 Hi, everyone, and thanks for coming to this Berkman Klein London series talk by Scott Bradner called A History of the Internet. Scott Bradner was involved in the design, operation, and use of data networks at Harvard University since the early days of the ARPANET. He was involved in the design of the original Harvard data networks, the Longwood Medical Area Network, LMANet, and New England Academic and Research Network, NEARNET. He was founding chair of the technical committees of LMANet, NEARNET, and the Corporation for Research and Enterprise Network. Mr. Bradner served in a number of roles at the IETF. He was the co-director of the operational requirements area, 1993 to 1997, IP&G area, 1993 to 1996, transport area, 1997 to 2003, and sub-IP area, 2001 to 2003. He was a member of the IESG, 1993 to 2003, and was an elected trustee of the Internet Society, 1993 to 1999, where he was the VP for standards from 1995 to 2003, and secretary to the board of trustees from 2003 to 2016. Scott was also a member of the IETF administrative support activity, as well as a trustee of the IETF trust from 2012 to 2016. Mr. Bradner retired from Harvard University in 2016. After 50 years working there in the areas of computer programming, system management, networking, IT security, and identity management. He still does some patent-related consulting. Before I bring him up here, I want to do just a couple of housekeeping notes. One is this talk is being live-streamed and recorded. Two, please silence your phones. I always forget that. During the talk, on those silenced phones, feel free to tweet us at BKC Harvard. Scott will take questions at any time. Please use the mic so that you can be heard on the stream. There will be a microphone runner to get that mic to you. Questions are not comments, and they end in question marks. And without further ado, Scott. Is this thing working? It's nice to get applause before. It's nicer to get it afterwards. OK, so I'm going to talk about a history of the internet. This is not the history of the internet. There isn't one. This is a very much Rashomon kind of environment. I've talked to a lot of people that were involved in this. I lived through part of it. The people that I talked to, some of their stories have changed over time. So even within one person, they have different memories. But I'm going to talk about a sort of an overview, episodic overview. This is certainly not a complete picture, but it's a basic picture. It all started with a grapefruit. The electronic grapefruit was sputnik. When the Russians popped up, sputnik, the US military industrial complex and the political complex went apeshit. Lyndon Johnson said, the skies were alien. Well, Lyndon had a way with words. But it was a big deal. But luckily at the time, we had a president that actually believed in science. Imagine that. President that believes in science. Well, Dwight Eisenhower, being a military guy, definitely believed in science. And he, along with his science advisor, put together in three months an advanced research project agency for the Department of Defense. The idea of putting together anything in Washington in three months is amazing. But he did it. He got it going. The advanced research project agency who had a mission of basically exploring new stuff was the formal mission. The informal mission, anything italics here is quotes from source material. The actual mission, according to contemporary materials, was prevent technological surprise like the launch of sputnik. So here's the ARPA setting up process starting to work on things. But it wasn't just ARPA that was being funded by the US government to do work. RAND was funded. The Air Force was funding RAND and Hughes Aircraft. And there's a fellow named Paul Barron that was working there. Paul Barron came up with the basic concept of what is today's internet. He came in 1960 paper. He described ways that you could build a reliable network out of unreliable components. So cheap, inexpensive devices we call routers today. You put enough of those things together with enough redundancy that can be a reliable environment. And then in 1962, his seminal paper followed up two years later by an 11 volume set described the full set of internet protocols that we have today. The full set of conceptual internet protocols. We started, except it wasn't internet. It was a network protocol. So he came up with the idea of breaking communications up into chunks. He called them standardized message blocks. But the idea was you had a big file, you made it into little pieces. You had a voice conversation. You made it into little pieces. Each of those little pieces had an address at the beginning, said where the piece was going. An address after that saying where the piece is coming from. Some other information and a payload, which was a chunk of voice or the chunk of data. It was a distributed network. The bottom picture is the distributed network. Bunch of little circ, little dots are the routers, the forwarding engines. All the messages, the blocks were soaring forward so they'd be forwarded to the next little router along the path. Saved there, figured out where to go next and then go next. Had the advantage that if you took out some of those dots, there were other paths around it. And the routers would figure out those other paths and forward the traffic along the other paths. This is indicative of the times. This is a chart out of his 1962 paper, which is plotting the probability of a message getting through against the probability of nodes being destroyed, an atomic attack. So you take out half the nodes. If you've got enough nodes, the message will still get through. So he was describing a resilient infrastructure to survive an atomic attack. Big advantages to the message block-based thing, the packet-based thing. You can put together multiple communications on the same wire. They can be running at different speeds. If there's a failure, if somebody blows up a router or a router dies or somebody hits a phone pole and takes out a fiber cable, it takes a minimum time to switch over to a new path. Many different applications, including voice. He described using it for voice in 1962. The environment he was working in was the Cold War. It was a pretty ugly environment. He very much wanted, he was very much at war. In his oral history, which is pointed to here, he talks about what he was trying to do was create a command and control network which would survive a first strike. But unlike most people, he wanted the Russians to have it too. He wanted it so that the Russians wouldn't think that the US wouldn't retaliate from a first strike and he wanted the US to not think the Russians could retaliate. So he wanted both to have it. It was primary. All of his documents were open. There were public, there was two that worked to do with security of the system. But basically his idea was to make a public environment which keeps both sides from killing the other. It didn't work. I mean, it didn't get built. He describes in his oral history of going into AT&T to tell him about this. And after he made his presentation, they said, son, let me tell you about how telephones work. And spent the next few hours telling him how a telephone works and he said, and at the end said, do you understand now and do you understand why this won't work? And he said, no. He moved on, but he was offered an opportunity to get it built. But the Defense Communications Agency would have been the ones that had to build it. And there was no conceivable way the Defense Communications Agency could have built it. He realized that they didn't believe in it and they would have built something which would have failed and then it would kill the whole concept for eons. So he declined to have them build it. That was him. Meanwhile, J.C. Licklider at MIT was talking about intergalactic networks, which was really talking about global networks, and talking about how people would interact with computers in the future. He wasn't the first to do that, but he was an important one to do that. In 1962, he argued for, created, and then started out the Information Processing Techniques Office in DARPA, in ARPA. By the way, this was called ARPA Advanced Research Project Agency some of the time, and DARPA, Defense Advanced Research Project Agency some of the time. Congress flipped its name from time to time for congressional reasons, nothing to do with logic. But he created this PTO, and IPTO will come along because that's the organization that built the ARPANET. Bob Taylor took over IPTO in 1966, or 1965, and got the money, got a million bucks to build what became the ARPANET. Now it didn't cost a million bucks, it cost a few million bucks to build the original ARPANET, but still, he got the initial money to do it. He was a pretty important person in this because he believed in the concept, he believed in what Licklider had been talking about, and he believed that networks and internet computers were an important thing. Meanwhile, over in England, Donald Davies was working on the same general concepts that Barron had. He developed them independently, but in a paper that he wrote later on, he says, Barron invented the whole thing. He invented more details than Davies had. Davies came up with the word packet instead of standardized message block, which doesn't roll off the tongue anywhere near as well. So packet is a nice thing. He was introduced to Barron's work, and he said, okay, that's it, and learned from Barron, but also taught Barron. Next person in his picture is Lyle Roberts. He was appointed by Taylor to actually build the ARPANET. A key thing that he was going to go, the idea here of the ARPANET was to be sharing large computing resources. The Defense Department and other government agencies were buying big computers for people, big computers for researchers, not for people, for research organizations. And the computers in those days were very big, very expensive, and there weren't many of them. So you want to share them. The ARPANET wanted to share them. Licklider and Company said, this is the right way to do it. We share these resources. But the people who owned those resources didn't want them, A, weren't all that interested in sharing them, but they were persuaded by the fact that the guys providing the money said, you will share. But they really didn't want their computers to be doing the processing of sending data around. They were okay if somebody wanted to remotely log in and use it, but they didn't like the idea that their computer was going to be processing all those packets that went through. And West Clark came up with the idea of building separate little boxes, which would do the processing. And that's rescued the basic concept because Robert wasn't going to get through the idea that they were going to have sharing resources where 20% of the machine was going to be taking up moving packets around. So Clark's idea of having a separate box that was now called a router was a key thing. Roberts had come up with the idea of building such a network, but he didn't know how to do it. He didn't know what techniques, what things to use to do it, what technology. He knew he didn't want circuits because circuits were going to be dedicated between links, and that was, he didn't have dedicated, he wanted to share those links among multiple researchers, so he didn't know how to circuit. But he didn't really know what to do. He went to a meeting, an ACM meeting in Gatleyburg, Tennessee in 1967. He presented a paper about the justification and the general concept of this ARPANET, this resource sharing network. Another person at that same meeting was a fellow from Donald Davies' group who presented about packet networks. Roberts told me personally that he had never heard of packet networks before this meeting. He had never heard of Barron's work and he had never heard of packet networks. Scandalberry told Roberts in the meeting about Barron's work, Barron, Roberts had Barron's, volumes of Barron's work on his shelf in his office, and he went back and read them. And he said that was a revelation and he basically took all of Barron's ideas, modified by Davies, updated by Davies, and that's what he decided to use. He used packet networks for the ARPANET. And he took over the IPTO in 1969. So here's the ARPANET, starting up. 68 and RFQ went out, 140 companies were asked to bid on building these little message processors, this is what turned out routers, and only 12 companies actually bid. IBM did not bid. Later on, they justified it by saying they didn't think it could be done cost-effectively. AT&T didn't bid, but they said they didn't bid because it wouldn't work. Packet Networking was never going to be useful for anything, just like they had told Barron. Beamian Ann out here in Fresh Pond won the contract, and within a year, less than a year, they delivered the first router, the first input. Interface Message Processor, even though Ted Kennedy called it the Interfaith Message Processor. So entirely different concept. So the first node was delivered to UCLA in 69, and a month later, a month and a half later, the first message was sent between two different imps in two different locations. Initial networks, four nodes, Kleinrock and others were there. It quickly expanded by the, next year it was on the East Coast. John Clemson reminds me that MIT got on before Harvard, so here it is, there you are, John. And then Beamian got in there someplace. By 73, it was international. By 80, it was a couple hundred hosts and 20,000 users, but those were users in the US, they were only people with US government funding, or who were staff on computers that were funded by the US government. Wasn't open to all people. I mean, I got my first account in 71 by conning my way into being staff on the PDP-10, that was the ARPANET interface in Aiken computer labs. But otherwise, I wouldn't have been able to get on because I wasn't government funded. I was time to time, but not at that particular time. Everything changed in 83, when CSNET made a deal with the ARPANET that any organization getting their CSNET connectivity, computer science network connectivity via the ARPANET could open up that pipe for every faculty, staff, and student for email, not for file transfer, but for email. So starting at 83, we started creating, teaching generations of students about this networking stuff before it was only the computer scientists, but now it's generations of students so that when they went out to business, when they graduated, went out to business, we need this, join up. So that was a huge forcing factor of getting the internet dispersed. So basic question, Baron's work was for 80 war, it was for protecting against the bomb. Robert's work was for sharing large resources. I've talked to a number of the people that were the worker bees at the time, and they all say from their point of view, it was sharing resources, it had nothing to do with the bomb, it had nothing to do with resilient infrastructure for the bomb. Yes, the resilient infrastructure was there, it's like what Bruce Davies said, that he was building his network over the French telecommunication system so he knew it was unreliable, it was a feature. And so he had to build a network that would not depend unreliability underneath. The packets could be lost, they could be duplicated, they could be reordered, and that was okay because the network was controlled from the edges. The source and destination machines took care of making it reliable, took care of reordering things, took care of asking for a retransmission if things were lost. So the worker bees at the time said, we're building a resilient resource sharing network. Steve Crocker said that this was what he believed, he was one of the people in there. But he later on a few years ago talked to one of the higher ups in the Pentagon at the time. And the guy said, you don't know what I was telling my boss. The money was coming from the Pentagon. Money was coming from the Defense Department. So it's very conceivable that the top levels of the Pentagon were telling Congress and they were telling each other, this is for anti-nuclear, this is for surviving nuclear war. Even though the people that were doing the work didn't think that. So the answer is ambiguous. And it's not something I can answer. This is, I can all you present you with the information. A lot of people were adamant that you need, you should use the microphone. That's the John Clemson I mentioned a minute ago. The reason why the DOD changed ARPA to DARPA and back several times was because Senator Mansfield who would have been one of the first people in Congress to be told this story thought that there was too much that was being done which was not directly defense related, i.e. building bombs. And he tried to force the ARPA and that project out as part of one of those transitions because it wasn't obviously defense related enough. And that was both the source of the name change and the rest of this. So I don't know who is telling who what but certainly nobody told Mansfield in a persuasive way that this was about bombs. It was not about bombs. It was not, well, told Mansfield that it was about bombs in order to convince him it was okay. And that's basically what Crocker told me was that going up to Congress, it was about bombs. But anyway, we'll move on. So now when Bob Kahn shows up in this picture he was a PhD Princeton but then went off to work for BBN and worked on the M. And moved to ARPA IPTO and then made a big demonstration in 1972 of what was then the ARPA net sharing resources in Washington. One of the people that were demonstrating there I think was Bob Metcalf said that he was demonstrating all day of how to do resource sharing and it was working perfectly. And a bunch of people from AT&T came in and late in the afternoon all dressed up in their pretty little suits and stood at the back of the room and while he was demonstrating in it, crashed for the first time. And he turned around just in time to see them all grinning and laughing. And he said, it was that moment that I realized they were the enemy. But so they didn't really believe in this stuff. But so Kahn is there. We'll come back to Kahn. Danny Cohen was a Harvard student and then MIT student. He was doing flight simulators work between Harvard and MIT and using the ARPA net link for that. He was mostly working on later on on speech and he did do speech work over the ARPA net. But he realized that the original ARPA net protocol, NCP, Network Control Program, was a reliable protocol and a reliable protocol is reliable because it knows when something's missing by timing out and saying, oh, I need that packet again. Well, if you're just in the middle of speech and you get a two second drop out of the middle, it doesn't work very well. So he said, we need an unreliable. We need something which is not part of the reliability picture. And he was given permission to use an unreliable pathway through NCP in order to get his speech work done. Over in France, Louis Poisson was working on his version of a packet network. Peer datagram network over France telecom lines. So it couldn't possibly be reliable. All of the stuff end and end. Bob Khan asked VentSurf to figure out a way to be able to make a scalable network. The ARPA net, as it began, was to connect computers together. Connect a computer at Harvard to a computer at Berkeley to a computer at MIT. It was not to connect networks together. Khan realized that scaling wise, you needed to connect networks together. And he worked with VentSurf to come up with a set of protocols to interconnect networks and the computers on those networks. And that's the internet protocol. The internet protocol, Laird protocol, the first version of this was a reliable protocol called ITCP, it was meant to be reliable because of course you want a reliable transfer of data. Danny Cohen and John Postel and David Reed, cornered VentSurf in a hallway in Southern California and said, we can't do this. We need a way to do something that's unreliable for voice and for management. If your network is all screwed up, you don't want to have forced reliability on your management protocol with just talking to your routers because the message will never get through. So you want an unreliable method. So they added what is called UDP, unreliable user data grant protocol, otherwise known as unreliable, that enables the voice over IP that we use today. The general idea that the network is just gonna deliver stuff and is not gonna be involved in the value decisions was articulated in the end-to-end arguments and system design from Salts, Reiser and Clark from MIT. Basically, you don't put stuff in the network which is redundant with what the end-to-end systems can do. The end systems knows what service they want out of the network. The network cannot know that. There's no way for the network to figure out what service they need, what level of reliability, what latency, all of that kind of stuff, they can't figure that out. But the telephone companies and the carriers thought it was a really good idea to have the network understand what's going on. That conflict between a stupid network that just delivers the bits and an intelligent network which the telephone company called their network or the intelligent network because all of the smarts were in the telephone switches. Star six nine was a telephone switch thing that took 10 years to deploy. So it's an article called Net Heads versus Bell Heads in Wired Magazine. If you've not read that, read it. It's a very good explanation of that conflict. David Eisenberg wrote The Rise of the Stupid Network. While he was an AT&T employee, he wasn't one for long because this message was basically AT&T, the network should be ignorant. The network should simply deliver the bits. Just get out of the way and deliver the bits. And of course AT&T didn't think that was a good idea and to this day, they won't let David post a copy of his paper on his website because he was an AT&T employee when he did that. If you're on his website, there are pointers to other people who are posting it. So the internet as it evolved from the ARPANET on really didn't need control. Didn't need centralization. It needed coordination. Everybody had to agree that this field in this packet when it said 25 meant email. They didn't care what the number was as long as everybody agreed the same number. So somebody wrote that down as the 25 means email. That was John Postella. Also have to say who has what IP addresses. And John wrote that down. When Harvard needed his first batch of IP addresses, I sent John Postella a message saying, we need IP addresses. And he sent me back a message saying, here they are. 128.103.0.0 class B. That's Harvard's. Harvard's gotten more since then. And then there's gotta be some pointers to when you resolve DNS names, harvard.edu or www.harvard.edu. There's gotta be somebody up there pointing at the top level. Where are the computers that resolve the names for .edu? There's a coordination, not governance. They're not control. All of that stuff was done by John. Evolved it into the IANA, Internet Association for Names and Numbers. That turned into authority for names and numbers. That turned into ICANN a little later on. So what did the powers that be think this Internet thing was? Who would think that was irrelevant? It's a toy. This is best effort, packet based, no guarantees, no reliability, no security, nothing. It's a toy. IBM, AT&T, and in particularly important, the regulators. This is a toy. It's a research toy. It's a research network and a research toy. So it's not important enough to regulate. That was key because if it was regulated, we'd have to get permission to do things. The bumblebee is on there because IBM, I was participating in a lot of IBM user groups at one point and IBM said, quote, you cannot build a corporate network out of TCP-IP. Direct quote from a fellow named Gray at IBM. I was the chair, the co-chair of the TCP project within chair, within this IBM user group. We chose a bumblebee for our badge symbol. You put badge symbols like if you were in communications and octopus because it goes everywhere. If you were numerically intensive computing, it's a circle with a 42, and if you didn't know what that meant, you didn't get one. We chose a bumblebee. Because IBM said TCP-IP wouldn't fly. Aerodynamic theoreticians have said that bumblebees couldn't fly. Bumblebees didn't give a crap. They kept flying. We didn't give a crap. We kept going. IP kept going. So connectivity evolution, let's start with the ARPA net, the CS net, and NSF net, the regional networks, near net which was mentioned in the introduction, bar net, people like that, and then migrated to commercial ISPs. The feds were completely out of the business of providing network connectivity for the public in the mid-1990s, and then we got carriers. It's moved from something which was feds to regional networks to commercial carriers to commercial internet service providers whose business was providing internet service to carriers whose business was content and is now doing internet provision. And that comes back to haunt us in a bit. Then the web expanded, the web exploded, and made it so that anybody could do anything. I was asked a few years back looking forward from the 1980s to the internet of the day, what was my biggest surprise? I thought that the internet was gonna be a big, important thing in the mid-1990s. It was gonna be an important thing for everybody. It was the telecommunications infrastructure. My answer was mom surfing. It never occurred to me my mother would use this because it took magic incantations to do it. The web changed all of that. The web made it so that anybody could use it and the use is exploded. So what made the internet important? What was the key things? Well, it was the end-to-end model and neutral networks. Network neutrality, the network just delivers the bits, doesn't get involved in deciding what bits should be delivered which way, doesn't get decided on, doesn't decide whose bits should be prioritized or whose bits should be let through. The networks are neutral. That allowed permissionless innovation. Anybody like Tim Berners-Lee could invent something like the web or somebody could invent Skype and you didn't need permission from the carriers to do it. You just did it. You didn't need permission from the regulators to do it. You just did it. Of course, capacity increase with Moore's law and the fiber bandwidth was a big deal. We didn't change the net for any particular application. The net was not designed to support some particular thing and the IETF we had when we first started getting the big push from telephone companies, they wanted to change the protocol to make it better for voice and we said no. The internet is good enough for stuff. The stuff it's good enough for use it for. If it's not good enough for it, don't use it. It's a pretty simple equation. Now it's gotten better for everything because of speech increases and no regulations meant we had permissionless deployment so we could innovate and we could deploy. And then very importantly, the Communications ECC Act was killed and overturned by the Supreme Court as unconstitutional. That was an act which required you personally to ensure that if you sent something on the internet that somebody anywhere in the world under 18 shouldn't see they didn't see it. Technically that's a bit of a tough, whoa. The fact that it was technically impossible it was irrelevant to the court case. The Supreme Court decided on it not because it was technically impossible but because it wasn't the best way to achieve those results. They thought the best way was to put filters on the individual kid's computers rather than trying to block the entire world because of course most of the world doesn't pay attention to the US. So under the First Amendment this was just too much of a burden on speakers to be able to be supported. Section 230 on the other hand of the Communications ECC Act is the section that permits Harvard to be an ISP and not be killed if some Harvard student downloads music. We can be a neutral thing. We're not responsible for the actions of our users as long as we do due diligence when we're informed of those actions. And of course the key thing that made the internet successful is that everything is bits. Everything is bits. Ada Lovelace figured out that in 1843 that computers, analytical machines and what she was dealing with weren't limited to dealing with numbers. They could deal with anything that those numbers could represent such as music and she talked about using it for music. So what's the future of the internet and technology-wise? All of these things, ATM, MPLS, Perfilo, Queuing and things like that are things that different people have proposed, different organizations have proposed to replace the internet that's here. Almost those are circuit-based. So get rid of the unpredictability. Put in circuits. That's ATM, MPLS, Perfilo, Queuing, et cetera. By the way, how many people have seen the movie Touch of Evil? It's a reasonable number. It's a great film-to-wa film from Orson Welles. This is a scene in there where Orson, as a corrupt sheriff, comes into Melina Daitric as a brothel runner, Madam, and says, tell me my future. Come on, read my future. You haven't any. What do you mean? Your future's all used up. Every one of those things, ATM, MPLS, Perfilo, et cetera, is some company, some organization, some standards body, some government, has said the internet's future's used up. We gotta do it right. The internet's kept going. The 3G, 4G, 5G thing here. First time I heard this was in Taiwan by one of the telephone people saying, 3G is around the corner. And when 3G shows up, you won't need any local area networks. You won't need any corporate networks. It'll all be provided by the telephone company. Same thing they said for 4G, and they're saying that now for 5G. I didn't believe it then, I don't believe it now. And other things, so. But this internet thing has been impactful. It's destroyed businesses, it's destroyed societies, it was a key facilitator of the Arab Spring. It's a big hit. This paragraph at the bottom here is from the first international telecommunications convention treaty of 1865. And it was the telegraph convention. There was required that telegraph operators watch their traffic, watch the telegrams, and block any that would be in danger of the security of the state, violate the laws of the country, public order or morals, and to report it. So the very first regulation we have on international telecommunications was a spy one, spy on your users and report those that are doing something immoral. That's something we do not have in the internet. We have been regulation free. The CDA tried to do that and was killed in the US. That doesn't mean that it doesn't happen in other countries, it does. So what's the future? On the control side. Carriers of course have been trying to control it ever since they figured out that there was something there to control. That it wasn't entirely bumblebees and uselessness, it was something that's really there that could actually add value. Some of the big carriers like Comcast are now getting more money from internet service than they're getting from video. Governments of course the same. They've been petitioning the ITU to take over governance of the internet so they can control it, so they can avoid the loss of telephone revenue for example, things like that. Certainly the current FCC is blowing away the Title II network neutrality rules are enabling carrier control. They are making it completely legal for the telephone carrier ISP to control anything like what you can talk to, who you can do, talk to, what you can use to do the talking. Newt Genders was once quoted as saying, a role of government is to guide technology. Not something I'd necessarily want the Congress to do. A guided or controlled internet wouldn't have been what the internet has been to date. And 20 years ago, 22 years ago or so, I testified in the communications decency act hearing, I was the witness for the American Library Association and I said that the power of the internet was chaotic. The power of the internet was the forum to innovate which was chaotic. You could not predict what would succeed and what would fail. It was chaotic. But that was the fundamental power of the net was represented by that chaos. And the question on the table today is will that forum continue? So questions, but use the microphone because the people out there in audio land need to hear. John? I'm going to skip quibbling about history and ask a more interesting question. You in my mind correctly denounce the efforts to fix the internet to make it better for particular applications. And similarly denounced some of the related issues about pulling things together and switching toward more circuit-like things again to facilitate particular applications. How do you feel about changing the internet to make it better for the web? Or to put it differently, would you like to comment on quick? I think that anything which makes it better for one application makes it worse for others. And there ain't any, there's no perfect unity. I mean, the general purpose computers are powerful because they're general purpose, not because they're good at math. Any other questions? Back in the back. No, no, you gotta use the microphone because the people out there in streaming land won't hear it. Oh, my bad. Can you talk more about Title II and what that has actually affected? Title II is the, well, the history there is that the Federal Communications Commission under Powell came up with four principles of a proper internet. They were that you could use any device you wanted to, any legal device, you could access any legal content and you could access any legal application and you had transparency of what was going on. The FCC put out those four principles, tried to enforce them, they got sued and they got kicked out because they had done no proper procedures to adopt those policies. They then went through a procedure to try and adopt something very similar. They got sued again and it got thrown out because they didn't undergo the right process. They didn't have the statutory authority to do it. They tried again and it got kicked out again and that last time the court said, the only thing you can do in order to be able to do this kind of regulation is to declare that the internet service is a common carrier. So it's under Title II of the Communications Act. The FCC did that. They got sued and it was upheld. The FCC was upheld in court. That it was legitimate for them to decide that internet service providers were common carriers and therefore had to abide by the rules of common carriage. The particular thing the FCC did was say they're subject to Title II but we will only actually enforce, we'll forebear on other rules that are relative to those four principles. So the 200 other rules out there for who you can give free tickets to, the movies to, literally. That's the kind of thing that's in the telephone regs. They don't apply and the FCC don't apply. A lot of people really didn't like the fact of Title II because it gives the FCC so much power. You have to pre-get how much you're gonna pay for things. The tariffs have to be pre-approved, the application has to be pre-approved. All of that kind of stuff is potential but the way the FCC approved this particular rule is they're gonna forebear on all of that. People don't like that because it's got the potential there. Somebody could sue the FCC later on and say these people are being unfair. Under Title II, you can do this. Go do this. And the courts could, in theory, tell them to do something like control prices. But the courts had said the only way they could do it is Title II. So the new FCC overturned that throughout that Title II stuff. But in their new regulations, say the ISPs can do anything they want except they have to be transparent about it. The rubric under which they said they have to be transparent is exactly the same rubric that the courts threw out when the FCC was trying to do the four principles. So it's in court today. It was argued last Friday. They could be overturned on a number of different reasons. The number, it could overturn the overturning of Title II but because the process the FCC used was bad, they literally ignored 20 million comments. They completely ignored them. Now it happens that a significant chunk of those comments were done by bots. Research later on showed that all of the bots hated network neutrality and wanted Title II overturned and all of the humans wanted it retained. It gives you some information about bots. But so it could be overturned on that basis. It also could be overturned on the basis that they're imposing this transparency requirement that they don't have the authority to impose. So but it's potentially very big blunderbuss but it's the only one the courts would let the FCC use. You need to wait for the microphone. Oh, there's one back in the back. There's one in the back first. Okay. The packet design of TCP, IP is necessarily somewhat arbitrary. I mean, you could design another packet that would still work and used to have deck net and bit net and Joe blows net. So is that packet design somehow superior or is it entirely arbitrary? And if we can design a better one somehow, not when, not in a party, not whether but when will it be installed to completely rewrite the internet? Well, first of all that the TCP, IP is as much a regulatory accident as anything. The defense department agreed to switch the ARP to TCP, IP and January 1st, 1983. The NSF net when it started up, the Dennis Jennings said, only IP is enabled. It can be run over the NSF net. You couldn't run deck net. Harvard had a deck net computer talking to a deck net computer at Princeton over the NSF sponsored regional network link. And the pieces that Harvard said, we would like to use deck net. Is that okay? And NSF said, no, you have to use TCP, IP. So it enforced that. Second thing is that ARPA paid BBNN, paid Berkeley to take the BBNN TCP, IP code and put it into UNIX. And then startups could buy a whole tape with all of UNIX, including a full TCP, IP stack for 50 grand or so. Much cheaper than it would be for them to build their own. So the department of defense basically pushed out IP to everybody, much cheaper than building your own. So that meant that all of the little startups had it. The big guys, they started moving to OSI, like deck net, tried to move to OSI. It was so much more complicated and so much more fragile that it just didn't work. And then finally the US Department of Defense Department which had the US government which had said, if you're gonna sell networking stuff to the US government, you have to support OSI. Backed off and said, you can buy whatever would do your job, including specifically TCP, IP. So it was somewhat of a regulatory accident that's IP. Deck was out there with deck net, IBM was out there with SNA, Xerox was out there with XNS, Novel was out there with IPX. They all had their strengths and weaknesses. IP is there because it became the common language. Do you think it'll ever be replaced by something which is somewhat back at Y superior? So the ARPANET switched from NCP to TCP IP officially on January 1st, 1983. It took six months. There were a couple of hundred computers. If you project that out, if you want to switch the internet of today which has a few billion computers to a new protocol, it would take a few hundred thousand times the projected lifetime of the universe to do. So don't bet on it. So 5G, I ran into a telecom guy at a party the other night who said that the internet was a Trojan horse for the telecoms and had been trying to get the horse out ever since. 5G is how they think they're going to do it and he's a telecom guy. The idea he said is you're going to get very high speed, you're going to get it on your phone, you get it on everything. It's going to have local storage. And they're going to compete with Amazon by blowing up their cloud and distributing it everywhere and they're going to have their cloud and they're going to charge. They're going to get it on Hollywood's business. They're going to be on Netflix's business and it's going to be all top-down distribution and then they finally win. So two questions. One is, is that true or close enough? And B, how do we fight that shit? Well, first of all, if it was a Trojan horse, it was against the telecoms. Yeah, yeah. So it wasn't for the telecoms. Oh, right, that's the episode. Yeah. And that's the same pipe dream that the carriers have had from the beginning. The carriers have considered it. There's a whole big tenu in the late 90s that content was king. That somehow that when the carrier started providing you with the ability to have a Tivo on demand, a couch potato heaven of Tivo, you're going to triple or quadruple your monthly fee to the telephone company. Well, content isn't king. And those days, the particular days when the telephone company were saying that, there was 10 times more revenue from telephone calls than there was for all of Hollywood put together. And yet they thought that once they got the movies, they would increase their revenue by five-fold. It didn't make any sense then. It doesn't make any sense now. You really want to depend on Verizon as your entire lifetime, for your entire life experience. I don't think so. It's not quite as bad as what a fellow from MIT said a few years ago when we used to have 9X was used to be the carry, the telephone carrier in the Boston area. And he said that was the air quarry word for moron. Verizon is quite that bad, but it's not good. The story from the party sounds sadly like the plot of the TV show Silicon Valley, which is satire, so I'm not sure I would treat it with a great deal of respect. But my real question is, whatever happened to IPv6 and IPsec? What do you have in your hand? That one. What kind of one? iPhone. You got IPv6. IPv6 is native on the iPhone, but it's native on most cell phones. It doesn't have to do anything right here because Harvard doesn't support IPv6. But if you're using the Verizon network, what network are you using? T-Mobile. T-Mobile. I'm not sure about T-Mobile. Verizon supports V6. Comcast does V6 to your house. The big thing with V6 was it was invented to solve a particular problem, which was running out of IP addresses, and try to do a little bit more at the same time. But it didn't provide a huge number of carats. So it didn't have a lot of reason to move there unless you're running out of addresses. We're running out of addresses. So big carriers like Verizon, like Comcast, can't use V4 addresses because there ain't enough of them. So they're using V6. It's transparent to you because it's infrastructure, just like mom surfing. It's completely transparent to you. But it's there. And it will continue to expand because there isn't any choice. There's no V4 addresses available, not in the US anyway. And IPsec? IPsec? That's VPN. Go to your system controls on the iPhone and press VPN. That's IPsec. Anything else? We got a question in the back. You at first mentioned that regulation would be a bad thing, but then you seem to agree that things like net neutrality were positive. So would you agree that there might be some types of regulation that could be positive? And what would those types be? What are we lacking right now in terms of regulation that perhaps promotes or enforces openness? Well, at this instant, what we're lacking is the enforcing factor to be sure that the ISP keeps out of the way. That's what net neutrality is all about. We had that with the Title II stuff. We don't have that anymore because that was overturned, though it's in court. So we'll see what happens there. That regulation is a, see, it's not telling you to do something. It's telling you not to do something. And that's sort of the difference. The regulation that telephone companies on to go, for example, is I want to bring up star 6'9". Well, I got to get permission. I got to go through a regulatory process to say how much that's going to cost. And then that's never been the case for the net. But with the ISPs now in the middle, it's not a regulatory thing. But the ISP can say, well, if you subscribe to this service, it doesn't count against your packet limit, your byte limit, or your download limit. But if you subscribe to this other service, which we don't have a business relationship with, then it does. And now the ISP is in the middle deciding who's winners and who's losers. And that's a regulation that says, keep out of the way. It's very different than a regulation that says, ask me first. At one point, a while back, Verizon wanted to regulate so that all voice over IP had to go through central servers. Now they said it was so that the FBI could enforce wire tapping, but really it's so that they could bill for it. And that kind of regulation to force that would inhibit the innovation, the ability to have that chaotic environment to innovate. Anything which gets in the way of that, asking permission first to innovate or permission to deploy, is an inhibition of the dynamics for which got us the internet. John, again? I want to give a slightly different answer to that last question, because I think another aspect of that is important. One of the things we have been lousy at as a society and a network and community is distinguishing between things which are ultimately legal matters and things which are ultimately technology matters. We try to substitute the technology to the legal issues. We try to substitute much more rarely lately, but more so in years past, legal solutions for technology issues. We're spending tremendous resources on the network, usually at the wrong end, trying to deal with spam and phishing and other kinds of attacks, which are basically legal problems. And if you think about those as part of the regulatory picture, then a lot of the answer to the answer Scott gave you to your question is we better get a lot better at figuring out the difference between the legal problems, the criminal code problems, and the technical solutions real or alleged. Question I wanted to ask you is that one of your slides you put up a picture of the logos of a large collection of relatively early entry commercial ISPs. And then a few more recent ones. And a few more recent ones. But in the last few years, we've seen significant consolidation there to the point that if that picture were drawn today, it would have a lot fewer icons on it. It basically has the bottom three, I think. Yeah. Do you think that's an issue going forward? And if not, why not? And if so, what do you think we do about it, if anything? Well, I think it's a significant issue not in and of itself. It's the pipe dream of the fiction that the FCC undergoes saying of this competition for the residential market. There isn't. And yet the FCC says that 60% of the country or 70% of the country has competition, realistic competition for high-speed internet service. Their definition of high-speed is five megabits. So that's questionable. Five megabits down and two up, I think, one up, which is piddling in Europe and Asia. But I think that fiction is a very serious issue, which is the fiction which is driving this, getting rid of Title II. It's the fiction that's driving the Congress in general. And a few years ago, I wrote a column that the people in Congress think that there's competition on the internet provider environment, but they also think that the driver of the internet is the carrier. And this is like you standing beside, you've been asked to evaluate highways. You're standing beside a highway, but your eye is in your ankle. And all you can see is the asphalt. You can't see the cars and the trucks and everything going over the highway. So your answer to better highways is to enable asphalt manufacturers, not to enable car manufacturers. And the Congress, eyes and their ankles, all they're seeing is the carriers, the telecommunication carriers. They're not seeing the power that the fact that those carriers are out of the way drove. That's the Googles and the Amazons and the Skypes and everything else that's riding over on top. That they can't see. They don't see that that's part of the internet. And that's, to me, the biggest problem. No competition, the myth that there is, and belief that the internet equals carriers. That's just crazy. On that positive note.