33,320 We found our mistakes, we fixed our mistakes, then we documented it. 286 00:22:33,320 --> 00:22:36,279 We sort of reversed that process. 287 00:22:36,279 --> 00:22:45,080 Around 1992, Bob Kahn and I started something called the Internet society, there is a whole 288 00:22:45,080 --> 00:22:51,049 story behind why we did that, but at that point I was President of the Internet society, 289 00:22:51,049 --> 00:22:56,389 and I'm still seeing these arguments about whether OSI or TCP/IP are going to be the 290 00:22:56,389 --> 00:22:58,789 preferred protocols. 291 00:22:58,789 --> 00:23:03,700 Governments were choosing OSI because after all, it was developed by an international 292 00:23:03,700 --> 00:23:06,960 standards organization, why wouldn't you want to do that. 293 00:23:06,960 --> 00:23:13,160 These or people are just a bunch of ragtag graduate students, loosely held together. 294 00:23:13,160 --> 00:23:19,360 Despite, and even the U.S. government decided that OSI would be the future and TCP/IP was 295 00:23:19,360 --> 00:23:23,919 not a blip but it was a demonstration of feasibility. 296 00:23:23,919 --> 00:23:30,269 By the time 1992 rolls around, I'm beginning to think that maybe we should allow TCP/IP 297 00:23:30,269 --> 00:23:34,129 to survive and thrive. 298 00:23:34,129 --> 00:23:41,730 In the exchange about OSI versus TCP/IP I finally wrote as President of the Internet 299 00:23:41,730 --> 00:23:47,850 society to the head of NIST, and I said I would appreciate it very much if you would 300 00:23:47,850 --> 00:23:57,169 convene a blue ribbon committee to evaluate TCP/IP versus OSI. 301 00:23:57,169 --> 00:24:02,399 Because there were a lot of people who were by now investing heavily in TCP/IP implementations, 302 00:24:02,399 --> 00:24:03,929 wanted to use it because it worked. 303 00:24:03,929 --> 00:24:12,029 The academic community was fully engaged, the energy sciences network is up, NASA science 304 00:24:12,029 --> 00:24:13,029 Internet is up. 305 00:24:13,029 --> 00:24:18,350 Our European colleagues by 1992 have made big investments already. 306 00:24:18,350 --> 00:24:25,019 I said I think that we should give them the opportunity to use the TCP/IP protocols and 307 00:24:25,019 --> 00:24:27,690 be authorized to do so. 308 00:24:27,690 --> 00:24:33,490 Up until that point, there was the government OSI profile or gossip that came out of NIST 309 00:24:33,490 --> 00:24:37,309 and that was the only guidance that had been given. 310 00:24:37,309 --> 00:24:43,899 The blue ribbon committee met and by 1993 concluded these two protocols were at least 311 00:24:43,899 --> 00:24:49,710 equivalent and therefore it should be accepted TCP/IP to be used in lieu of OSI. 312 00:24:49,710 --> 00:24:55,169 At that point we turned the corner, most of the products and services that were being 313 00:24:55,169 --> 00:24:58,730 built adopted the TCP/IP. 314 00:24:58,730 --> 00:25:05,240 The other thing that happened in that interval is that Tim Berners Lee came along at Cern, 315 00:25:05,240 --> 00:25:11,789 and against many of the colleagues' opinions at Cern chose to use the Internet as his base 316 00:25:11,789 --> 00:25:16,860 to put the hypertext transport protocols for the Worldwide Web on, so he released that 317 00:25:16,860 --> 00:25:23,940 in December of '91, just as the Internet society is getting started, and nobody notices. 318 00:25:23,940 --> 00:25:27,909 Except a couple guys at the national center for super computer applications, Mark and 319 00:25:27,909 --> 00:25:34,410 Eric, and they say, that is interesting, why don't we do a graphical version of the browser, 320 00:25:34,410 --> 00:25:38,570 so they developed the mosaic browser. 321 00:25:38,570 --> 00:25:42,149 Everybody knows because it turns the Internet into something that looks like a magazine, 322 00:25:42,149 --> 00:25:47,190 with formatted text, imagery, streaming audio and video. 323 00:25:47,190 --> 00:25:49,279 It's a smash hit. 324 00:25:49,279 --> 00:25:51,549 Everybody is downloading the mosaic browser. 325 00:25:51,549 --> 00:25:58,970 A guy named Jim Clark who had been the founder of silicon graphics based on geometry engine 326 00:25:58,970 --> 00:26:04,999 shipped which he had built while at Stanford thanks to the metal oxide implementation that 327 00:26:04,999 --> 00:26:11,350 DARPA set up for people to do VLSI design, he takes one look at this and says there is 328 00:26:11,350 --> 00:26:12,399 something going on there. 329 00:26:12,399 --> 00:26:20,980 He goes to Champaign Urbana and takes others to start Netscape communications in 1994. 330 00:26:20,980 --> 00:26:23,990 They go public in 1995. 331 00:26:23,990 --> 00:26:28,399 The stock goes through the roof and the dot boom is on. 332 00:26:28,399 --> 00:26:31,570 Venture capital people are throwing money at anything that looks like it might have 333 00:26:31,570 --> 00:26:36,620 something to do with the Internet, until around April of 2000. 334 00:26:36,620 --> 00:26:38,619 We had the dot bust. 335 00:26:38,619 --> 00:26:44,100 What happened there, the Internet didn't go away but a lot of companies did, because the 336 00:26:44,100 --> 00:26:51,360 brilliant young men who were CEOs of the companies had not taken economics 101, which says first 337 00:26:51,360 --> 00:26:55,480 of all there is a difference between revenue and capital. 338 00:26:55,480 --> 00:26:57,809 Revenue is supposed to keep going. 339 00:26:57,809 --> 00:26:58,809 Capital is finite. 340 00:26:58,809 --> 00:27:03,509 When they ran out of capital they didn't know what happened and the company went under. 341 00:27:03,509 --> 00:27:08,809 The other thing they should learn is don't spend more money than you make, because your 342 00:27:08,809 --> 00:27:11,280 company goes out of business. 343 00:27:11,280 --> 00:27:18,799 The end story is that after all of the visibility of the Worldwide Web, all the various other 344 00:27:18,799 --> 00:27:26,519 companies, we started to see significant investment in Internet infrastructure, on a global scale. 345 00:27:26,519 --> 00:27:30,519 Today it's 50 percent penetrated around the world, so I still have a lot of work to do 346 00:27:30,519 --> 00:27:34,559 to get the other 3.9 billion people that aren't connected. 347 00:27:34,559 --> 00:27:38,820 But also lots of other companies began to build on top of the web platform. 348 00:27:38,820 --> 00:27:44,340 It took off in the mid 1990s, and it's still going. 349 00:27:44,340 --> 00:27:47,419 >> This is a good transition, by the way, in about ten minutes we will take questions 350 00:27:47,419 --> 00:27:48,419 from the audience. 351 00:27:48,419 --> 00:27:51,259 We have three mics, and I'll let you know. 352 00:27:51,259 --> 00:27:53,059 Get your questions ready. 353 00:27:53,059 --> 00:28:00,490 You were there at the very beginning, in the '70s of the Internet, first of all I have 354 00:28:00,490 --> 00:28:03,070 a series of questions on that. 355 00:28:03,070 --> 00:28:08,600 Did you envision then or approximately at that time in the '70s what this thing would 356 00:28:08,600 --> 00:28:15,589 become, and how it would literally completely take over the world and change humanity? 357 00:28:15,589 --> 00:28:21,910 >> VINT CERF: You would have to be a megalomaniac to have imagined that in the 1970s. 358 00:28:21,910 --> 00:28:28,830 I was focused and patient and persistent, but I didn't I hoped this would do something 359 00:28:28,830 --> 00:28:30,419 useful. 360 00:28:30,419 --> 00:28:35,239 Honestly though, we had a inkling about some of the things that have happened. 361 00:28:35,239 --> 00:28:42,980 In the ARPANET period, electronic mail got invented, Ray Tomlinson came up with idea 362 00:28:42,980 --> 00:28:47,399 of electronic mail in mid 1971. 363 00:28:47,399 --> 00:28:53,259 He figured out that if you want to send something to someone at a different machine, you had 364 00:28:53,259 --> 00:28:58,080 to say which machine it was going to, and you had to say which user it was going to, 365 00:28:58,080 --> 00:29:04,649 and he was trying to figure out how do I separate those two, user, something, and he found that 366 00:29:04,649 --> 00:29:08,770 the only character that wasn't already used by the other operating systems was the at 367 00:29:08,770 --> 00:29:15,779 sign, so user @ host was his way of saying send this to that computer and get it to that 368 00:29:15,779 --> 00:29:16,779 user. 369 00:29:16,779 --> 00:29:23,029 We saw E mail pop up in mid 1971, everybody goes crazy, because it suddenly means that 370 00:29:23,029 --> 00:29:27,919 we can coordinate with each other without both having to be awake at the same time because 371 00:29:27,919 --> 00:29:30,690 otherwise it's phone conference calls. 372 00:29:30,690 --> 00:29:37,340 Indeed, we thought that this would allow us to manage more far flung projects because 373 00:29:37,340 --> 00:29:39,669 of the time zone accommodation. 374 00:29:39,669 --> 00:29:43,780 This is computer mediated communication. 375 00:29:43,780 --> 00:29:48,580 The other thing that happened is that mailing lists were very quickly created. 376 00:29:48,580 --> 00:29:53,450 The first mailing list that I can remember was called sci fi lovers, because we were 377 00:29:53,450 --> 00:29:57,159 all geeks and we all argued over who is the best science fiction writer, which were the 378 00:29:57,159 --> 00:29:58,159 best books. 379 00:29:58,159 --> 00:29:59,590 That was a distribution list. 380 00:29:59,590 --> 00:30:03,669 We saw immediately, that had social character to it. 381 00:30:03,669 --> 00:30:09,919 We had a mailing list called yum yum, which is a restaurant review list at Stanford for 382 00:30:09,919 --> 00:30:13,590 the restaurants in the Palo Alto area which expanded over time. 383 00:30:13,590 --> 00:30:18,889 We saw the social net working element of this even before Internet. 384 00:30:18,889 --> 00:30:27,250 We saw what we think of today as Internet of Things, was nascent in the system. 385 00:30:27,250 --> 00:30:33,210 We were building sensory systems, we were using radars, we were using Voice over IP, 386 00:30:33,210 --> 00:30:39,090 we were using videoconferencing, in the early 1980s. 387 00:30:39,090 --> 00:30:47,370 So a lot of what we see today has simply become more affordable, and physically smaller, therefore, 388 00:30:47,370 --> 00:30:48,929 more portable. 389 00:30:48,929 --> 00:30:50,929 But we could see a lot of that coming. 390 00:30:50,929 --> 00:30:54,129 I wouldn't have predicted everything that happened. 391 00:30:54,129 --> 00:30:59,649 But a awful lot of this was visibly nascent and of course as time went on, as the decades 392 00:30:59,649 --> 00:31:04,770 rolled along, there would be milestones when you would realize, holey moley, this is really 393 00:31:04,770 --> 00:31:10,929 serious, Voice over IP we are experimenting with it in the 1970s but it starts showing 394 00:31:10,929 --> 00:31:14,440 up in commercial terms in 1996. 395 00:31:14,440 --> 00:31:21,010 It doesn't actually catch on until later, but the telephone companies began discarding 396 00:31:21,010 --> 00:31:27,159 their circuit switching gear and using packet switch voice because it was cheaper. 397 00:31:27,159 --> 00:31:32,029 A lot of this was hiding in there, even if, I don't claim that we imagined everything 398 00:31:32,029 --> 00:31:37,360 that happened on the Internet but a lot of it you could anticipate. 399 00:31:37,360 --> 00:31:41,980 >> One more question, so folks can start lining up if you would like to ask questions from 400 00:31:41,980 --> 00:31:44,440 the audience. 401 00:31:44,440 --> 00:31:45,509 What about the future? 402 00:31:45,509 --> 00:31:49,940 You have a wonderful title, I didn't mention in the opening, but your current title, you 403 00:31:49,940 --> 00:31:55,480 work at Google among other things you do, and you are a Internet evangelist, is that 404 00:31:55,480 --> 00:31:56,480 on your business card? 405 00:31:56,480 --> 00:31:57,480 >> VINT CERF: It is. 406 00:31:57,480 --> 00:31:58,480 Chief Internet evangelist. 407 00:31:58,480 --> 00:31:59,480 >> That is fantastic. 408 00:31:59,480 --> 00:32:06,029 Achieve evangelist, that is great title. 409 00:32:06,029 --> 00:32:12,239 I'm sure that you spent a lot of time thinking about the future. 410 00:32:12,239 --> 00:32:15,340 In our, in your domain, what do you foresee? 411 00:32:15,340 --> 00:32:18,760 Where would we be let's say another 40 years from now? 412 00:32:18,760 --> 00:32:22,570 >> VINT CERF: First of all, I didn't ask for that title, by the way. 413 00:32:22,570 --> 00:32:26,590 I asked for arch duke. (laughter). 414 00:32:26,590 --> 00:32:31,049 But Larry and Eric and Sergei reminded me that the previous arch duke was Ferdinand 415 00:32:31,049 --> 00:32:36,169 and he was assassinated in 1914 and it started World War I. 416 00:32:36,169 --> 00:32:38,149 So that is probably a bad title to have. 417 00:32:38,149 --> 00:32:40,230 Why don't you be our chief Internet evangelist. 418 00:32:40,230 --> 00:32:43,759 Okay, I can do that. 419 00:32:43,759 --> 00:32:50,299 I'm fortunate that Google allows me a great deal of latitude, one of my missions is to 420 00:32:50,299 --> 00:32:55,169 get more Internet built, Google has contributed to that in a variety of different ways, including 421 00:32:55,169 --> 00:33:01,210 investment in infrastructure in Africa, program, with fiber networks and in India with wi fies 422 00:33:01,210 --> 00:33:06,659 along the railroad stations, because fiber is running along the railroad tracks. 423 00:33:06,659 --> 00:33:14,299 I also sit on the NASA advisory committee, and have been a visiting scientist at the 424 00:33:14,299 --> 00:33:17,359 jet propulsion laboratory since 1998. 425 00:33:17,359 --> 00:33:22,249 We have been developing new protocols for interplanetary operation of the Internet since 426 00:33:22,249 --> 00:33:24,080 1998. 427 00:33:24,080 --> 00:33:29,799 We have those protocols, TCP didn't work well between earth and Mars, because when there 428 00:33:29,799 --> 00:33:32,690 is a 20 minute one way >> Don't you need somebody else there? 429 00:33:32,690 --> 00:33:35,059 >> VINT CERF: We have equipment there. 430 00:33:35,059 --> 00:33:38,970 We are not building a network and hoping somebody will come. 431 00:33:38,970 --> 00:33:46,210 We are building the network to service our manned and robotic space exploration needs 432 00:33:46,210 --> 00:33:50,379 to provide them with a richer communication environment and a point to point radio link 433 00:33:50,379 --> 00:33:56,610 which is how we have been exploring the solar system since the 1960s. 434 00:33:56,610 --> 00:34:01,580 Those protocols needed some other features, TCP/IP didn't have. 435 00:34:01,580 --> 00:34:02,580 We developed those. 436 00:34:02,580 --> 00:34:06,570 We have gone through the usual mistakes and reimplementations and design. 437 00:34:06,570 --> 00:34:12,200 It is operating now between earth and Mars and the international space station. 438 00:34:12,200 --> 00:34:16,810 The gateway program which is a new orbiter around the moon which is to be launched in 439 00:34:16,810 --> 00:34:21,030 mid 2020s will have those protocols on board. 440 00:34:21,030 --> 00:34:25,190 We have offered them, and standardized them through the consultative committee on space 441 00:34:25,190 --> 00:34:26,419 systems. 442 00:34:26,419 --> 00:34:27,500 Protocols are freely available. 443 00:34:27,500 --> 00:34:31,750 There is no restrictions, no patents or anything else. 444 00:34:31,750 --> 00:34:40,470 We didn't want there to be any barrier or excuse to not adopt them, concern over payment 445 00:34:40,470 --> 00:34:41,660 of some kind. 446 00:34:41,660 --> 00:34:44,690 They were released freely and openly. 447 00:34:44,690 --> 00:34:46,929 That is one thing for sure that is happening. 448 00:34:46,929 --> 00:34:53,390 The second thing, you are already carrying a mobile with a camera, is it a prototype 449 00:34:53,390 --> 00:34:58,850 of Internet enabled thing of which there will be billions, many of them are already in the 450 00:34:58,850 --> 00:35:07,120 market today, web cams, security systems, heating ventilation control systems, all kinds 451 00:35:07,120 --> 00:35:15,360 of appliances, and everyone should be very concerned about safety and security and reliability 452 00:35:15,360 --> 00:35:19,310 and privacy, and autonomy of these devices. 453 00:35:19,310 --> 00:35:23,250 These are nontrivial problems to solve. 454 00:35:23,250 --> 00:35:30,350 Part of what I can foresee over the next ten, 20, 30 years is a lot of struggle and challenge 455 00:35:30,350 --> 00:35:34,630 to make sure the stuff actually behaves the way you expect it to. 456 00:35:34,630 --> 00:35:38,410 Some people are making heavy investments in machine learning and artificial intelligence. 457 00:35:38,410 --> 00:35:40,600 We certainly are at Google. 458 00:35:40,600 --> 00:35:44,950 But when you start applying those things and then relying on either the machine learning 459 00:35:44,950 --> 00:35:51,780 mechanisms or just the ordinary software that you give the device control, that animates 460 00:35:51,780 --> 00:35:57,730 the device and you hand it authority to control things, there are some risks associated with 461 00:35:57,730 --> 00:36:00,160 that, especially if there are bugs. 462 00:36:00,160 --> 00:36:06,650 I foresee a period of time when we have challenges to fix the bugs, either they get exploited 463 00:36:06,650 --> 00:36:13,490 in which case we have malware problems, or privacy is invaded, or the headline I worry 464 00:36:13,490 --> 00:36:18,450 about is the one that says 100,000 web cams take over Bank of America. 465 00:36:18,450 --> 00:36:24,520 Unfortunately, that is not funny anymore, because 500,000 web cams did take over dyne 466 00:36:24,520 --> 00:36:30,840 corporation or knocked it over in a denial of service attack a couple of years ago. 467 00:36:30,840 --> 00:36:36,020 There is a lot that is going to happen with programmable devices, that can communicate. 468 00:36:36,020 --> 00:36:38,320 The Internet is part of that. 469 00:36:38,320 --> 00:36:44,831 Our big job at Google and elsewhere is to try to make sure that we do this in a way 470 00:36:44,831 --> 00:36:49,850 that protects everybody's interests, and that is a nontrivial challenge. 471 00:36:49,850 --> 00:36:56,930 On the other side though and on the positive side, I'm thrilled to see these tools coming, 472 00:36:56,930 --> 00:37:04,220 because as Doug, the guy that invented the mouse and many other things, said, his thought 473 00:37:04,220 --> 00:37:09,610 was the computers should be partners with people in order to augment our ability to 474 00:37:09,610 --> 00:37:15,630 think, just like electric motors augmented our muscle power, computers can augment our 475 00:37:15,630 --> 00:37:19,730 brain power, if we learn how to cooperate with them. 476 00:37:19,730 --> 00:37:24,240 I'm all excited about what is going to happen over the next 20, 30 or 40 I just wish I could 477 00:37:24,240 --> 00:37:25,240 see it. 478 00:37:25,240 --> 00:37:30,530 I sometimes wish I was 8 years old so I could see what happens in 2090. 479 00:37:30,530 --> 00:37:35,010 >> At the speed of innovation nowadays, 40 years might actually happen, in five or ten 480 00:37:35,010 --> 00:37:37,390 or >> VINT CERF: Faster than we think. 481 00:37:37,390 --> 00:37:38,670 >> Questions, folks. 482 00:37:38,670 --> 00:37:44,040 >> VINT CERF: Qualify questions I'm going to put my headset on so I can hear you better. 483 00:37:44,040 --> 00:37:45,230 >> Great. 484 00:37:45,230 --> 00:37:49,970 There are three mics. 485 00:37:49,970 --> 00:37:52,350 Very good. 486 00:37:52,350 --> 00:37:54,010 I don't know that the mics are on. 487 00:37:54,010 --> 00:37:55,560 >> VINT CERF: That mic isn't on. 488 00:37:55,560 --> 00:37:57,370 Try that again. 489 00:37:57,370 --> 00:38:06,170 How many engineers does it take to turn on a microphone? 490 00:38:06,170 --> 00:38:07,170 >> It's not on. 491 00:38:07,170 --> 00:38:10,690 Is that one on, on that side? 492 00:38:10,690 --> 00:38:13,569 >> Hello. 493 00:38:13,569 --> 00:38:14,569 >> It's on. 494 00:38:14,569 --> 00:38:15,569 >> Go ahead. 495 00:38:15,569 --> 00:38:18,240 >> I wanted to say hi, Vint. 496 00:38:18,240 --> 00:38:19,240 >> Hi. 497 00:38:19,240 --> 00:38:20,240 >> Hi. 498 00:38:20,240 --> 00:38:25,740 Actually, I have a different last name than I used to have before, but these days I am 499 00:38:25,740 --> 00:38:33,280 with Ernst & Young EY, a while back and we do advisory services at USPTO but a while 500 00:38:33,280 --> 00:38:39,880 back I was part of [inaudible] >> We share a common disaster. 501 00:38:39,880 --> 00:38:48,620 >> I found in TCP/IP distribution list, one distribution you were on it, I was on it. 502 00:38:48,620 --> 00:38:54,780 We were pretty much the first company or one of the first companies who made TCP/IP commercial 503 00:38:54,780 --> 00:39:00,250 and available to everybody in the world for small fee. 504 00:39:00,250 --> 00:39:07,710 In '70s you had no idea of the impact, neither had we in '80s. 505 00:39:07,710 --> 00:39:13,451 I talk to some of my former colleagues and we never imagined what would happen, what 506 00:39:13,451 --> 00:39:15,220 would be the impact. 507 00:39:15,220 --> 00:39:18,360 So what is the next thing? 508 00:39:18,360 --> 00:39:23,370 Is it human genome who may have this kind of impact or something else, any opinion on 509 00:39:23,370 --> 00:39:24,370 your part? 510 00:39:24,370 --> 00:39:31,160 >> VINT CERF: If we go outside of communications for a moment, probably the most important 511 00:39:31,160 --> 00:39:37,520 technology which has emerged is crisper class 9 which is allowing people to do editing of 512 00:39:37,520 --> 00:39:38,520 genetic sequences. 513 00:39:38,520 --> 00:39:48,210 Even that we have to be super careful about, it turns out it doesn't work perfectly. 514 00:39:48,210 --> 00:39:54,010 The modifications can be made in the genetic sequence where you didn't intend. 515 00:39:54,010 --> 00:39:59,270 Any time you see somebody making claims about solving people's medical problems by editing 516 00:39:59,270 --> 00:40:02,540 their genes, be very careful about that. 517 00:40:02,540 --> 00:40:08,940 However, I will say that we are at a period of time now where our depth of understanding 518 00:40:08,940 --> 00:40:16,710 of biology is dramatically improving rapidly, partly because of our ability to measure things 519 00:40:16,710 --> 00:40:19,160 in ways we could not measure before. 520 00:40:19,160 --> 00:40:25,100 At this point we have the ability to take a single cell, select a single cell and sequence 521 00:40:25,100 --> 00:40:28,200 the DNA in that cell. 522 00:40:28,200 --> 00:40:33,540 Imagine if you have cancer, for example, you can select a cell which you recognize to be 523 00:40:33,540 --> 00:40:41,030 cancerous because of its appearance, and you can select that DNA and sequence it and use 524 00:40:41,030 --> 00:40:46,140 the information that you get to figure out how to target cells with those DNA sequences. 525 00:40:46,140 --> 00:40:48,370 That is one thing that is happening. 526 00:40:48,370 --> 00:40:52,850 The second thing that is happening, and it's fun from the computer science point of view, 527 00:40:52,850 --> 00:41:00,270 is that almost all sciences, that you can think of today, have a computational analog, 528 00:41:00,270 --> 00:41:06,730 so we talk about computational linguistics, about computational astronomy, computational 529 00:41:06,730 --> 00:41:12,380 biology, so suddenly we are starting to use computers to help us do things we couldn't 530 00:41:12,380 --> 00:41:18,210 do before, and it gets even more exciting when you start looking at new computing technology 531 00:41:18,210 --> 00:41:21,810 for example, quantum computing. 532 00:41:21,810 --> 00:41:26,600 These are all current buzzwords, so nobody will be too surprised at that. 533 00:41:26,600 --> 00:41:30,690 But I'm proud of the fact that Google invested heavily in quantum computing. 534 00:41:30,690 --> 00:41:37,390 We have a 72Q bit system running in Santa Barbara which is well beyond the, what is 535 00:41:37,390 --> 00:41:43,580 normally called the quantum supremacy limits which is around 50 qubits so that is becoming 536 00:41:43,580 --> 00:41:50,140 real to say nothing of machine learning and neural networks and things of that sort. 537 00:41:50,140 --> 00:41:57,440 >> What about combining crisper and DNA technology with TCP/IP, to make a invention right here. 538 00:41:57,440 --> 00:41:58,440 Okay. 539 00:41:58,440 --> 00:41:59,770 Here is the thought. 540 00:41:59,770 --> 00:42:07,430 You scan the human for teleportation, okay, you scan the human and transmit the DNA data 541 00:42:07,430 --> 00:42:10,020 instantaneously and it gets recreated on the other side. 542 00:42:10,020 --> 00:42:11,100 What do you think? 543 00:42:11,100 --> 00:42:12,190 >> Move them to Mars. 544 00:42:12,190 --> 00:42:13,760 >> VINT CERF: Several things come to mind. 545 00:42:13,760 --> 00:42:21,070 First if you have a bad problem and arrive inverted, that is why some of the "Star Trek" 546 00:42:21,070 --> 00:42:25,690 people didn't want to get into the teletransporter. 547 00:42:25,690 --> 00:42:31,180 What you just said is not as silly as it sounds, not the teleportation part, but I've been 548 00:42:31,180 --> 00:42:37,750 working with a biologist at EZ San Diego who sent me a E mail a year and a half ago saying 549 00:42:37,750 --> 00:42:44,770 I found a Internet inside the cell, of course my reaction was, okay, can I come out and 550 00:42:44,770 --> 00:42:46,320 talk about that. 551 00:42:46,320 --> 00:42:53,570 What she found is that there are layers of biological, chemical communications going 552 00:42:53,570 --> 00:42:59,380 on between the organelles inside a cell and between cells. 553 00:42:59,380 --> 00:43:05,530 There really is a communication structure associated with biological systems. 554 00:43:05,530 --> 00:43:12,570 Once we understand that, we have an opportunity to make use of that for therapeutic purposes. 555 00:43:12,570 --> 00:43:15,910 I'm all excited about this stuff that we can do now. 556 00:43:15,910 --> 00:43:16,980 >> Thank you. 557 00:43:16,980 --> 00:43:18,600 >> Thank you, ma'am. 558 00:43:18,600 --> 00:43:19,600 >> Hi. 559 00:43:19,600 --> 00:43:22,870 My question has to do with one of the topics you mentioned about AI. 560 00:43:22,870 --> 00:43:30,110 I wonder what you thought the Internet was going to, its role, other than being a transport 561 00:43:30,110 --> 00:43:37,910 between AI endpoints, or do you think that you could see it being part of the integral 562 00:43:37,910 --> 00:43:44,600 part of AI, like having its own >> VINT CERF: This is the scenario where one 563 00:43:44,600 --> 00:43:49,930 day, when no one is looking, the Internet suddenly becomes conscious. 564 00:43:49,930 --> 00:43:50,930 >> Yes. 565 00:43:50,930 --> 00:43:57,930 >> VINT CERF: The scene in the movie is, it's alive! 566 00:43:57,930 --> 00:44:01,400 Sky net. 567 00:44:01,400 --> 00:44:08,870 So I don't think that we are likely to see that exactly happen. 568 00:44:08,870 --> 00:44:10,640 Two things are very important here. 569 00:44:10,640 --> 00:44:21,451 One of my colleagues at Google has been speculating about scaling of computing, he wrote a, it's 570 00:44:21,451 --> 00:44:25,370 Ray Kurzweil, called the singularity is near. 571 00:44:25,370 --> 00:44:29,820 He was arguing once you get up to a certain number of processors you are getting close 572 00:44:29,820 --> 00:44:32,180 to the same scale as a human brain. 573 00:44:32,180 --> 00:44:37,470 I think that may not be the right metric, because it's the connectivity of the human 574 00:44:37,470 --> 00:44:40,060 brain that gives it its unique character. 575 00:44:40,060 --> 00:44:45,040 I'm not sure that we will ever see that degree of connectivity, even in the Internet today. 576 00:44:45,040 --> 00:44:51,000 It really tends to be very hierarchical whereas the brain is anything but that. 577 00:44:51,000 --> 00:44:52,280 I don't think that we will see that. 578 00:44:52,280 --> 00:44:59,300 I think that the Internet for the present is simply a wonderful road system, on top 579 00:44:59,300 --> 00:45:04,810 of which we are able to build a wide range of applications. 580 00:45:04,810 --> 00:45:09,800 When people ask for metaphors, I think of the Internet as kind of a road system that 581 00:45:09,800 --> 00:45:15,680 Bob and I helped to design, we said don't build the cars longer than this or wider than 582 00:45:15,680 --> 00:45:21,260 this or taller than that or heavier than this and it should work on this road system. 583 00:45:21,260 --> 00:45:26,630 By the way, we won't tell you what kind of vehicle to design, one wheel or two or four 584 00:45:26,630 --> 00:45:32,400 wheels or 18 wheels, we won't say anything about what buildings you put at the side of 585 00:45:32,400 --> 00:45:35,670 the road or what you put in them. 586 00:45:35,670 --> 00:45:39,500 There is enormous freedom in this architecture. 587 00:45:39,500 --> 00:45:43,340 I'm sure there will be some surprises, but I honestly don't think that the artificial 588 00:45:43,340 --> 00:45:48,030 intelligence of the kind that you are implying will arise out of the network itself. 589 00:45:48,030 --> 00:45:51,430 >> Follow up question real quick. 590 00:45:51,430 --> 00:45:58,720 What do you think the residence to adopt IPv6 or why we are slow moving in that direction. 591 00:45:58,720 --> 00:46:03,070 >> VINT CERF: Same reason that the guy with the hole in his roof doesn't fix it, when 592 00:46:03,070 --> 00:46:07,240 it's raining, he doesn't want to do it because it's dangerous to go up there, when the sun 593 00:46:07,240 --> 00:46:10,010 is shining it's not a problem so why bother. 594 00:46:10,010 --> 00:46:16,950 A lot of people are saying I still have IPv4 address space so I don't need IPv6. 595 00:46:16,950 --> 00:46:21,140 What they misunderstand, most of the software and routers, software in edge devices, desktop, 596 00:46:21,140 --> 00:46:23,560 laptop, mobile can all do IPv6. 597 00:46:23,560 --> 00:46:29,110 ISPs have not turned it on and they say nobody is asking for it. 598 00:46:29,110 --> 00:46:32,960 My reaction is nobody should know about it. 599 00:46:32,960 --> 00:46:35,910 Ordinary users shouldn't have to care what IPv6 is. 600 00:46:35,910 --> 00:46:37,830 We should just do it and get it over with. 601 00:46:37,830 --> 00:46:41,980 It is about 30 percent penetrated on the average. 602 00:46:41,980 --> 00:46:47,140 You can all help by the way, if you go home and whoever your ISP carrier is, ask them 603 00:46:47,140 --> 00:46:50,150 when can I get my IPv6 allocation. 604 00:46:50,150 --> 00:46:54,260 That will help, because you will fulfill their complaint that, satisfy the complaint that 605 00:46:54,260 --> 00:46:57,180 nobody was asking for it. 606 00:46:57,180 --> 00:46:59,800 >> Understand it too. 607 00:46:59,800 --> 00:47:02,000 >> VINT CERF: They have to know they have to call and ask for it. 608 00:47:02,000 --> 00:47:04,230 That is all you have to know. 609 00:47:04,230 --> 00:47:07,210 The other problem we get is I have enough IPv4 address space. 610 00:47:07,210 --> 00:47:08,630 There is not a problem. 611 00:47:08,630 --> 00:47:13,200 The if somebody else on the other side of the world ran out of IPv4 address and has 612 00:47:13,200 --> 00:47:19,800 to use IPv6 and you want to make sure you can use IPv6 across the map so everybody needs 613 00:47:19,800 --> 00:47:22,300 to implement it. 614 00:47:22,300 --> 00:47:28,770 The smart phones by the way, and the 5G world is all IPv6. 615 00:47:28,770 --> 00:47:35,670 I'm fairly confident that we will get to the point where the IPv4 space will be in the 616 00:47:35,670 --> 00:47:36,670 minority. 617 00:47:36,670 --> 00:47:38,960 But it is just taking a long time to get there. 618 00:47:38,960 --> 00:47:39,960 >> Thank you. 619 00:47:39,960 --> 00:47:42,770 >> Vint, I have a question from somebody on line. 620 00:47:42,770 --> 00:47:46,990 We have more than 1,000 folks watching on line. 621 00:47:46,990 --> 00:47:52,800 What are some of the obstacles in connecting the remaining 50 percent of the world's population, 622 00:47:52,800 --> 00:47:54,220 about four billion people. 623 00:47:54,220 --> 00:47:56,440 >> VINT CERF: There are a number of them. 624 00:47:56,440 --> 00:47:58,310 The first one is cost. 625 00:47:58,310 --> 00:48:00,170 We have to get cost out. 626 00:48:00,170 --> 00:48:02,100 Think about the way in which the Internet grew. 627 00:48:02,100 --> 00:48:06,570 It went to places that could afford it first. 628 00:48:06,570 --> 00:48:12,050 As you push farther out to places where the Internet isn't, that is where you discover 629 00:48:12,050 --> 00:48:17,600 you need to drive cost of communications down, drive the cost of the equipment down. 630 00:48:17,600 --> 00:48:18,860 That is one thing. 631 00:48:18,860 --> 00:48:21,420 Second, you have to make it useful. 632 00:48:21,420 --> 00:48:23,440 Just having it doesn't do anything. 633 00:48:23,440 --> 00:48:25,820 You need local information that is useful. 634 00:48:25,820 --> 00:48:30,790 It doesn't help if you are in Caracas trying to find a plumber and you get back from the 635 00:48:30,790 --> 00:48:34,790 Google search a plumber in New York, it doesn't help. 636 00:48:34,790 --> 00:48:38,540 You need local content in local languages to be useful. 637 00:48:38,540 --> 00:48:46,420 You need applications that maybe the government can use to help service the population. 638 00:48:46,420 --> 00:48:49,750 Think about India for a moment, where Internet was not heavily penetrated. 639 00:48:49,750 --> 00:48:52,540 It is getting better now. 640 00:48:52,540 --> 00:48:59,230 One thing the Indian government did was provide identity for some 800 million people of its 641 00:48:59,230 --> 00:49:03,760 population, so they could be visible to the government and be able to get access to government 642 00:49:03,760 --> 00:49:05,340 services. 643 00:49:05,340 --> 00:49:09,870 There are a bunch of variables that have to be satisfied to make the Internet propagate 644 00:49:09,870 --> 00:49:10,950 further. 645 00:49:10,950 --> 00:49:17,820 At Google, we have tried to make demonstration investments, we built a optical fiber network 646 00:49:17,820 --> 00:49:23,700 for the City of Kampala in Uganda, and we made it available at wholesale rates to people 647 00:49:23,700 --> 00:49:28,260 who could offer retail services and compete with each other for customers. 648 00:49:28,260 --> 00:49:32,920 That particular tactic worked out so well, we created a joint venture called C squared 649 00:49:32,920 --> 00:49:37,480 to replicate that in other parts of Africa, by way of example. 650 00:49:37,480 --> 00:49:39,670 But there is a lot of work to be done there. 651 00:49:39,670 --> 00:49:45,640 Policies have to be adopted in the governments where these networks need to be built, that 652 00:49:45,640 --> 00:49:49,150 will encourage private sector investment, for example. 653 00:49:49,150 --> 00:49:51,520 And encourage competition. 654 00:49:51,520 --> 00:49:57,270 In some parts of the world, Telecom has been a piggy bank for the government and for a 655 00:49:57,270 --> 00:50:03,670 long time termination charges were generated, a lot of revenue for the government, and so 656 00:50:03,670 --> 00:50:08,980 as the Internet comes along and Voice over IP is essentially free, suddenly that model 657 00:50:08,980 --> 00:50:10,260 doesn't work anymore. 658 00:50:10,260 --> 00:50:15,780 It's a classic example of something that people here at the Patent and Trademark Office are 659 00:50:15,780 --> 00:50:21,250 familiar with, and that is that new technologies disrupt old business models. 660 00:50:21,250 --> 00:50:26,580 You have to be agile to figure out what to do, and in particular, if you are in a business 661 00:50:26,580 --> 00:50:31,780 that is disruptable, you should be asking yourself, what did somebody else do to my 662 00:50:31,780 --> 00:50:36,130 business to make the business model fail, and then you learn the lesson, if somebody 663 00:50:36,130 --> 00:50:40,310 else is going to eat your lunch or if someone is going to eat your lunch it might as well 664 00:50:40,310 --> 00:50:43,210 be you. 665 00:50:43,210 --> 00:50:44,210 >> Thank you. 666 00:50:44,210 --> 00:50:45,210 >> Ma'am. 667 00:50:45,210 --> 00:50:47,320 >> Over here. 668 00:50:47,320 --> 00:50:55,500 >> I'm also a geneticist, and one of the questions that comes to my mind about ability to sequence 669 00:50:55,500 --> 00:51:04,030 single cells, crisper and all of that how do we go and design something where we are 670 00:51:04,030 --> 00:51:13,730 able to analyze and make use of the tremendous amount of biological data in a cost effective 671 00:51:13,730 --> 00:51:17,740 and manner where people will be able to use it. 672 00:51:17,740 --> 00:51:21,770 >> VINT CERF: That is a interesting question. 673 00:51:21,770 --> 00:51:27,520 To extend the question a little bit, there is a substantial amount of data being generated 674 00:51:27,520 --> 00:51:33,210 now by all of the various scientific enterprises, and one of the big questions is how do we 675 00:51:33,210 --> 00:51:39,050 capture the data, how do we capture the metadata so we know what it means, so that the information 676 00:51:39,050 --> 00:51:42,570 coming from multiple sources can be combined together. 677 00:51:42,570 --> 00:51:49,080 In your world, you have done better than everyone else except maybe the astronomers who have 678 00:51:49,080 --> 00:51:53,710 done a good job of standardizing the way in which they capture data and represent it and 679 00:51:53,710 --> 00:51:55,440 store it and retrieve it. 680 00:51:55,440 --> 00:52:02,960 In the genetics world the three major genetic sequence banks have standardized the representations. 681 00:52:02,960 --> 00:52:07,890 The publishing companies have insisted that people share their genetic discoveries, sequences 682 00:52:07,890 --> 00:52:13,210 that they discovered before the papers are published, which is a brilliant way of creating 683 00:52:13,210 --> 00:52:14,680 an openly available database. 684 00:52:14,680 --> 00:52:21,060 But you are putting your finger on a problem that for many of the sciences which is there 685 00:52:21,060 --> 00:52:24,390 is no common standards for representing the data. 686 00:52:24,390 --> 00:52:32,270 Imagine for a moment that you have a series of sensors systems, NASA has this problem, 687 00:52:32,270 --> 00:52:37,220 there are specialized instruments that have been sent out to gather data around the solar 688 00:52:37,220 --> 00:52:44,500 system, that data is being sent back, being captured locally, recorded in data centers. 689 00:52:44,500 --> 00:52:49,350 But a hundred years from now, it's not clear whether we will know what the data means. 690 00:52:49,350 --> 00:52:55,070 I have a bunch of numbers here, but was this pressure, temperature, radiation levels, something 691 00:52:55,070 --> 00:52:56,250 else. 692 00:52:56,250 --> 00:52:57,470 Metadata has to be captured. 693 00:52:57,470 --> 00:53:01,780 When was it captured, on what spacecraft, where was it captured, all those questions 694 00:53:01,780 --> 00:53:03,580 have to be answered. 695 00:53:03,580 --> 00:53:09,590 I'm a huge fan of standardization of data representation in order to make sure that 696 00:53:09,590 --> 00:53:12,720 we can use that data over long periods of time. 697 00:53:12,720 --> 00:53:15,200 Digital preservation is part of the challenge. 698 00:53:15,200 --> 00:53:16,770 >> Thank you. 699 00:53:16,770 --> 00:53:19,760 I'm afraid we have >> VINT CERF: We are running out of time, 700 00:53:19,760 --> 00:53:20,760 wow. 701 00:53:20,760 --> 00:53:27,710 >> Feel free to come afterwards, but let me end with this, Vint. 702 00:53:27,710 --> 00:53:33,800 You have been in Silicon Valley or part of the Silicon Valley culture, there is actually 703 00:53:33,800 --> 00:53:41,670 a wonderful scene in Walter Isaacson's innovators book, you make a appearance there and he describes 704 00:53:41,670 --> 00:53:45,190 you, I don't know if it's accurate but describes you as being dressed in a three piece suit, 705 00:53:45,190 --> 00:53:51,200 tie, briefcase in hand in the '70s in Silicon Valley, perhaps a little bit out of place, 706 00:53:51,200 --> 00:53:54,660 a lot of things have changed in Silicon Valley. 707 00:53:54,660 --> 00:53:56,610 The way you dress hasn't, obviously. 708 00:53:56,610 --> 00:53:57,610 (laughter). 709 00:53:57,610 --> 00:54:03,420 >> VINT CERF: Because it's the only suits I have. 710 00:54:03,420 --> 00:54:09,060 >> But on a final thought here what do you think overall in terms of the Silicon Valley 711 00:54:09,060 --> 00:54:17,960 overall tech culture in the United States, has it changed for the better or how has it 712 00:54:17,960 --> 00:54:23,369 changed and what is your view on the state of the tech culture in the States now. 713 00:54:23,369 --> 00:54:27,480 >> VINT CERF: This is a very interesting question, because I actually moved out of the Silicon 714 00:54:27,480 --> 00:54:30,320 Valley in 1976. 715 00:54:30,320 --> 00:54:38,420 I was in it for, I was in California for 30 years, of those, eight years were at Stanford. 716 00:54:38,420 --> 00:54:43,630 I was around in the early periods of Silicon Valley's evolution. 717 00:54:43,630 --> 00:54:53,120 I think that there are a lot of peculiar phenomena that have led to the Silicon Valley phenomena 718 00:54:53,120 --> 00:55:00,250 that have been harmful in some respects, because it's chasing dollars. 719 00:55:00,250 --> 00:55:07,620 I don't know whether any of you have heard the expression, that the worst possible economic 720 00:55:07,620 --> 00:55:13,150 mistake was made when somebody said, the sole purpose of the company was to increase shareholder 721 00:55:13,150 --> 00:55:15,040 value. 722 00:55:15,040 --> 00:55:21,390 And then you give to the CEOs maximum amounts of stocks so they will be incented to improve 723 00:55:21,390 --> 00:55:26,940 shareholder value which means it drives the stock up and all kinds of behaviors stem from 724 00:55:26,940 --> 00:55:27,970 that. 725 00:55:27,970 --> 00:55:33,220 I think we have seen more and more of the consequences of that showing up, in Silicon 726 00:55:33,220 --> 00:55:34,220 Valley. 727 00:55:34,220 --> 00:55:39,860 On the positive side though Silicon Valley is a amazing phenomenon, it is a economic 728 00:55:39,860 --> 00:55:45,450 engine of extraordinary proportions and it's very important to understand what the ingredients 729 00:55:45,450 --> 00:55:49,210 were and are, that keep that engine running. 730 00:55:49,210 --> 00:55:54,640 I remember Tony Blair coming out to Silicon Valley while he was still Prime Minister, 731 00:55:54,640 --> 00:55:59,480 and he had a dozen of us sitting around the table and he says how do I turn London into 732 00:55:59,480 --> 00:56:01,830 Silicon Valley and there is dead silence. 733 00:56:01,830 --> 00:56:08,200 I'm embarrassed, and I thought the Prime Minister just asked a question, maybe we should say 734 00:56:08,200 --> 00:56:13,580 something about what it was that made us, what did we have in common, what was it about 735 00:56:13,580 --> 00:56:15,680 Silicon Valley that we all shared. 736 00:56:15,680 --> 00:56:22,780 Steve Jobs was still alive then, he raised his hand and said, one thing we all have in 737 00:56:22,780 --> 00:56:27,060 common is we have all failed at one time or another, and it didn't leave a permanent mark 738 00:56:27,060 --> 00:56:29,050 on our forehead. 739 00:56:29,050 --> 00:56:33,820 Whereas in Europe, a failure, business failure is actually quite harmful. 740 00:56:33,820 --> 00:56:40,461 The real answer in that case was that failure was forgiven in the Silicon Valley, that may 741 00:56:40,461 --> 00:56:43,190 not be true in a lot of other cultures. 742 00:56:43,190 --> 00:56:48,270 What will be important though is to find what Steve Case is trying to do for example is 743 00:56:48,270 --> 00:56:53,950 to find ways of bringing serious investment into other parts of the country, but it can 744 00:56:53,950 --> 00:56:55,440 be just money flowing in. 745 00:56:55,440 --> 00:57:02,370 We have to have education, we have to have fresh educated people in business and finance 746 00:57:02,370 --> 00:57:09,350 and science and technology and marketing and sales, to go along with that investment. 747 00:57:09,350 --> 00:57:14,300 And we have to have places like the Patent and Trademark Office to help people protect 748 00:57:14,300 --> 00:57:21,130 the intellectual property that they have created, to give them some time to extract the value 749 00:57:21,130 --> 00:57:22,170 from that. 750 00:57:22,170 --> 00:57:26,590 Silicon Valley has taught us a bunch of lessons, and it's still teaching them. 751 00:57:26,590 --> 00:57:32,360 I think we need to learn from them so we don't make everything be exactly like Silicon Valley, 752 00:57:32,360 --> 00:57:36,390 because that may not be the right answer. Well, thank you very much. 753 00:57:36,390 --> 00:57:44,130 Warm hand for Dr. Cerf. (applause). 754 00:57:44,130 --> 00:57:50,240 Vint is a great friend of the USPTO and Department of Commerce in general, actually, and you 755 00:57:50,240 --> 00:57:54,000 are on the board of our sister agency at NIST and so on. 756 00:57:54,000 --> 00:58:00,280 If you haven't visited the museum upstairs, right in front there is a cardboard cartoon 757 00:58:00,280 --> 00:58:06,850 of Vint Cerf as a superhero, he really is a superhero of the innovation world. 758 00:58:06,850 --> 00:58:11,390 And an honor for us to have you here with us today. Thank you so much. 759 00:58:11,390 --> 00:58:12,640 Thank you very much. 760 00:58:12,650 --> 00:58:13,650 I appreciate it.